JP2005004657A - Information classification method and device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To classify information mainly in consideration of a user, to realize the classification such that positioning of the information or the user is understood, and to allow the user to passively receive information related to a classification result thereof. <P>SOLUTION: A server device 110 comprises an information management part 111, a characteristic amount calculation part 112 and an automatic transmission/reception part 113. A client device 120 comprises an information management part 121, an information classification part 122, a group classification part 123, and an automatic transmission/reception part 124. The information management part 111 manages the information about a classification target in a lump, and the characteristic amount calculation part 112 calculates a characteristic amount of the information. The user obtains the information or its characteristic amount registered in the server device via a network by the information management part 121. The information classification part 122 and the group classification part 123 classify the information and a group, and each classification result can be displayed and viewed on each the client device. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

【０１０７】
分類は次のように行われる。すべての入力データｘ_ｉに対し、繰り返し回数ｔだけ以降の処理を行う。入力データｘ_ｉに最も近い参照ベクトルをもつセルｃを探索する。探索は式（２）に従って行う。
【０１０８】
【数２】

【０１０９】
すなわち、ｉ番目の入力ベクトルｘ_ｉと出力層のｊ番目のセルの参照ベクトルｍ_ｊの距離が最小となる出力層のセルｊを見つけ、それをｃとする。ここで、入力ベクトルと参照ベクトルの距離は、例えば式（３）のようなユークリッド距離で表現する。セルｃの近傍集合が形成される。この近傍集合Ｎ_ｃ（ｔ）は、例えば式（４）で定義する。
【０１１０】
【数３】

【０１１１】
式（４）で、ｄｉｓ（ｊ、ｃ）はセルｊとセルｃの距離を表し、セルｃからの距離がｒ（ｔ）より近いセルｊを集めたものを近傍集合とする。例えば、近傍集合１５０２はセルｃの周りに形成されている。次に、近傍集合に属するセルの参照ベクトルを式（５）、（６）に従って更新する。
【０１１２】
【数４】

【０１１３】
すなわち、近傍集合Ｎ_ｃ（ｔ）に属するセルの参照ベクトルはその値を式（５）に従って更新し、それ以外のセルの参照ベクトルは何もしない。ここで、距離ｒ（ｔ）と係数α（ｔ）は、繰り返し回数ｔに対して一様に減少させる。最後に、入力データに対して式（２）のように最も近い参照ベクトルを持つセルを探索してデータを配置する。以上の処理によって、参照ベクトルの各元の値は変化していくが、更新後も参照ベクトルと入力ベクトルとは関連付けされているため、いずれの情報がいずれの場所にマッピングされているかを知ることができる。
【０１１４】
従って、情報の特徴量ベクトルを入力データとして情報の分類を行い、処理後に特徴量ベクトルに最も近い参照ベクトルを持つセルに情報を配置した結果の可視化が行われることとなって分類結果を得ることができる。さらに、特定の情報を中心に分類するために、その情報に対しては自己組織化マップのｏ×ｐの中心セルをセルｃとして近傍集合を形成して分類処理を行う。このように処理することによって、常に特定の情報が中心になるように分類が実施される。
【０１１５】
なお、ｋ−ｍｅａｎｓ法は情報をｋ個のクラスタ（グループ）に分類する方法であり、最初に分類する数ｋを設定して次のようなステップで分類を行う。
【０１１６】
ステップ１：ｋ個の初期クラスタ中心ｚ_１（１），ｚ_２（１），・・・，ｚ_ｋ（１）を適当に決める。
ステップ２：ｎ回目の繰り返しステップにおいて、すべての入力データｘを次の方法でｋ個のクラスタに分類する。すべてのｌ＝１，２，・・・，ｋ （ｌ≠ｍ）について、
【０１１７】
【数５】

【０１１８】
とする。ここでｓ_ｍ（ｎ）はｚ_ｍ（ｎ）をクラスタ中心とする入力データの集合である。
ステップ３：ステップ２で得たＳ_ｍ（ｎ）の新しいクラスタ中心ｚ_ｍ（ｎ＋１）を
【０１１９】
【数６】

【０１２０】
とする。ここで、Ｎ_ｍは、Ｓ_ｍ（ｎ）の数である。
ステップ４：すべてのｍ＝１，２，・・・，ｋに対して、ｚ_ｍ（ｎ＋１）＝ｚ_ｍ（ｎ）となればアルゴリズムは収束したものとして終了する。そうでなければステップ２に戻る。
【０１２１】
また、ＩＳＯＤＡＴＡ法はクラスタ数ｋを与えることなく、入力データの性質によって変えていく方法である。これは、複数のパラメータを与えてパラメータを変化させながらクラスタ分類を行っていく方法である。しかし、パラメータが入力データに依存して決まり、その設定も煩雑であるという欠点がある。
【０１２２】
ｋ−ｍｅａｎｓ法やＩＳＯＤＡＴＡ法で分類を行った後は、クラスタ毎に２次元や３次元にベクトル量子化を行い、２次元や３次元上に配置して可視化を行う。
【０１２３】
図１６は、クライアント装置１２０のグループ分類処理について説明する図である。グループ分類処理は、ユーザのリクエストに応じて実行される。また、情報が追加された時や定期的に実行されるようにすることもできる。
【０１２４】
グループ分類処理では、基本グループ３２５に属する情報特徴量３２４を用いて、特定の所有者を中心として基本グループの特徴量を計算し、その特徴量から基本グループを分類する。ここで、所有者とは、情報の所有者のことであり、対象情報が文献であれば執筆者あるいは著作者のことである。
【０１２５】
グループ分類処理は、まずクライアントデータベース１２５から情報特徴量３２４と基本グループ３２５とを読み込む（Ｓ１６０１）。次にユーザによって、情報特徴量の調整、分類の中心となる所有者および中心となる基本グループを決定する分類基準である分類の設定を行う（Ｓ１６０２）。ここで、特徴量の調整とは、ユーザの好みや分類の内容に応じて、サーバ装置で作成した特徴量を修正することである。次にこの情報の特徴量を用いて基本グループの特徴量を計算する（Ｓ１６０３）。これはグループ毎に行われ、Ｓ１６０２で設定した分類の中心となる所有者と基本グループにより決定する。この計算した基本グループの特徴量から基本グループの分類を行う（Ｓ１６０４）。
【０１２６】
次に各基本グループについて、基本グループに属する所有者の分類を行い（Ｓ１６０５）、最終的な分類結果をデータベース１２５に格納し（Ｓ１６０６）、終了する。
【０１２７】
なお、グループの分類やグループ内の分類は、前述した自己組織化マップ等の分類方法を用いて行う。
【０１２８】
図１７は、基本グループの特徴量の例について説明する図である。各基本グループは、グループ名１７０１と代表ベクトル１７０２とグループの重み１７０３をもつ。代表ベクトル１７０２は、基本グループに属する全ての登録情報の特徴量ベクトルの平均などで表現される。グループの重みは、初期状態はシステムの管理者が基本グループの関連度により決め、分類時にはユーザによって変更することができるようにする。基本グループに属する情報は所有者毎（１７０４〜１７０６）にまとめておき、所有者毎に基本グループに属する情報数、その所有者の全情報数、情報数を全情報数で割った所属率、および情報の特徴量ベクトルの平均を取った平均ベクトルをもつ。また、所有者毎に類似度を設定し、例えば所有者１７０４を分類の中心とする特定所有者とするとき、図１２のα／βなどを類似度とする。ここで、図１７の網掛けの部分が基本グループの特徴量を表し、例えば１７０７のように基本グループの情報特徴量ベクトルを生成する。すなわち、基本グループの代表ベクトルと分類中心となる基本グループの代表ベクトルの距離（例えばユークリッド距離）、基本グループの重み、分類中心となる所有者の所属率、または基本グループに属する所有者の類似度の総和などを元としたベクトルを基本グループの特徴量ベクトルとする。ここで、ベクトルの項目は自由であり、システムの特性に応じて決めることとする。ただし、ある所有者が中心となるような類似度を用いる必要がある。また、特徴量ベクトルは便宜上規格化などを施す必要がある。
【０１２９】
図１８は、クライアント装置１２０がサーバ装置１１０から登録情報を取得するときの情報管理処理について説明する図である。ユーザが分類結果などを用いて所望の情報を探し出し、実際にその情報を取得する場合は、次のように情報管理処理を実行する。
【０１３０】
まず、クライアントデータベース１２５から分類結果などのシステムの状態を取得する（Ｓ１８０１）。次にユーザが所望の情報を選択する（Ｓ１８０２）。そしてその情報をネットワークを介してサーバ装置１１０から取得し（Ｓ１８０３）、サーバ装置１１０はクライアント装置１２０からの要求に応じてサーバ装置のデータベース１１５から登録情報を読み出してクライアント装置１２０に送信する（Ｓ１８０５）。最後に取得した情報をユーザに提示して（Ｓ１８０４）終了する。
【０１３１】
図１９は、ユーザがシステムを利用する時の例を説明する図である。上記に説明した方法により、クライアント装置１２０上で情報と情報が属する基本グループの分類が行われ、ユーザはその分類結果に応じてサーバ装置１１０に登録されている情報を取得することができる。
【０１３２】
具体的には次のようにユーザはシステムを利用する。まず、図６で説明したようにクライアント装置１２０を起動してシステムの利用をスタートする。ここで、既にクライアント装置１２０が起動している場合は、ユーザがログインするのみで処理が実行されるようにすることができる。次にユーザが情報の分類を要求し、図１４で説明した情報分類処理により登録情報を分類する（Ｓ１９０１）。次にユーザがグループの分類を要求し、図１６で説明したグループ分類処理により登録情報が属する基本グループを分類する（Ｓ１９０２）。次にユーザが分類結果の表示を要求し、クライアント装置１２０の情報管理部により、分類結果がクライアント装置１２０の画面等に表示される（Ｓ１９０３）。さらにユーザが特定の情報の取得を要求し、図１８で説明した情報管理処理においてユーザが選択した情報をサーバ装置１１０から取得し、ユーザに提示する（Ｓ１９０４）。次にユーザが新しく追加された情報を自動的に設定する自動送受信の設定を要求し、第４実施形態で説明する方法により自動送受信設定を行い（Ｓ１９０５）、最後に情報管理部によりユーザの利用履歴をログとしてクライアント装置のデータベースに格納して、システムを終了するか、ユーザがログオフを行う。
【０１３３】
なお、図１９で説明した手順はあくまで例であり、要求を入れ替えたり、一部の要求だけとすることもできる。また、情報やグループの分類は、クライアント装置１２０上で自動的に実行するように設定しておけば、ユーザは処理の時間待つことなく最新の分類結果を取得することができる。
【０１３４】
（第２実施形態）
情報がシステムに追加される場合の分類処理について図面を参照して説明する。
【０１３５】
図２０は、情報の追加処理の流れを説明する概念図である。サーバ装置１１０で情報の追加があった場合に、サーバ装置１１０の情報管理処理においてシステムに情報を登録し、登録情報３１２、基本グループ３１５および更新のログ３１３をサーバ装置のデータベース１１５に格納する。次にサーバ装置の特徴量計算処理Ｓ２００２により、各登録情報３１２の特徴量をすべて計算して再処理するか、追加された情報についてだけ計算して追加処理を行って特徴量３１４をデータベース１１５に格納する。そして、サーバ装置の情報管理処理Ｓ２００１において変更をクライアント装置１２０に送信し、クライアント装置１２０の情報管理処理Ｓ２００４において更新のログ３２３、メタ情報３２２、情報の特徴量３２４および基本グループ３２５をクライアント装置のデータベース１２５に格納してサーバ装置１１０はアイドル状態となる。
【０１３６】
もし、サーバ装置へ情報が追加された際にクライアント装置１２０が起動していない場合は、次にクライアント装置１２０が起動すると図６で説明したようにその追加が反映される。
【０１３７】
図２１は、情報の追加時のサーバ装置１１０の特徴量計算処理の追加処理を説明する図である。図２０における特徴量計算処理は、すべての登録情報について特徴量を再計算する場合と、追加された登録情報についてのみ特徴量を計算する場合がある。前者は、図９に説明したように処理を行い、後者は次に説明するように図２１に示す方法で行う。
【０１３８】
まず、サーバ装置１１０のデータベース１１５から追加された登録情報３１２を読み込む（Ｓ２１０１）。次に追加された登録情報について特徴量の計算を行う（Ｓ２１０２）。この時、前述した方法で特徴量を計算するときの参照情報は適宜データベースから取得する。最後に計算した情報の特徴量をデータベース１１５格納して（Ｓ２１０３）終了する。
【０１３９】
図２２は、情報の追加後のクライアント装置の情報分類処理の追加処理を説明する図である。情報追加後の情報分類は、登録情報３１２の特徴量がすべて再計算された場合は前述の図１４の方法で行い、追加された登録情報についてのみ特徴量が計算された場合は図２２に示す方法で行う。
【０１４０】
まず、クライアント装置１２０のデータベース１２５から更新のログ３２３を取得し（Ｓ２２０１）、追加された情報の特徴量３２４とその時点での分類結果３２６をデータベース１２５から取得する（Ｓ２２０２）。そして、分類結果に対して追加された情報を配置し（Ｓ２２０３）、これを新たな分類結果としてデータベース１２５に格納して（Ｓ２２０４）終了する。ここで、処理Ｓ２２０３は、自己組織化マップを用いる場合、分類の処理を行うことなしに、最後の特徴量ベクトルに最も近い参照ベクトルを探して配置することを行う。なお、分類の設定は以前の分類時の設定を用いる。
【０１４１】
図２３は、情報の追加後のグループの分類部の追加処理を説明する図である。情報追加後のグループ分類は、登録情報の特徴量がすべて再計算された場合は前述の図１６に示す方法で行い、追加された登録情報についてのみ特徴量が計算された場合は図２３に示す方法で行う。ただしこの場合、追加された登録情報によって新たな基本グループが作成されない時は、グループ分類の追加処理は行わない。
【０１４２】
まず、クライアント装置のデータベース１２５から更新のログを取得し（Ｓ２３０１）、追加された情報の特徴量、新たに追加された基本グループおよびその時点でのグループの分類結果をデータベース１２５から取得する（Ｓ２３０２）。次に追加された基本グループの特徴量を計算し（Ｓ２３０３）、分類結果に対して追加された基本グループを配置し（Ｓ２３０４）、追加された基本グループ内の分類を行う（Ｓ２３０５）。最後に最後の分類結果をデータベース１２５に格納して（Ｓ２３０６）終了する。
【０１４３】
なお、基本グループの特徴量や分類の方法は図１６で説明した方法を用いる。なお、分類基準等の設定は以前の分類時に設定した設定を用いる。
【０１４４】
以上のように、情報が新たにサーバ装置に追加された場合でも、情報と基本グループの分類を行うことができる。また、既に登録されている情報が更新された場合においても、同様の方法で適用することができる。
【０１４５】
（第３実施形態）
次に、前述の方法を応用した所有者の分類について説明する。図１６、１７において説明した基本グループ内の分類を用いれば、情報の所有者を分類することができる。すなわち、所有者毎に所有者が持つ登録情報の特徴量ベクトルを平均した平均ベクトルを用いて、情報の分類と同様の方法を用いて、所有者を分類することができる。分類に自己組織化マップを用いるならば、この場合も情報の分類と同様に、分類の中心にしたい所有者の平均ベクトルをマップの中央に位置するようにして分類を行うことにより、特定の所有者を中心とした分類が可能となる。
【０１４６】
（第４実施形態）
サーバ装置１１０とクライアント装置１２０との間での情報の自動送受信について、図面を参照して説明する。自動送受信は、サーバ装置１１０に情報が新たに追加された場合に、ユーザが欲する情報を自動的にクライアント装置１２０に送信するという機能である。
【０１４７】
図２４は、自動送受信の設定を説明する該略図である。まず、クライアント装置データベース１２５から情報とグループの分類結果を取得する（Ｓ２４０１）。次に、自動的に取得する情報の対象を選択し、自動送受信設定３２８としてデータベース１２５に格納する（Ｓ２４０２）。そして、この自動送受信設定３２８をサーバ装置１１０に登録し（Ｓ２４０３）、サーバ装置１１０では自動送受信設定３１８をサーバ装置１１０のデータベース１１５に格納する。
【０１４８】
ここで、自動的に取得する情報の対象の選択では、ある特定の情報の特徴量ベクトルおよび取得する情報の特徴量ベクトルの距離、または特定の基本グループもしくは特定の基本グループの特徴量ベクトルとの距離、または特定の所有者、もしくは特定の所有者との距離を指定して自動送受信設定とし、それを満たす物を取得することとする。
【０１４９】
図２５は、自動送受信の方法の第１の例を説明する図である。ここでは図面の簡単化の為に、各データベースを省略して説明する。
【０１５０】
今、クライアント装置Ｂがサーバ装置に新たに情報を登録し、これに対し、クライアント装置Ａが自動送受信設定を行っているとする。まず、クライアント装置Ｂがサーバ装置に情報を登録する（Ｓ２５０１）。サーバ装置では図２０で説明した情報の追加処理Ｓ２５０２を行い、そこで得られた情報の特徴量ベクトル等と自動送受信設定を比較して、クライアント装置Ａへの送信の判断を行う（Ｓ２５０３）。設定を満たした場合サーバ装置はクライアント装置Ａに新着情報を通報し（Ｓ２５０４）、クライアント装置Ａはその情報を取得するかの判断を行う（Ｓ２５０５）。ここで、自動送受信設定３２８を行っていても、処理Ｓ２５０５によってクライアント装置Ａは情報の取得を拒絶することができる。情報を取得する場合は、情報の発信者であるクライアント装置Ｂに情報を送信できるかどうかの問合を行い（Ｓ２５０６）、クライアント装置Ｂはそれに答える（Ｓ２５０７）。この際、送信できるかどうかは、クライアント装置Ｂが直接送信するかどうか、またはクライアント装置Ａからサーバ装置とクライアント装置Ａからクライアント装置Ｂまでのネットワークの状態などから判断する。クライアント装置Ａがクライアント装置Ｂから情報を取得する場合は処理Ｓ２５０８とＳ２５０９が実行され、クライアント装置Ａがサーバ装置から情報を取得する場合は処理Ｓ２５１０とＳ２５１１が実行されて処理が終了する。
【０１５１】
図２６は、自動送受信の方法の第２の例を説明する該略図である。ここでは図面の簡単化の為に、各データベースを省略して説明する。図２５と同様に、クライアント装置Ｂがサーバ装置に新たに情報を登録し、これに対し、クライアント装置Ａが自動送受信設定を行っているものとする。
【０１５２】
まず、クライアント装置Ｂがサーバ装置に情報を登録する（Ｓ２６０１）。サーバ装置では図２０で説明した情報の追加処理Ｓ２６０２を行い、新着情報をクライアント装置Ａに通報する（Ｓ２６０３）。クライアント装置Ａでは図２２と図２３で説明した情報の分類Ｓ２６０４とグループの分類Ｓ２６０５を行い、情報を取得するかどうかの判断を行う（Ｓ２６０６）。これは自動送受信設定３１８に従って行なわれる．次に情報を取得する場合は、情報の発信者であるクライアント装置Ｂに情報を送信できるかどうかの問合せを行い（Ｓ２６０７）、クライアント装置Ｂはそれに答える（Ｓ２６０８）。この際、送信できるかどうかについては、クライアント装置Ｂが直接送信するかどうか、またはクライアント装置Ａからサーバ装置までもしくはクライアント装置Ａからクライアント装置Ｂまでのネットワークの状態などから判断する。クライアント装置Ａがクライアント装置Ｂから情報を取得する場合は処理Ｓ２６０９とＳ２６１０によって実行し、クライアント装置Ａがサーバ装置から情報を取得する場合は処理Ｓ２６１１とＳ２６１２によって実行されて処理を終了する。
【０１５３】
以上のように、ユーザは特に意識して処理することなく、ユーザに関連する最新の登録情報を取得することができる。
【０１５４】
【発明の効果】
以上説明したように、本発明によれば、複数の分類対象情報を含む分類対象情報群を格納するサーバ記憶手段、複数の分類対象情報の各々の間で情報特徴量を相互に算出してサーバ記憶手段に格納するサーバ制御手段、および情報特徴量送信要求を受信し、サーバ記憶手段から読み出された情報特徴量を送信するサーバ通信手段を含むサーバ装置と、ユーザによって設定された、情報グループを分類する場合基準となる分類基準を予め格納するクライアント記憶手段、サーバ装置に情報特徴量送信要求を送信し、情報特徴量をサーバ装置から受信するクライアント通信手段、およびユーザの要求によりまたは所定の条件が満たされると自動的に、クライアント記憶手段から分類基準を読み出し、通信手段において受信した情報特徴量に基づいて当該読み出された分類基準を用いて分類対象情報を分類し、分類された結果をクライアント記憶手段に格納するクライアント制御手段を含むクライアント装置とを備えているため、情報やユーザの位置づけがわかるような分類を実現することができ、また分類結果に関連する情報を受動的にユーザが受け取ることができる。
【図面の簡単な説明】
【図１】本実施形態の分類システムの全体構成を説明する図である。
【図２】本実施形態のシステムが動作する場合の配置構成を説明する図である。
【図３】本実施形態のサーバ装置とクライアント装置のデータベースについて説明する図である。
【図４】本実施形態のシステムのサーバ装置の起動時の処理を説明する図である。
【図５】本実施形態のサーバ装置起動時の情報管理処理を説明する図である。
【図６】本実施形態のシステムのクライアント装置の起動時の処理を説明する図である。
【図７】クライアント装置起動時の情報管理部を説明する図である。
【図８】本実施形態のシステムの分類対象となる情報の定義を説明する図である。
【図９】サーバ装置の特徴量計算処理を説明する図である。
【図１０】情報特徴量を計算するための参照ツリーの作成方法を説明する図である。
【図１１】参照ツリーへ新しいセルを追加する方法を説明するための参照ツリーを示す図である。
【図１２】参照ツリーの解析による情報特徴量の算出方法を説明するための図である。
【図１３】参照ツリーの比較による情報特徴量の算出方法を説明するための図である。
【図１４】クライアント装置の情報分類処理について説明する図である。
【図１５】自己組織化マップによる情報の分類方法を説明する図である。
【図１６】クライアント装置のグループ分類処理について説明する図である。
【図１７】基本グループの特徴量の例について説明する図である。
【図１８】本実施形態のクライアント装置がサーバ装置から登録情報を取得するときの情報管理処理について説明する図である。
【図１９】ユーザがシステムを利用する時の例を説明する図である。
【図２０】情報の追加処理の流れを説明する概念図である。
【図２１】情報の追加時のサーバ装置の特徴量計算処理の追加処理を説明する図である。
【図２２】情報の追加後のクライアント装置の情報分類処理の追加処理を説明する図である。
【図２３】情報の追加後のグループの分類部の追加処理を説明する図である。
【図２４】自動送受信の設定を説明する該略図である。
【図２５】自動送受信の方法の第１の例を説明する図である。
【図２６】自動送受信の方法の第２の例を説明する該略図である。
【符号の説明】
１１０ サーバ装置
１２０ クライアント装置
１１１、１２１ 情報管理部
１１２ 特徴量計算部
１１３、１２４ 自動送受信部部
１２２ 情報分類部
１２３ グループ分類部
１１５、１２５ データベース
２０１、２０２、２０３ クライアント装置
２２０ ネットワーク
３１１、３２１ ユーザ設定
３１２ 登録情報
３１３、３２３ ＬＯＧ
３１４、３２４ 特徴情報量
３１５、３２５ 基本グループ
３１８、３２８ 自動送受信設定
３２２ メタ情報
３２６、３２７ 情報、グループ分類結果
８０１ 情報
８０２ 情報名部
８０３ 参照情報部
８０４ キーワード部
８０５ 内容部
８０６ 所有者
８０７ グループ部
８０８ 情報群
１００１〜１０１０、１１０１〜１１０９ セル
１０１１〜１０１４、１１１０ 階層
１２０１ 参照ツリー
１２０２ 解析結果の表
１２０３ 類似度ベクトル
１３０１ 類似度計算式
１３０２ 類似度ベクトルの表
１５０１ 自己組織化マップ
１５０２ 近傍集合
１７０１ グループ名
１７０２ 代表ベクトル
１７０３ 重み
１７０４〜１７０６ 所有者
１７０７ 特徴量ベクトル[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an information classification method and apparatus for classifying information, and more particularly to an information classification method and apparatus for classifying information based on the similarity between pieces of information included in a classification target information group.
[0002]
[Prior art]
As a conventional information classification system, in particular, a system for automatically classifying using the appearance frequency of words appearing in a document as a document classification system, there is a “sentence automatic classification system” (for example, see Patent Document 1). This is classified by the following method. First, an appearance frequency vector composed of word appearance frequencies in a document is formed by statistically processing the word appearance frequencies in the input sentence. Then, using this appearance frequency vector, the sentence to be classified is divided into clusters based on a predetermined information amount criterion, and the division is repeated until the divided cluster cannot be further divided. I do.
[0003]
On the other hand, there is a “user adaptive information distribution system” (see, for example, Patent Document 2) that enables a user to selectively receive desired information as an information push type service. This provides information to the user in the following manner. Information indicating the statistical characteristics of the document information accessed by the user in the client is input as the user's access characteristics, the statistical characteristics are extracted for each stored document information, and the user's access characteristics and extraction results are extracted. Based on this, the similarity between the document information accessed by the user and each of the document information to be distributed is evaluated, and the document information having a high similarity is selectively distributed to the user according to the evaluation result.
[0004]
[Patent Document 1]
JP-A-8-263510
[Patent Document 2]
JP-A-11-232287
[0005]
[Problems to be solved by the invention]
However, the conventional information classification method has the following problems.
[0006]
In the "automatic sentence classification system" that automatically classifies using the appearance frequency of words appearing in the above-mentioned document, a key for classifying information such as sentences is extracted, and the similarity is calculated from the key. Since the clustering is performed, the documents are only classified into appropriate groups. However, with this method, although the overall positioning and classification of information can be understood, it is difficult for the user to understand how much information is related and how difficult it is to use information effectively.
[0007]
On the other hand, in the “user adaptive information distribution system” that enables a user to selectively receive desired information, when the user passively receives information, the user needs to access the information to be associated. There is a problem that it is difficult to distribute related information in a push type using the classification result of information as in the “automatic text classification system”.
[0008]
Accordingly, the present invention has been made in view of the above-described conventional problems, and the object of the present invention is to realize classification that allows information and user positioning to be understood by focusing on classification of information. It is another object of the present invention to provide an information classification system and method that enables a user to passively receive information related to the classification result.
[0009]
[Means for Solving the Problems]
In order to achieve such an object, the invention according to claim 1 is the information classification system, wherein the server storage means for storing the classification target information group including the plurality of classification target information, each of the plurality of classification target information Server control means for mutually calculating information feature quantities between them and storing them in the server storage means, and server communication means for receiving the information feature quantity transmission request and transmitting the information feature quantities read from the server storage means The server device, client storage means for storing in advance a classification criterion that is set by the user and used as a criterion for classifying the information group, transmits an information feature amount transmission request to the server device, and receives the information feature amount from the server device The client communication means and automatically read the classification criteria from the client storage means upon user request or when a predetermined condition is met A client device including client control means for classifying the classification target information using the read classification criteria based on the information feature quantity received in the communication unit and storing the classified result in the client storage means. It is characterized by that.
[0010]
According to a second aspect of the present invention, in the information classification system according to the first aspect, the classification standard includes an instruction of central information required to be classified as the center of the classification target information, and the client control means It is characterized by classifying information indicated in the information instruction as a center.
[0011]
According to a third aspect of the present invention, in the information classification system according to the first or second aspect, the classification target information includes information content and an owner, and the client control means includes a classification target between each of the classification target information. The content and owner of the information included in each piece of information are classified based on information feature amounts calculated mutually.
[0012]
According to a fourth aspect of the present invention, in the information classification system, a server storage unit that stores a classification target information group including a plurality of classification target information, and an information feature amount is calculated between each of the plurality of classification target information. A server control unit that stores the information in the server storage unit and generates a plurality of information groups in which the classification target information included in the classification target information group is classified according to a predetermined rule, and stores the information group in the server storage unit, and an information feature amount transmission request And a server apparatus including server communication means for transmitting the information feature quantity read from the server storage means and a plurality of information groups, and a classification standard set by the user as a standard for classifying the information groups An information feature amount transmission request is transmitted to the client storage means and the server device stored in advance, and the information feature amount and the plurality of information groups are transmitted to the server device. The client communication means received from the server, and the classification criteria are automatically read from the client storage means when requested by the user or when a predetermined condition is satisfied, and is calculated using the information feature quantity and the plurality of information groups received by the communication unit And a client control unit that classifies a plurality of information groups using the classification criteria read out based on the group information feature amount and stores the classified results in the client storage unit.
[0013]
According to a fifth aspect of the present invention, in the information classification system according to the fourth aspect, the classification standard includes an instruction of central information required to be classified as the center of the classification target information, and the client control means An information group including information indicated in the information instruction is classified as a center.
[0014]
According to a sixth aspect of the present invention, in the information classification system according to the fourth or fifth aspect, the information to be classified includes the content and owner of the information, and the client control means includes the information group between each of the information groups. The content and the owner of the information included in each are classified based on the mutually calculated group information feature amounts.
[0015]
According to a seventh aspect of the present invention, in the information classification system according to any one of the first to sixth aspects, the classification standard includes a feature amount adjustment value indicating an adjustment amount when the information feature amount is adjusted for each user. The client control means classifies the information feature quantity based on the value adjusted by the feature quantity adjustment value.
[0016]
The invention according to claim 8 is the information classification system according to any one of claims 1 to 7, wherein the predetermined condition is that additional information that is new information is added to the target information group, and the client The control means classifies based on the information feature amount calculated between the additional information and each of the classification target information.
[0017]
The invention according to claim 9 is the information classification system according to any one of claims 1 to 7, wherein the predetermined condition is that additional information as new information is added to the target information group, and the client The communication unit receives the additional information from the server device, and the client control unit includes the received additional information in the target information based on the already received information feature amount and classifies the information.
[0018]
A tenth aspect of the present invention is the information classification system according to any one of the first to seventh aspects, wherein the server control unit is an object in which the client device stores the information feature amount at a predetermined time interval. The classification included in the target information group that is calculated by adding each additional information to the target information group, or calculating each other among the classification target information included in the information group. It is characterized by mutually calculating between each piece of object information.
[0019]
According to an eleventh aspect of the present invention, in the information classification system according to any one of the first to tenth aspects, when the server device receives a notification that there is additional information from one of the client devices, When a client device other than the client device that has notified is notified that there is additional information, the client device requests the additional information from either the server device or the client device that transmitted the notification that there is additional information. It is characterized by receiving additional information.
[0020]
According to the twelfth aspect of the present invention, in the server device, an information feature amount is calculated between the server storage unit that stores the classification target information group including the plurality of classification target information and each of the plurality of classification target information. Server control means for storing in the server storage means, and server communication means for receiving the information feature quantity transmission request and transmitting the information feature quantity read from the server storage means.
[0021]
According to a thirteenth aspect of the present invention, in the server device, the server storage means for storing the classification target information group including the plurality of classification target information and each of the plurality of classification target information mutually calculate information feature amounts. A server control unit that stores the information in the server storage unit and generates a plurality of information groups in which the classification target information included in the classification target information group is classified according to a predetermined rule, and stores the information group in the server storage unit, and an information feature amount transmission request And server communication means for transmitting the information feature amount read from the server storage means and a plurality of information groups.
[0022]
According to a fourteenth aspect of the present invention, in the server device according to the twelfth or thirteenth aspect, the server control unit includes a classification target included in the target information group stored at the time of calculation of the information feature amount at predetermined time intervals. Between each of the classification target information included in the target information group added by adding additional information to the target information group each time information is calculated between each other or added. It is characterized by mutual calculation.
[0023]
According to the fifteenth aspect of the present invention, in the client device, an information feature amount transmission request is transmitted, and information feature amounts calculated mutually among a plurality of pieces of classification target information included in the classification target information group are received. Client communication means, client storage means for preliminarily storing a classification standard that is set by the user and used as a reference when classifying a plurality of classification target information, and automatically when a user's request or a predetermined condition is satisfied, Client control means for reading out classification criteria from the client storage means, classifying the classification target information using the classification criteria read out based on the information feature received by the communication unit, and storing the classified results in the client storage means It is characterized by comprising.
[0024]
According to a sixteenth aspect of the present invention, in the client device according to the fifteenth aspect, the classification criterion includes an indication of central information that is required to be classified as the center of the classification target information, and the client control means includes the central information It is characterized in that the information shown in the instruction is classified mainly.
[0025]
According to a seventeenth aspect of the present invention, in the client device according to the fifteenth or sixteenth aspect, the classification target information includes information content and an owner, and the client control means includes the classification target information between each of the classification target information. The contents and owner of the information included in each of the information are classified based on information feature amounts calculated mutually.
[0026]
According to an eighteenth aspect of the present invention, in the client device, an information feature amount transmission request is transmitted, and an information feature amount calculated between each of a plurality of pieces of classification target information included in the classification target information group, and a classification Client communication means for receiving a plurality of information groups obtained by classifying classification target information included in the target information group according to a predetermined rule, and a classification standard set by the user as a standard for classifying the plurality of classification target information in advance The client storage means to be stored, and when the user's request or a predetermined condition is satisfied, the classification criteria are automatically read from the client storage means and calculated using the information feature quantity and the plurality of information groups received in the communication unit. A plurality of information groups are classified using the read classification criteria based on the group information feature amount. Characterized in that the results and a client control means for storing in the client storage means.
[0027]
According to a nineteenth aspect of the present invention, in the client device according to the eighteenth aspect, the classification criterion includes an indication of central information that is required to be classified as the center of the classification target information, and the client control means includes the central information. It classifies | categorizes centering on the information group containing the information shown in instruction | indication.
[0028]
20. The classification target information in the client device according to claim 18 or 19, including the content and owner of the information, and the client control means includes the content of the information included in each of the information groups among the information groups. And the owners are classified based on the mutually calculated group information feature amounts.
[0029]
According to a twenty-first aspect of the present invention, in the client device according to any one of the fifteenth to twentieth aspects, the classification criterion includes a feature amount adjustment value indicating an adjustment amount when the information feature amount is adjusted for each user, The client control means classifies the information feature quantity based on a value adjusted by the feature quantity adjustment value.
[0030]
According to a twenty-second aspect of the present invention, in the client device according to any one of the fifteenth to twenty-first aspects, the predetermined condition is that additional information that is new information is added to the target information group. The means is characterized in that the classification is performed based on the information feature amount calculated between the additional information and each of the classification target information.
[0031]
According to a twenty-third aspect of the present invention, in the client device according to any one of the fifteenth to twenty-first aspects, the predetermined condition is that additional information that is new information is added to the target information group. The means receives the additional information, and the client control means includes the received additional information in the target information based on the already received information feature quantity and classifies the information.
[0032]
The invention according to claim 24 is a program for executing a step of classifying classification target information based on information feature amounts calculated mutually between the classification target information included in the classification target information group, A feature quantity storing step in which the server control means stores the information feature quantity between each of the plurality of classification target information and the server storage means stores it, and an information feature quantity transmission request is received and read from the server storage means And a transmission step in which the server communication means transmits the information feature amount.
[0033]
The invention according to claim 25 is a program for executing a step of classifying the classification target information based on information feature amounts calculated mutually between the classification target information included in the classification target information group. A plurality of information feature quantities calculated between each of a plurality of classification target information, stored in the server storage means by the server control means, and classified target information included in the classification target information group according to a predetermined rule The feature amount storing step of generating the information group and storing it in the server storage means and the information feature amount transmission request are received, and the server communication means reads out the feature information amount and the plurality of information groups from the server storage means and transmits them. And a transmission step.
[0034]
The invention according to claim 26 is the program according to claim 24 or 25, wherein the feature amount storing step is included in the target information group in which the information feature amount is stored at a predetermined time interval. Each of the classification target information included in the target information group that is calculated by mutual calculation between each of the classification target information or adding additional information to the target information group each time additional information is added It is characterized by being calculated mutually.
[0035]
The invention according to claim 27 is a program for executing a step of classifying classification target information based on information feature amounts calculated mutually between the classification target information included in the classification target information group, A feature amount receiving step in which an information feature amount is transmitted between each of a plurality of pieces of classification target information included in the classification target information group by transmitting an information feature amount transmission request, and set by the user A classification criteria storage step in which the client storage means stores in advance the classification criteria that serve as the criteria for classifying a plurality of classification target information, and the client control means automatically upon request of the user or when a predetermined condition is satisfied In addition, the classification standard is read from the client storage unit, and the classification standard read based on the information feature amount received by the communication unit is used for the classification. Classifying the target information, and executes the classification result storing step of storing the classified results to the client storage means.
[0036]
The invention according to claim 28 is the program according to claim 27, wherein the classification standard includes an instruction of center information required to be classified as the center of the classification target information, and the classification result storing step includes: It is characterized in that the information shown in the center information instruction is classified into the center.
[0037]
The invention according to claim 29 is the program according to claim 27 or 28, wherein the classification target information includes the content and owner of the information, and the classification storage step includes a classification target between each of the classification target information. The content and owner of the information included in each piece of information are classified based on information feature amounts calculated mutually.
[0038]
The invention according to claim 30 is a program for executing a step of classifying the classification target information based on information feature amounts calculated mutually between the classification target information included in the classification target information group, The information feature amount transmission request is transmitted, and the information feature amount calculated between each of the plurality of classification target information included in the classification target information group and the classification target information included in the classification target information group are determined according to a predetermined rule. A feature receiving step in which the client communication means receives a plurality of information groups classified by the client, and a classification criterion in which the client storage means stores in advance a classification criterion that is set by the user and serves as a reference when classifying the plurality of classification target information The storing step and the client control means are automatically separated from the client storage means by a user request or when a predetermined condition is satisfied. A classification result storage that reads out the reference, classifies the classification target information using the classification feature read based on the information feature amount and the plurality of information groups received in the communication unit, and stores the classified result in the client storage unit And executing the step.
[0039]
The invention according to claim 31 is the program according to claim 30, wherein the classification criterion includes an instruction of center information required to be classified as the center of the classification target information, and the classification result storing step includes: An information group including information indicated in the instruction of the center information is classified as a center.
[0040]
A thirty-second aspect of the invention is the program according to the thirty-third or thirty-first aspect, wherein the classification target information includes the content and owner of the information, and the classification result storing step is performed between the information groups. The contents and owners of the information included in each are classified based on mutually calculated group information feature amounts.
[0041]
A thirty-third aspect of the invention is the program according to any one of the twenty-seventh to thirty-second aspects, wherein the classification standard includes a feature amount adjustment value indicating an adjustment amount when the information feature amount is adjusted for each user. The classification result storing step is characterized in that the information feature amount is classified based on a value adjusted by the feature amount adjustment value.
[0042]
The invention according to claim 34 is the program according to any one of claims 27 to 33, wherein the predetermined condition is that additional information which is new information is added to the target information group, and classification The result storing step is characterized in that classification is performed based on the information feature amount calculated between the additional information and each of the classification target information.
[0043]
The invention according to claim 35 is the program according to any one of claims 27 to 33, wherein the predetermined condition is that additional information as new information is added to the target information group, and the client The communication means further executes a step of receiving additional information, and the classification result storing step is characterized in that the received additional information is included in the target information and classified based on the already received information feature quantity.
[0044]
A thirty-sixth aspect of the present invention is a computer-readable recording medium in which each program according to the twenty-fourth to thirty-fifth aspects is recorded.
[0045]
The invention according to claim 37 is the information classification method for classifying the classification target information based on the information feature amount calculated mutually between the classification target information included in the classification target information group. A setting step for presetting and storing a classification standard as a standard for classification, and a classification standard that is automatically read out and read out when new information is added to the classification target information group upon user request. And a classification step of classifying the classification target information by using.
[0046]
According to a thirty-eighth aspect of the present invention, in the information classification method according to the thirty-seventh aspect, the classification criterion includes an instruction of central information that is required to be classified as the center of the classification target information, and the classification step includes the central information. It is characterized in that the information shown in the instruction is classified mainly.
[0047]
The invention according to claim 39 is an information classification method for classifying classification target information based on information feature amounts calculated mutually between the classification target information included in the classification target information group. Generating a plurality of information groups by classifying the classification target information included in the information according to a predetermined rule, a setting step for presetting and storing a classification criterion as a reference when the information group is classified, and a user request Or, when a predetermined condition is satisfied, the classification standard is automatically read out, and the plurality of information groups are used based on the information feature amount calculated mutually between the plurality of information groups. And a classification step for classifying the data.
[0048]
The invention according to claim 40 is the information classification method according to claim 39, wherein the classification standard includes an instruction of center information required to be classified as the center of the classification target information, and the classification step includes the center information. It classifies | categorizes centering on the information group containing the information shown in instruction | indication.
[0049]
The invention according to claim 41 is the information classification method according to any one of claims 37 to 40, wherein the classification target information includes information contents and an owner, and the classification step is performed between each of the classification target information. And classifying information contents and owners of information included in each of the classification target information based on mutually calculated information feature amounts.
[0050]
The invention according to claim 42 is the information classification method according to any one of claims 37 to 41, wherein the classification standard includes a feature amount adjustment value indicating an adjustment amount when the information feature amount is adjusted for each user. The classification step is characterized in that the information feature amount is classified based on a value adjusted by the feature amount adjustment value.
[0051]
The invention according to claim 43 is the information classification method according to any one of claims 37 to 42, wherein the predetermined condition is that additional information which is new information is added to the target information group, and the classification is performed. The step is characterized in that the classification is performed based on the information feature amount calculated between the additional information and each of the classification target information.
[0052]
According to a 44th aspect of the present invention, in the information classification method according to any of the 37th to 42nd aspects, the information feature amount is included in the target information group stored at the time of calculation at every predetermined time interval. Each of the classification target information included in the target information group that is calculated by mutual calculation between each of the classification target information or adding additional information to the target information group each time additional information is added It is characterized by being calculated mutually.
[0053]
According to a 45th aspect of the present invention, in the information classification method according to any of the 37th to 44th aspects, when the server device receives a notification that there is additional information from one of the client devices, the notification is provided. Additional information notification step for notifying the client device other than the client device that there is additional information, and when the client device requests additional information, the server device and the client device that has transmitted the notification that there is additional information. And an additional information receiving step of receiving additional information from any of them.
[0054]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described with reference to the drawings.
(First embodiment)
FIG. 1 is a diagram for explaining the overall configuration of the classification system of the present embodiment. The system according to the present embodiment is composed of one server device and a plurality of client devices, each of which has a database. However, there may be a configuration in which there are a plurality of server devices and operate in cooperation. In the present embodiment, a case where there is one server device will be described.
[0055]
The server device 110 includes an information management unit 111, a feature amount calculation unit 112, and an automatic transmission / reception unit 113. The client device 120 includes an information management unit 121, an information classification unit 122, a group classification unit 123, and an automatic transmission / reception unit 124.
[0056]
In the literary system of this embodiment, the information management unit 111 of the server device 110 collectively manages the information to be classified, and the feature amount calculation unit 112 calculates the feature amount of the information. The user obtains information registered in the server device and its feature amount via the network by the information management unit 121 of the client device 120. In addition, the information classification unit 122 and the group classification unit 123 classify information and groups. Each classification result can be displayed and viewed on each client device by the information management unit 121.
[0057]
In addition, the automatic transmission / reception units 113 and 124 of the server apparatus 110 and the client apparatus 120 realize a function of automatically informing the user of newly registered information via the network and acquiring the information. Details will be described later.
[0058]
The information to be classified, the feature amount of the information, the classification result, and the like are stored in the databases 115 and 125 of the server device and the client device, and each unit is appropriately read and used.
[0059]
FIG. 2 is a diagram illustrating an arrangement configuration when the system of this embodiment operates. Components similar to those in FIG. 1 are denoted by the same reference numerals. The server apparatus 110 and the client apparatuses A 201, B 202, and C 203 are connected to the network 220, and the server apparatus 110 and the client apparatus 120 exchange information via the network 220 using the information management units 111 and 121. .
[0060]
The client devices A, B, and C communicate with the server device 110 through network paths 211, 212, and 213, respectively. In the automatic transmission / reception units 113 and 124, the client apparatuses directly exchange information through the network path 214 and the like. Details will be described in a fourth embodiment.
[0061]
FIG. 3 is a diagram for explaining the databases of the server device 115 and the client device 125. Components similar to those in FIG. 1 and the like are denoted by the same reference numerals. The server database 115 stores user settings 311 of users participating in the system and registration information 312 to be classified. Here, the user setting is an example of what is a category standard when performing classification, and includes at least a user name and information owned by the user.
[0062]
These are registered and changed by the information management unit of the server apparatus 110, and the update state is stored as a log 313. Further, the server database 115 includes a feature amount 314 of information calculated by the feature amount calculation unit 112 of the server device 110, a basic group 315 generated by the information management unit 111 of the server device 110, and an automatic transmission / reception unit 113 of the server device 110. The automatic transmission / reception setting 318 set in is stored. These data are read and written by each unit as appropriate. Here, the automatic transmission / reception setting is a setting for selecting which information is automatically acquired, for example, specific information and a threshold value of a feature vector from the information.
[0063]
The client database 125 stores user settings 321 of users who use the system. In addition, a log 323 representing an update state when the meta information 322 such as an owner included in the registration information 312 acquired and generated by the information management unit of the client device and the registration information 312 are added or changed, and the server device The information feature quantity 324 acquired from 110 and the basic group 325 acquired from the server apparatus 110 are stored.
[0064]
Furthermore, the information classification result 326 calculated by the information classification unit of the client device, the basic group classification result 327 calculated by the group classification unit 123 of the client device 120, and the automatic transmission / reception set by the automatic transmission / reception unit 125 of the client device 120 Settings 328 are stored. These data are read and written by each unit as appropriate.
[0065]
FIG. 4 is a diagram illustrating processing at the time of starting the server device of the system according to the present embodiment. Components similar to those in FIG. 1 and the like are denoted by the same reference numerals. First, it is assumed that information to be classified is stored in the database 115 of the server device 110. When the server apparatus 110 is activated for the first time, registration information 312 and a basic group 315 are generated from information to be classified on the database 115 and stored in the information management process S301. Next, the feature amount of each information is calculated in the feature amount calculation process S302, the feature amount is stored in the database 115, and the idle state, which is a waiting state for the next processing, is entered.
[0066]
In addition, when there is no information in the database 115 of the server apparatus 110, it will be in an idle state, without performing the above process.
[0067]
In the second and subsequent activations of the server device 110, the information management unit 111 reads the system state such as the registration information 312 and the log 313 and enters an idle state.
[0068]
FIG. 5 is a diagram for explaining information management processing when the server apparatus according to the present embodiment is activated. Components similar to those in FIG. 1 and the like are denoted by the same reference numerals. First, classification target information is read from the server database 115 (S501). Next, a basic group 315 is created from the group information to which each information belongs (S502). At this time, information belonging to the basic group is listed and associated with the basic group. Next, each piece of information to be classified is registered information 312, and this registration information 312 and the created basic group 315 are stored in the database 115, and an update log 313 is created and stored in the database 115 (S503). Exit.
[0069]
The basic group 315 is preferably added according to a specific rule when the user registers information or later by the system administrator. As these rules, in the case of a predetermined field or a paper defined in advance by an administrator, a technical field predetermined by an academic society can be used, but is not limited thereto.
[0070]
FIG. 6 is a diagram illustrating processing at the time of starting the client device of the system according to the present embodiment. Components similar to those in FIG. 1 and the like are denoted by the same reference numerals. When the client device 120 is activated, the information management unit 121 of the client device 120 and the information management unit 111 of the server device 110 perform user settings, logs, meta information, and information feature amounts in the databases 115 and 125 via the network 220. , The basic group is exchanged and is in an idle state waiting for a user request.
[0071]
FIG. 7 is a diagram for explaining the information management unit when the client device is activated. The client device 120 reads the user setting 321 from the database 125, confirms the user setting, registers if it is a new user, updates the setting if necessary, and stores the user setting in the database (S721). . Note that user login and the like are also realized using this user setting. When the user setting is changed or updated, the client device 120 notifies the server device 110 via the network 220 (S722). If the server device 110 is different from the user setting 311 of the server device 110, the user setting is performed. Is updated and stored in the database 115 of the server apparatus 110 (S711). Next, the client device 120 acquires the state of the server device 110 (S712), and in response to this, the server device 110 reads a log from the database 115 and transmits it to the client device 120 (S713). The client device 120 reads the log 323 from the database 125, compares it with the log 313 of the server device, acquires update information if the state of the server device 110 has been updated (S713), and the server device 110 updates from the database 115. The registered information 312, the information feature amount 314, and the basic group 315 are transmitted to the client device 120 (S713), the client device 120 stores the information in the database 125 (S725), and the information management process ends.
[0072]
FIG. 8 is a diagram for explaining the definition of information to be classified in the system of the present embodiment. The information 801 requires an information name part 802, an information owner 806, a group part 807 to which information for classifying information and a group belongs, and a content part 805 which is information contents. Furthermore, for the classification method described in the present embodiment, it is assumed to have a reference information part 803 for information related to information and a keyword part 804 for expressing information.
[0073]
In addition, the classification target information group 808, which is a set of information, is a combination of the classification target information and the information referred to by the information.
[0074]
FIG. 9 is a diagram illustrating the feature amount calculation processing of the server device. In the feature amount calculation processing, first, the registration information 312 in the database 115 of the server apparatus 110 is read (S901), the feature amount of the registration information 312 is calculated (S902), and the calculated feature amount 314 is stored in the database 115. (S903) The process ends.
[0075]
10 to 13 are diagrams for describing a specific example of the feature amount calculation processing of the server device in the case of calculating the information feature amount according to the present embodiment. Here, a method of calculating the information feature amount 314 from information referred to by information, information contents, and keywords will be described.
[0076]
FIG. 10 is a diagram illustrating a method for creating a reference tree for calculating the information feature quantity 314. A basic reference tree associates information referred to by information in a tree structure. A reference tree is created specifically for the information A1001 by the following method.
[0077]
It is assumed that information A1001 refers to information B1002, C1003, and D1004. That is, the information hierarchy 1011 represents the reference information of the information A1001. First, the information A1001 is linked to each information B1002, C1003, and D1004.
[0078]
Next, links are established from the information B1002 to the information C1005, E1006, and F1007 referred to by the information B1002. Here, each piece of information is called a cell of the reference tree, and a link from the information to the information is called a branch of the reference tree. As a result of generating the reference tree as described above, the generation of the reference tree is ended when a predetermined end condition is satisfied.
[0079]
The end conditions used in this embodiment are the following five end conditions as shown in FIG.
[0080]
End condition <1>: When information that appears once in the reference tree is referenced
End condition <2>: When information with date and time outside the specified range is referenced
End condition <3>: When the reference tree exceeds the specified hierarchy value
Termination condition <4>: When the number of cells included in the reference tree exceeds a predetermined number
End condition <5>: When the reference information does not exist in the information group
[0081]
When these five conditions are satisfied in all cells, the creation of the reference tree is finished. For example, since the information C has appeared once in the cell C1003, the information C is ended by the end condition <1>. In this embodiment, if the information H is information having a date and time outside the predetermined range, the information H referred to by the information E satisfies the end condition <2>, and thus the cell H1008 linked from the cell E1006. Terminates tree generation without further linking. Also, assuming that the predetermined hierarchy value indicated in the end condition <3> of this embodiment is 4, in FIG. 4, the hierarchy 1011 is the hierarchy value 1, the hierarchy 1012 is the hierarchy value 2, the hierarchy 1013 is the hierarchy value 3, and the hierarchy 1014 is Since the hierarchy value is 4, generation of the tree ends because the cell K1010 corresponding to the information K referred to by the information I satisfies the end condition <3>.
[0082]
Note that the termination condition is not limited to the above, and can be additionally modified according to the characteristics of the information. In the present embodiment, the reference tree is created for each hierarchy of the reference tree. Therefore, the reference tree does not become deeper only for specific reference information, and when information reappears in the reference tree, it appears at a position that is never shallower (hierarchy value is smaller) than the first occurrence of the information. Never do.
[0083]
FIG. 11 is a diagram illustrating a reference tree for explaining a method of adding a new cell to the reference tree. Since the reference tree generation method described above with reference to FIG. 10 uses a technique for expanding the reference information of information A on the tree, information that is not referred to by information A is not incorporated into the reference tree. Therefore, since information newer than information A is not added, it is necessary to incorporate in another method, that is, to add a cell corresponding to the new information to the reference tree. Such information is referred to as additional reference information in this specification.
[0084]
Similarly to the reference tree shown in FIG. 10, the reference tree shown in FIG. 11 creates a reference tree with the information A as the center of classification (cell A 1101 as the root cell). Since information A refers to information B, a link is established from cell A 1101 to cell B 1102. At this time, information having high similarity with information B is searched, and addition of a cell is executed when a predetermined cell addition condition (described later) is satisfied. In FIG. 11, a cell L1103 corresponding to new information L that is additional reference information is added to the same hierarchy 1110 as information B.
[0085]
For the information L, a tree is generated in the same manner as the above-described reference tree generation. The information L refers to the information A and the information M, but since the information A has already appeared, the generation of the reference tree is terminated by the termination condition <1>. Further, information M refers to information C, information N, and information F. Since the cell C1106 corresponding to the information C satisfies the end condition <1>, the generation of the tree ends here, the information N refers to the information O, and the cell O1108 corresponding to the information O satisfies the end condition <3>. Thus, the generation of the tree ends, and the cell F1109 corresponding to the information F satisfies the end condition <1>, and the generation of the tree ends.
[0086]
In the present embodiment, the following three conditions are defined as conditions for adding a cell to the reference tree.
Additional condition <1>: The number of cells added to one cell is less than or equal to a predetermined value
Additional condition <2>: The hierarchical value of the reference tree to which the cell is added is less than or equal to the value
Additional condition <3>: dissimilarity of information to be added is lower than a predetermined value
[0087]
The addition condition <1> defines the number of cells that can be added to each cell of the reference tree. The additional condition <2> defines in which hierarchy of the reference tree the cell can be added. For example, if the number of cells that can be added to one cell of the reference tree is 1, and the hierarchical value of the reference tree that can be added is 1, only the cells B1002, C1003, and the cell D1004 in the reference tree shown in FIG. Each cell can be added one by one.
[0088]
The additional condition <3> specifies that the dissimilarity between the information already appearing in the reference tree and the cell to be added needs to be a certain level or less. That is, even if the conditions <1> and <2> are satisfied, information that is not very similar (dissimilarity is more than a certain level) has no meaning to be added to the reference tree.
[0089]
The additional conditions are not limited to the above, and additional corrections can be made according to the characteristics of the information. In this embodiment, addition of a cell to the reference tree is performed when a cell is created in each layer when creating the reference tree.
[0090]
12 and 13 are diagrams for explaining a method for creating a feature vector from a reference tree.
[0091]
FIG. 12 is a diagram for explaining a method for calculating an information feature amount by analyzing a reference tree. In this embodiment, the information feature quantity vector is calculated based on the reference tree 1201. The reference tree 1201 adds a cell L1103 corresponding to the information L as shown in FIG. 11 to the reference tree shown in FIG. Is.
[0092]
The information feature amount vector obtained by analyzing the reference tree of the present invention is calculated by analyzing the reference tree based on the reference information or the keyword or content included in the information, and calculating a parameter indicating similarity. In this embodiment, the appearance frequency α, the average layer value β, the reappearance frequency γ, the weight 1δ, and the weight 2 (1 / ε) are used as parameters.
[0093]
First, each parameter included in the reference tree 1201 is subjected to analysis such as appearance frequency to calculate the above parameters. The re-appearance frequency γ of each information is given, for example, to a hierarchy in which information appears, with a score of 5 for the hierarchy value 0 and 4 for the hierarchy value 1 so that the score decreases by 1 for each increase of the hierarchy value by 1. Then, it can be calculated as follows.
[0094]
Number of points in the hierarchy where the information first appears ÷ Hierarchy value of the hierarchy where the information appears again
Considering information C as an example, information B refers to information C that appears in a hierarchy of layer value 2, and information A also refers to information C in a layer of layer value 1. Therefore, the hierarchy in which the information C first appears is the hierarchy value 1 and the score is 4. Next, the re-appearance layers are the layer value 2 and the layer value 3, and the re-appearance frequency γ is (4/2 + 4/3 + 4/3) because it appears twice in the layer of the layer value 3.
[0095]
In the example of the present embodiment shown in FIG. 12, the weight 1 (δ) is a numerical value obtained when the keywords included in the information match, and the weight 2 (1 / ε) is a figure for the information A. A numerical value similar to the dissimilarity indicated by 11 is used.
[0096]
Next, based on these analysis values, each information is represented as a vector to obtain a similarity vector 1203. Here, the vector element can be normalized or parameterized as necessary.
[0097]
Note that the items to be analyzed are not limited to this, and can be additionally modified according to the characteristics of the information.
[0098]
FIG. 13 is a diagram for explaining a method of calculating information feature amounts by comparing reference trees. This calculation method is a calculation method using the principle that the same information is referred to when the information is similar.
[0099]
The information feature quantity vector based on the comparison of the reference trees of the present invention is obtained by applying the analysis value of each information obtained by analyzing the reference tree to the classification target information to a predetermined calculation formula, and calculating the calculated value as the information feature quantity vector. Is generated by In the example of the present embodiment shown in FIG. 13, a calculation formula 1301 is used. As an analysis value by the reference tree analysis, a sum of numerical values obtained by normalizing the analysis values shown in FIG. 12 is used. A table 1302 shown in FIG. 13 is a vector representation of values calculated for each piece of information.
[0100]
FIG. 14 is a diagram for explaining the information classification process of the client device 120. The information classification process is executed in response to a user request. It can also be executed periodically when information is added.
[0101]
The information classification process has two main functions: one is a function that classifies information by calculating the similarity from the feature quantity of the information to be classified, and the other is an existing function that uses the similarity of information. It is a function to classify according to the classification result. The former will be described below. The latter is an operation when information is added, and will be described in the second embodiment.
[0102]
In the information classification process, first, the information feature quantity 324 is read from the database 125 of the client device 120 (S1401). Next, the user adjusts the feature amount and sets a classification standard, which is a classification of information that is the center of the classification (S1402). Here, the adjustment of the feature amount is to correct the feature amount created by the server device 110 according to the user's preference or the content of the classification. The information that is the center of the classification is determined by which information the user wants to perform the classification, and is applied to the information classification process S1403. Next, information is classified using this feature quantity, and the classification result is operated so as to be visualized (S1404). The final classification result is stored in the database 125 (S1405), and the process ends.
[0103]
At this time, as a method for classifying information, a method such as a k-mean method, an ISODATA method, or a self-organizing map can be applied. The purpose of the information classification process of the system of the present embodiment is not only to simply classify information but also to classify specific information so that the user can obtain similar information, and also to visualize the classification result. An important classification function.
[0104]
For example, in a self-organizing map (T. Kohonen: “Self-Organizing Map”, Springer, 1997), information is treated as a multidimensional vector and classified while being arranged in a two-dimensional plane or a three-dimensional space. Since the classification result is arranged in a two-dimensional plane or a three-dimensional space, the classification result is visualized as it is. Therefore, the self-organizing map is applied to the above-described feature quantity vector, and the result can be used as the information classification result.
[0105]
FIG. 15 is a diagram for explaining a method of classifying information using a self-organizing map. Here, a method for classifying into a two-dimensional plane will be described. The method of classifying into a three-dimensional space can be realized in the same way. Now, consider placing information on an o × p cell 1501. The information feature vector of the central information is used as input data and is an n-dimensional vector. At that time, each cell j has the following n-dimensional reference vector.
[0106]
[Expression 1]

[0107]
Classification is performed as follows. All input data x _i On the other hand, the subsequent processing is performed by the number of repetitions t. Input data x _i The cell c having the reference vector closest to is searched. The search is performed according to equation (2).
[0108]
[Expression 2]

[0109]
That is, the i-th input vector x _i And the reference vector m of the jth cell in the output layer _j Find the cell j in the output layer that minimizes the distance, and let c be that. Here, the distance between the input vector and the reference vector is expressed by, for example, a Euclidean distance as shown in Equation (3). A neighborhood set of cells c is formed. This neighborhood set N _c (T) is defined by, for example, Expression (4).
[0110]
[Equation 3]

[0111]
In equation (4), dis (j, c) represents the distance between cell j and cell c, and a collection of cells j whose distance from cell c is closer than r (t) is a neighborhood set. For example, the neighborhood set 1502 is formed around the cell c. Next, the reference vectors of the cells belonging to the neighborhood set are updated according to equations (5) and (6).
[0112]
[Expression 4]

[0113]
That is, the neighborhood set N _c The value of the reference vector of the cell belonging to (t) is updated according to equation (5), and the reference vectors of the other cells do nothing. Here, the distance r (t) and the coefficient α (t) are reduced uniformly with respect to the number of repetitions t. Finally, the cell having the closest reference vector is searched for the input data as shown in Expression (2), and the data is arranged. By the above process, each original value of the reference vector changes, but the reference vector and the input vector are still associated after the update, so it is possible to know which information is mapped to which location. Can do.
[0114]
Therefore, the information is classified using the feature vector of the information as input data, and the result of placing the information in the cell having the reference vector closest to the feature vector after the processing is visualized to obtain the classification result. Can do. Furthermore, in order to classify specific information as a center, the information is subjected to classification processing by forming a neighborhood set with the center cell of o × p of the self-organizing map as cell c. By processing in this way, classification is performed so that specific information is always at the center.
[0115]
The k-means method is a method of classifying information into k clusters (groups), and the number k to be classified first is set and classification is performed in the following steps.
[0116]
Step 1: k initial cluster centers z ₁ (1), z ₂ (1), ..., z _k Determine (1) appropriately.
Step 2: In the n-th iteration step, all input data x is classified into k clusters by the following method. For all l = 1, 2,..., K (l ≠ m),
[0117]
[Equation 5]

[0118]
And Where s _m (N) is z _m This is a set of input data with (n) as the cluster center.
Step 3: S obtained in Step 2 _m New cluster center z in (n) _m (N + 1)
[0119]
[Formula 6]

[0120]
And Where N _m S _m It is the number of (n).
Step 4: z for all m = 1, 2,..., K _m (N + 1) = z _m If (n), the algorithm ends as converged. Otherwise, go back to step 2.
[0121]
In addition, the ISODATA method is a method of changing according to the nature of input data without giving the number k of clusters. This is a method of performing cluster classification while giving a plurality of parameters and changing the parameters. However, there is a disadvantage that the parameters are determined depending on the input data, and the setting is complicated.
[0122]
After classification by the k-means method or the ISODATA method, vector quantization is performed in 2D or 3D for each cluster, and visualization is performed by arranging in 2D or 3D.
[0123]
FIG. 16 is a diagram for explaining the group classification process of the client device 120. The group classification process is executed in response to a user request. It can also be executed periodically when information is added.
[0124]
In the group classification process, using the information feature quantity 324 belonging to the basic group 325, the feature quantity of the basic group is calculated around a specific owner, and the basic group is classified from the feature quantity. Here, the owner is the owner of the information, and if the target information is a document, it is the author or the author.
[0125]
In the group classification process, first, the information feature quantity 324 and the basic group 325 are read from the client database 125 (S1601). Next, the user adjusts the information feature amount, and sets the classification, which is a classification standard for determining the owner who is the center of the classification and the basic group which is the center (S1602). Here, the adjustment of the feature amount is to correct the feature amount created by the server device in accordance with the user's preference and the content of the classification. Next, the basic group feature quantity is calculated using the feature quantity of this information (S1603). This is performed for each group, and is determined by the owner and basic group that are the center of the classification set in S1602. Based on the calculated basic group feature amount, the basic group is classified (S1604).
[0126]
Next, for each basic group, the owners belonging to the basic group are classified (S1605), the final classification result is stored in the database 125 (S1606), and the process ends.
[0127]
The group classification and the group classification are performed by using the classification method such as the self-organizing map described above.
[0128]
FIG. 17 is a diagram for describing an example of the feature amount of the basic group. Each basic group has a group name 1701, a representative vector 1702, and a group weight 1703. The representative vector 1702 is expressed by an average of feature quantity vectors of all registered information belonging to the basic group. The initial weight of the group is determined by the system administrator based on the relevance of the basic group, and can be changed by the user during classification. The information belonging to the basic group is collected for each owner (1704 to 1706), the number of information belonging to the basic group for each owner, the total number of information of the owner, the affiliation rate obtained by dividing the number of information by the total number of information, And an average vector obtained by averaging the feature vector of information. Also, a similarity is set for each owner. For example, when the owner 1704 is a specific owner with the classification as the center, α / β in FIG. 12 is set as the similarity. Here, the shaded portion in FIG. 17 represents the feature amount of the basic group. For example, an information feature amount vector of the basic group is generated as 1707. That is, the distance between the representative vector of the basic group and the representative vector of the basic group serving as the classification center (for example, Euclidean distance), the weight of the basic group, the membership rate of the owner serving as the classification center, or the similarity of the owners belonging to the basic group A vector based on the sum of, etc. is used as a feature vector of the basic group. Here, the item of the vector is free and determined according to the characteristics of the system. However, it is necessary to use a similarity that is centered on a certain owner. The feature vector needs to be normalized for convenience.
[0129]
FIG. 18 is a diagram for describing information management processing when the client device 120 acquires registration information from the server device 110. When a user searches for desired information using a classification result or the like and actually acquires the information, information management processing is executed as follows.
[0130]
First, the system state such as the classification result is acquired from the client database 125 (S1801). Next, the user selects desired information (S1802). The information is acquired from the server apparatus 110 via the network (S1803), and the server apparatus 110 reads the registration information from the database 115 of the server apparatus in response to a request from the client apparatus 120 and transmits the registration information to the client apparatus 120 (S1805). ). The finally acquired information is presented to the user (S1804), and the process ends.
[0131]
FIG. 19 is a diagram illustrating an example when a user uses the system. By the method described above, the information and the basic group to which the information belongs are classified on the client device 120, and the user can acquire the information registered in the server device 110 according to the classification result.
[0132]
Specifically, the user uses the system as follows. First, as described with reference to FIG. 6, the client apparatus 120 is activated to start using the system. Here, when the client device 120 is already activated, the processing can be executed only by the user logging in. Next, the user requests information classification, and the registered information is classified by the information classification processing described with reference to FIG. 14 (S1901). Next, the user requests group classification, and the basic group to which the registration information belongs is classified by the group classification processing described with reference to FIG. 16 (S1902). Next, the user requests display of the classification result, and the information management unit of the client device 120 displays the classification result on the screen of the client device 120 (S1903). Further, the user requests acquisition of specific information, acquires the information selected by the user in the information management processing described with reference to FIG. 18 from the server device 110, and presents the information to the user (S1904). Next, the user requests the automatic transmission / reception setting for automatically setting newly added information, performs automatic transmission / reception setting by the method described in the fourth embodiment (S1905), and finally uses the user by the information management unit. The history is stored as a log in the database of the client device, and the system is terminated or the user logs off.
[0133]
Note that the procedure described with reference to FIG. 19 is merely an example, and the requests can be replaced or only a part of the requests can be used. If the information and group classification are set to be automatically executed on the client device 120, the user can acquire the latest classification result without waiting for the processing time.
[0134]
(Second Embodiment)
A classification process when information is added to the system will be described with reference to the drawings.
[0135]
FIG. 20 is a conceptual diagram illustrating the flow of information addition processing. When information is added to the server apparatus 110, the information is registered in the system in the information management process of the server apparatus 110, and the registration information 312, the basic group 315, and the update log 313 are stored in the database 115 of the server apparatus. Next, in the feature amount calculation process S2002 of the server device, all the feature amounts of each registered information 312 are calculated and reprocessed, or only the added information is calculated and added, and the feature amount 314 is stored in the database 115. Store. Then, the change is transmitted to the client device 120 in the information management processing S2001 of the server device. In the information management processing S2004 of the client device 120, the update log 323, the meta information 322, the information feature amount 324, and the basic group 325 are stored. The data is stored in the database 125, and the server device 110 enters an idle state.
[0136]
If the client apparatus 120 is not activated when information is added to the server apparatus, the addition is reflected as described with reference to FIG. 6 when the client apparatus 120 is activated next time.
[0137]
FIG. 21 is a diagram for explaining the addition processing of the feature amount calculation processing of the server apparatus 110 when information is added. In the feature amount calculation processing in FIG. 20, there are a case where feature amounts are recalculated for all registered information and a case where feature amounts are calculated only for added registration information. The former is performed as described in FIG. 9, and the latter is performed by the method shown in FIG. 21 as described below.
[0138]
First, the added registration information 312 is read from the database 115 of the server apparatus 110 (S2101). Next, a feature amount is calculated for the added registration information (S2102). At this time, reference information for calculating the feature value by the method described above is appropriately acquired from the database. The feature quantity of the last calculated information is stored in the database 115 (S2103), and the process ends.
[0139]
FIG. 22 is a diagram for explaining the addition process of the information classification process of the client device after the addition of information. The information classification after the information addition is performed by the method of FIG. 14 when all the feature amounts of the registration information 312 are recalculated, and when the feature amount is calculated only for the added registration information, it is shown in FIG. By the way.
[0140]
First, an update log 323 is acquired from the database 125 of the client device 120 (S2201), and the feature amount 324 of the added information and the classification result 326 at that time are acquired from the database 125 (S2202). Then, information added to the classification result is arranged (S2203), this is stored in the database 125 as a new classification result (S2204), and the process is terminated. Here, in the case of using the self-organizing map, the process S2203 searches for and arranges the reference vector closest to the last feature vector without performing the classification process. The classification setting uses the previous classification setting.
[0141]
FIG. 23 is a diagram illustrating the addition processing of the group classification unit after the addition of information. The group classification after the addition of information is performed by the method shown in FIG. 16 when all the feature amounts of the registration information are recalculated, and when the feature amount is calculated only for the added registration information, it is shown in FIG. By the way. However, in this case, when a new basic group is not created by the added registration information, the group classification addition process is not performed.
[0142]
First, an update log is acquired from the database 125 of the client device (S2301), and the feature amount of the added information, the newly added basic group, and the group classification result at that time are acquired from the database 125 (S2302). ). Next, the feature amount of the added basic group is calculated (S2303), the added basic group is arranged with respect to the classification result (S2304), and the added basic group is classified (S2305). Finally, the last classification result is stored in the database 125 (S2306), and the process ends.
[0143]
Note that the method described with reference to FIG. 16 is used as the feature amount and classification method of the basic group. Note that the settings set during the previous classification are used for setting the classification criteria and the like.
[0144]
As described above, even when information is newly added to the server device, the information and the basic group can be classified. In addition, even when already registered information is updated, it can be applied in the same manner.
[0145]
(Third embodiment)
Next, the classification of owners applying the above-described method will be described. If the classification within the basic group described with reference to FIGS. 16 and 17 is used, the owner of the information can be classified. That is, for each owner, the owner can be classified using the same method as the information classification using the average vector obtained by averaging the feature quantity vectors of the registered information possessed by the owner. If a self-organizing map is used for classification, as in the case of classification of information, the classification is performed by classifying the owner so that the average vector of the owner to be the center of the classification is located at the center of the map. Classification centered on the person.
[0146]
(Fourth embodiment)
Automatic transmission / reception of information between the server apparatus 110 and the client apparatus 120 will be described with reference to the drawings. The automatic transmission / reception is a function of automatically transmitting information desired by the user to the client device 120 when information is newly added to the server device 110.
[0147]
FIG. 24 is a schematic diagram illustrating automatic transmission / reception settings. First, information and group classification results are acquired from the client device database 125 (S2401). Next, a target of information to be automatically acquired is selected and stored in the database 125 as an automatic transmission / reception setting 328 (S2402). The automatic transmission / reception setting 328 is registered in the server device 110 (S2403), and the server device 110 stores the automatic transmission / reception setting 318 in the database 115 of the server device 110.
[0148]
Here, in selecting the target of information to be automatically acquired, the distance between the feature vector of specific information and the feature vector of information to be acquired, or a specific basic group or a specific basic group feature vector A distance or a specific owner, or a distance from a specific owner is designated to set automatic transmission / reception, and an object that satisfies the automatic transmission / reception setting is acquired.
[0149]
FIG. 25 is a diagram illustrating a first example of an automatic transmission / reception method. Here, in order to simplify the drawing, each database is omitted for explanation.
[0150]
Now, it is assumed that the client apparatus B newly registers information in the server apparatus, and the client apparatus A performs automatic transmission / reception settings. First, the client apparatus B registers information with the server apparatus (S2501). The server apparatus performs the information addition process S2502 described with reference to FIG. 20, compares the obtained feature amount vector with the automatic transmission / reception setting, and determines transmission to the client apparatus A (S2503). When the setting is satisfied, the server device reports new arrival information to the client device A (S2504), and the client device A determines whether to acquire the information (S2505). Here, even if the automatic transmission / reception setting 328 is performed, the client apparatus A can reject the acquisition of information through the processing S2505. When acquiring the information, an inquiry is made as to whether or not the information can be transmitted to the client apparatus B that is the sender of the information (S2506), and the client apparatus B answers it (S2507). At this time, whether or not transmission is possible is determined based on whether or not the client apparatus B directly transmits, or the state of the network from the client apparatus A to the server apparatus and from the client apparatus A to the client apparatus B. When the client apparatus A acquires information from the client apparatus B, processes S2508 and S2509 are executed. When the client apparatus A acquires information from the server apparatus, processes S2510 and S2511 are executed and the process ends.
[0151]
FIG. 26 is a schematic diagram for explaining a second example of the automatic transmission / reception method. Here, in order to simplify the drawing, each database is omitted for explanation. Similarly to FIG. 25, it is assumed that the client apparatus B newly registers information in the server apparatus, and the client apparatus A performs automatic transmission / reception settings.
[0152]
First, the client apparatus B registers information in the server apparatus (S2601). The server apparatus performs the information addition process S2602 described with reference to FIG. 20, and notifies the client apparatus A of new arrival information (S2603). The client apparatus A performs the information classification S2604 and the group classification S2605 described with reference to FIGS. 22 and 23, and determines whether to acquire information (S2606). This is done according to the automatic transmission / reception setting 318. Next, when acquiring information, an inquiry is made as to whether or not the information can be transmitted to the client apparatus B, which is the sender of the information (S2607), and the client apparatus B answers it (S2608). At this time, whether or not transmission is possible is determined based on whether or not the client apparatus B directly transmits, or the state of the network from the client apparatus A to the server apparatus or from the client apparatus A to the client apparatus B. When the client apparatus A acquires information from the client apparatus B, the processes are executed in steps S2609 and S2610. When the client apparatus A acquires information from the server apparatus, the processes are completed by steps S2611 and S2612.
[0153]
As described above, the user can acquire the latest registration information related to the user without particularly processing.
[0154]
【The invention's effect】
As described above, according to the present invention, server storage means for storing a classification target information group including a plurality of classification target information, a server that mutually calculates information feature amounts between each of the plurality of classification target information Server control means for storing in the storage means, a server apparatus including server communication means for receiving the information feature quantity transmission request and transmitting the information feature quantity read from the server storage means, and an information group set by the user Client storage means for storing in advance a classification criterion as a reference, a client communication means for transmitting an information feature amount transmission request to the server device and receiving an information feature amount from the server device, and a user request or a predetermined When the condition is satisfied, the classification standard is automatically read from the client storage means, and based on the information feature amount received by the communication means Since it includes a client device including client control means for classifying the classification target information using the read classification criteria and storing the classified result in the client storage means, the information and the position of the user can be understood. Classification can be realized, and information related to the classification result can be passively received by the user.
[Brief description of the drawings]
FIG. 1 is a diagram illustrating an overall configuration of a classification system according to an embodiment.
FIG. 2 is a diagram illustrating an arrangement configuration when the system of the present embodiment operates.
FIG. 3 is a diagram illustrating a database of a server device and a client device according to the present embodiment.
FIG. 4 is a diagram illustrating processing at the time of activation of a server device of the system according to the present embodiment.
FIG. 5 is a diagram illustrating an information management process when the server apparatus according to the present embodiment is activated.
FIG. 6 is a diagram illustrating processing at the time of activation of a client device of the system according to the present embodiment.
FIG. 7 is a diagram illustrating an information management unit when a client device is activated.
FIG. 8 is a diagram illustrating the definition of information to be classified in the system of the present embodiment.
FIG. 9 is a diagram illustrating a feature amount calculation process of the server device.
FIG. 10 is a diagram illustrating a method for creating a reference tree for calculating an information feature amount.
FIG. 11 is a diagram illustrating a reference tree for explaining a method of adding a new cell to the reference tree.
FIG. 12 is a diagram for explaining a method for calculating an information feature amount by analyzing a reference tree.
FIG. 13 is a diagram for explaining a method for calculating an information feature amount by comparing reference trees.
FIG. 14 is a diagram illustrating information classification processing of a client device.
FIG. 15 is a diagram for explaining a method of classifying information using a self-organizing map;
FIG. 16 is a diagram illustrating group classification processing of a client device.
FIG. 17 is a diagram for describing an example of a basic group feature amount;
FIG. 18 is a diagram illustrating an information management process when the client device of the present embodiment acquires registration information from the server device.
FIG. 19 is a diagram illustrating an example when a user uses a system.
FIG. 20 is a conceptual diagram illustrating a flow of information addition processing.
FIG. 21 is a diagram illustrating an addition process of a feature amount calculation process of a server device when information is added.
FIG. 22 is a diagram for explaining an adding process of an information classification process of a client device after adding information;
FIG. 23 is a diagram for describing an adding process of a group classification unit after adding information;
FIG. 24 is a schematic diagram illustrating automatic transmission / reception settings.
FIG. 25 is a diagram illustrating a first example of an automatic transmission / reception method.
FIG. 26 is a schematic diagram illustrating a second example of an automatic transmission / reception method.
[Explanation of symbols]
110 Server device
120 Client device
111, 121 Information Management Department
112 Feature amount calculator
113, 124 Automatic transmission / reception unit
122 Information classification part
123 Group classification part
115, 125 database
201, 202, 203 Client device
220 network
311, 321 User setting
312 Registration Information
313, 323 LOG
314, 324 Feature information
315, 325 Basic group
318, 328 Automatic transmission / reception setting
322 meta information
326, 327 information, group classification results
801 Information
802 Information name part
803 Reference information section
804 Keyword part
805 Contents
806 owner
807 Group Department
808 information group
1001-1010, 1101-1109 cells
1011-1014, 1110 hierarchy
1201 Reference tree
1202 Analysis result table
1203 similarity vector
1301 Similarity calculation formula
1302 Similarity vector table
1501 Self-organizing map
1502 Neighbor set
1701 Group name
1702 Representative vector
1703 weight
1704-1706 owner
1707 feature vector

Claims (45)

Server storage means for storing a classification target information group including a plurality of classification target information, server control means for mutually calculating information feature amounts between each of the plurality of classification target information, and storing the information feature quantities in the server storage means; and A server device including a server communication unit that receives an information feature amount transmission request and transmits the information feature amount read from the server storage unit;
Client storage means for storing in advance a classification criterion that is set by a user and serving as a criterion for classifying the information group, transmits the information feature amount transmission request to the server device, and receives the information feature amount from the server device The client communication means and the user's request or when a predetermined condition is satisfied, the classification standard is automatically read out from the client storage means and read out based on the information feature amount received in the communication unit. An information classification system comprising: a client device including client control means for classifying the classification target information using the classification criteria and storing the classified result in the client storage means. The classification criteria includes an indication of center information that is required to be classified as the center of the classification target information,
The information classification system according to claim 1, wherein the client control unit classifies information indicated in the instruction of the center information as a center. The classification target information includes information content and owner,
The client control means classifies the content of information included in each of the classification target information and the owner among the classification target information based on mutually calculated information feature amounts. The information classification system according to claim 1 or 2. Server storage means for storing a classification target information group including a plurality of classification target information, information feature amounts calculated between each of the plurality of classification target information, stored in the server storage means, and the classification target A server control unit that generates a plurality of information groups in which information to be classified included in the information group is classified according to a predetermined rule and stores the information group in the server storage unit, and receives an information feature amount transmission request and reads it from the server storage unit A server device including server communication means for transmitting the issued information feature quantity and the plurality of information groups;
Client storage means that stores in advance a classification criterion that is set by the user and serves as a criterion for classifying the information group, the information feature amount transmission request is transmitted to the server device, and the information feature amount and the plurality of information groups Client communication means for receiving the information from the server device, and when the user requests or when a predetermined condition is satisfied, the classification criteria are automatically read from the client storage means, and the information feature amount received by the communication unit and Client that classifies the plurality of information groups using the read classification criteria based on the group information feature amount calculated using the plurality of information groups, and stores the classified result in the client storage unit And a client device including a control unit. Broadcast classification system. The classification criteria includes an indication of center information that is required to be classified as the center of the classification target information,
5. The information classification system according to claim 4, wherein the client control unit classifies the information group including information indicated in the instruction of the center information as a center. The classification target information includes information content and owner,
The client control means classifies information contents and owners included in each of the information groups among the information groups based on mutually calculated group information feature amounts. Item 4. The information classification system according to item 4 or 5. The classification standard includes a feature amount adjustment value indicating an adjustment amount when adjusting the information feature amount for each user,
The information classification system according to claim 1, wherein the client control unit classifies the information feature amount based on a value adjusted by the feature amount adjustment value. The predetermined condition is that additional information that is new information is added to the target information group,
8. The information classification system according to claim 1, wherein the client control unit performs classification based on an information feature amount calculated between the additional information and each of the classification target information. . The predetermined condition is that additional information that is new information is added to the target information group,
The client communication means receives the additional information from the server device;
8. The information classification system according to claim 1, wherein the client control means classifies the received additional information by including it in the target information based on the already received information feature amount. . The server control unit may calculate the information feature amount between each of the classification target information included in the target information group stored at the calculation time point for each predetermined time interval between the client device, Alternatively, each time the additional information is added, the additional information is added to the target information group and is calculated between each of the classification target information included in the target information group to which the addition has been performed. The information classification system according to any one of claims 1 to 7. When the server device receives a notification that there is additional information from one of the client devices, the server device notifies a client device other than the client device that made the notification that the additional information is present,
The client device, when requesting the additional information, receives the additional information from either the server device or a client device that has transmitted a notification that there is the additional information. The information classification system according to any one of 10. Server storage means for storing a classification target information group including a plurality of classification target information;
Server control means for mutually calculating an information feature quantity between each of the plurality of classification target information and storing it in the server storage means;
A server apparatus comprising: a server communication unit that receives an information feature amount transmission request and transmits the information feature amount read from the server storage unit. Server storage means for storing a classification target information group including a plurality of classification target information;
A plurality of pieces of information obtained by mutually calculating information feature amounts between each of the plurality of classification target information and storing them in the server storage means, and classifying the classification target information included in the classification target information group according to a predetermined rule Server control means for generating a group and storing it in the server storage means;
A server apparatus comprising: a server communication unit that receives an information feature amount transmission request and transmits the information feature amount read from the server storage unit and the plurality of information groups. The server control unit mutually calculates the information feature amount between each of the classification target information included in the target information group stored at the calculation time point for each predetermined time interval, or the additional information 13. The additional information is added to the target information group each time a target is added, and mutual calculation is performed between each of the classification target information included in the target information group to which the addition is performed. Or the server apparatus of 13. Client communication means for transmitting an information feature amount transmission request and receiving information feature amounts calculated mutually between each of a plurality of classification target information included in the classification target information group;
Client storage means for preliminarily storing a classification standard that is set by a user and serves as a reference when classifying the plurality of classification target information;
When the user's request or a predetermined condition is satisfied, the classification standard is automatically read from the client storage unit, and the read classification standard is used based on the information feature amount received by the communication unit. A client apparatus comprising: a client control unit that classifies the classification target information and stores the classified result in the client storage unit. The classification criteria includes an indication of center information that is required to be classified as the center of the classification target information,
16. The client device according to claim 15, wherein the client control means classifies information indicated in the instruction of the center information as a center. The classification target information includes information content and owner,
The client control means classifies the content of information included in each of the classification target information and the owner among the classification target information based on mutually calculated information feature amounts. The client device according to claim 15 or 16. An information feature amount transmission request is transmitted, and information feature amounts calculated mutually among each of a plurality of classification target information included in the classification target information group and the classification target information included in the classification target information group A client communication means for receiving a plurality of information groups classified according to a rule;
Client storage means for preliminarily storing a classification standard that is set by a user and serves as a reference when classifying the plurality of classification target information;
A group calculated by using the information feature amount and the plurality of information groups received by the communication unit by automatically reading the classification criteria from the client storage unit upon a user request or when a predetermined condition is satisfied Client control means for classifying the plurality of information groups using the read classification criteria based on the information feature quantity and storing the classified results in the client storage means apparatus. The classification criteria includes an indication of center information that is required to be classified as the center of the classification target information,
19. The client apparatus according to claim 18, wherein the client control means classifies the information group including information indicated in the instruction of the center information as a center. The classification target information includes information content and owner,
The client control means classifies information contents and owners included in each of the information groups among the information groups based on mutually calculated group information feature amounts. Item 20. The client device according to Item 18 or 19. The classification standard includes a feature amount adjustment value indicating an adjustment amount when adjusting the information feature amount for each user,
21. The client device according to claim 15, wherein the client control unit classifies the information feature quantity based on a value adjusted by the feature quantity adjustment value. The predetermined condition is that additional information that is new information is added to the target information group,
The client device according to any one of claims 15 to 21, wherein the client control unit performs classification based on an information feature amount calculated between the additional information and each of the classification target information. The predetermined condition is that additional information that is new information is added to the target information group,
The client communication means receives the additional information;
The client device according to any one of claims 15 to 21, wherein the client control unit classifies the received additional information by including it in the target information based on the already received information feature amount. A program for executing a step of classifying the classification target information based on information feature amounts calculated mutually between the classification target information included in the classification target information group,
A feature amount storing step in which the server control unit stores the information feature amount between each of the plurality of classification target information and stores it in the server storage unit;
A program for receiving an information feature amount transmission request and executing a transmission step in which a server communication unit transmits the information feature amount read from the server storage unit. A program for executing a step of classifying the classification target information based on information feature amounts calculated mutually between the classification target information included in the classification target information group,
An information feature amount is calculated between each of the plurality of classification target information, the server control means stores it in the server storage means, and the classification target information included in the classification target information group is classified according to a predetermined rule A feature amount storing step of generating a plurality of information groups and storing them in the server storage means;
A program characterized by receiving an information feature amount transmission request and executing a transmission step in which server communication means reads and transmits the feature information amount and the plurality of information groups from the server storage means. The feature amount storing step calculates the information feature amount between each of the classification target information included in the target information group stored at the calculation time point at predetermined time intervals, or adds the information feature amount The additional information is added to the target information group each time information is added, and is calculated between each of the classification target information included in the target information group to which the addition is performed. The program according to 24 or 25. A program for executing a step of classifying the classification target information based on information feature amounts calculated mutually between the classification target information included in the classification target information group,
A feature amount receiving step in which the client communication means receives an information feature amount that is calculated between each of a plurality of pieces of classification target information included in the classification target information group by transmitting an information feature amount transmission request;
A classification criterion storage step in which the client storage means stores in advance a classification criterion that is set by the user and serves as a criterion when classifying the plurality of classification target information;
The client control means automatically reads out the classification criteria from the client storage means when requested by a user or when a predetermined condition is satisfied, and the read out based on the information feature amount received in the communication unit. A classification result storing step of classifying the classification target information using a classification criterion and storing the classified result in the client storage means. The classification criteria includes an indication of center information that is required to be classified as the center of the classification target information,
28. The program according to claim 27, wherein the classification result storing step classifies the information indicated in the instruction of the center information as a center. The classification target information includes information content and owner,
In the classification storing step, classification is performed based on information feature amounts calculated mutually with respect to contents and owners of information included in each of the classification target information among the classification target information. The program according to claim 27 or 28. A program for executing a step of classifying the classification target information based on information feature amounts calculated mutually between the classification target information included in the classification target information group,
An information feature amount transmission request is transmitted, and information feature amounts calculated mutually among each of a plurality of classification target information included in the classification target information group and the classification target information included in the classification target information group are set as predetermined A feature receiving step in which the client communication means receives a plurality of information groups classified according to the rules;
A classification criterion storage step in which the client storage means stores in advance a classification criterion that is set by the user and serves as a criterion when classifying the plurality of classification target information;
The client control means automatically reads out the classification criteria from the client storage means according to a user request or when a predetermined condition is satisfied, and based on the information feature amount and the plurality of information groups received in the communication unit And a classification result storing step of classifying the classification target information using the read classification standard and storing the classified result in the client storage means. The classification criteria includes an indication of center information that is required to be classified as the center of the classification target information,
31. The program according to claim 30, wherein the classification result storing step classifies the information group including information indicated in the instruction of the center information as a center. The classification target information includes information content and owner,
The classification result storing step classifies the content of information included in each of the information groups and the owner based on the mutually calculated group information feature amounts among the information groups. The program according to claim 30 or 31. The classification standard includes a feature amount adjustment value indicating an adjustment amount when adjusting the information feature amount for each user,
33. The program according to claim 27, wherein the classification result storing step classifies the information feature quantity based on a value adjusted by the feature quantity adjustment value. The predetermined condition is that additional information that is new information is added to the target information group,
The program according to any one of claims 27 to 33, wherein the classification result storing step performs classification based on an information feature amount calculated between the additional information and each of the classification target information. The predetermined condition is that additional information that is new information is added to the target information group,
The client communication means further executes a step of receiving the additional information;
34. The program according to claim 27, wherein the classification result storing step classifies the received additional information by including it in the target information based on the already received information feature amount. 36. A computer-readable recording medium on which each program according to claim 24 is recorded. In an information classification method for classifying the classification target information based on information feature amounts calculated mutually between the classification target information included in the classification target information group,
A setting step for preliminarily setting and storing a classification standard as a standard when classifying the classification target information;
A classification step of automatically reading out the classification standard and classifying the classification target information using the read classification standard, upon request of a user or when new information is added to the classification target information group An information classification method characterized by that. The classification criteria includes an indication of center information that is required to be classified as the center of the classification target information,
38. The information classification method according to claim 37, wherein the classification step classifies the information indicated in the instruction of the central information as a center. In an information classification method for classifying the classification target information based on information feature amounts calculated mutually between the classification target information included in the classification target information group,
A generation step of classifying classification target information included in the classification target information group according to a predetermined rule to generate a plurality of information groups; a setting step of presetting and storing a classification standard as a reference when classifying the information group;
When the user's request or when a predetermined condition is satisfied, the classification criteria are read out, and the plurality of information groups are determined based on information feature amounts calculated mutually between the plurality of information groups. An information classification method comprising: a classification step of classifying using the read classification standard. The classification criteria includes an indication of center information that is required to be classified as the center of the classification target information,
40. The information classification method according to claim 39, wherein the classification step classifies the information group including information indicated in the instruction of the center information as a center. The classification target information includes information content and owner,
The classification step is characterized in that classification is performed on the basis of information features calculated for each content and owner included in each of the classification target information among the classification target information. Item 41. The information classification method according to any one of Items 37 to 40. The classification standard includes a feature amount adjustment value indicating an adjustment amount when adjusting the information feature amount for each user,
42. The information classification method according to claim 37, wherein the classification step classifies the information feature quantity based on a value adjusted by the feature quantity adjustment value. The predetermined condition is that additional information that is new information is added to the target information group,
43. The information classification method according to claim 37, wherein the classification step performs classification based on an information feature amount calculated between the additional information and each of the classification target information. The information feature amount is calculated between each of the classification target information included in the target information group stored at the calculation time point for each predetermined time interval, or whenever the additional information is added 43. The method according to claim 37, wherein the additional information is added to the target information group, and the classification target information included in the target information group to which the addition has been performed is mutually calculated. Information classification method described in 1. When the server device receives a notification that there is additional information from one of the client devices, an additional information notification step for notifying that there is the additional information to a client device other than the client device that has made the notification When,
When the client device requests the additional information, the client device further includes an additional information receiving step of receiving the additional information from either the server device or the client device that has transmitted a notification that there is the additional information. 45. The information classification method according to any one of claims 37 to 44, wherein:

Images