JP4005925B2 - Document processing method, document processing apparatus, and program - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a document processing method and apparatus for replacing a proper noun portion related to privacy in a document with a prone character to make it unidentifiable.
[0002]
[Prior art]
If you try to share or divert existing electronic documents such as e-mails, internal reports, daily reports, weekly reports, etc., the privacy information of proper nouns may be infringed depending on the document. For this reason, it has been necessary to deal with it by deleting or hiding the proper nouns related to the privacy information in the document manually.
[0003]
On the other hand, there is a technique in which a proper noun part related to privacy is extracted from a document, and the extracted proper noun part is processed as a letter (for example, refer to Patent Document 1 and Patent Document 2). In this method, a word to be made unidentifiable is registered in a word dictionary, and a proper noun portion related to privacy is extracted by performing morphological analysis using the dictionary.
[0004]
However, this method does not provide a correction means when the extracted proper noun part is incorrect or a word dictionary update means.
[0005]
[Patent Document 1]
JP 2002-259363 A
[0006]
[Patent Document 2]
JP 2002-259368 A
[0007]
[Problems to be solved by the invention]
Thus, conventionally, when detecting a proper noun part from a document and processing it in a prose, it is impossible to correct when the detected proper noun part is incorrect, and to update the word dictionary used for detecting the proper noun part. There was a point.
[0008]
In view of the above problems, an object of the present invention is to provide a document processing method and apparatus capable of easily confirming and correcting a proper noun portion detected from a document.
[0009]
Another object of the present invention is to provide a document processing method and apparatus capable of easily updating a word dictionary used for detecting proper noun parts in a document.
[0010]
[Means for Solving the Problems]
The present invention detects a masking target location from an input document based on a dictionary storing a character string to be masked or a part thereof, stores the detected masking target location in a storage means, and stores the masking target location. The masking target portion displayed on the display screen and stored in the storage means is rewritten to the masking target portion corrected by the user on the display screen, and the masking target portion stored in the storage means in the input document is masked. As a result, the proper noun part detected from the document can be easily confirmed and corrected.
[0011]
Further, in the document displayed on the display screen, the new masking target location designated by the user is stored in the storage means, and the character string of the new masking target location stored in the storage means is stored in the dictionary. By storing the word dictionary, the word dictionary used for detecting the proper noun part in the document can be easily updated.
[0012]
Further, the dictionary is based on at least one of the number of characters of each character string, the character type constituting each character string, and the appearance frequency of each character string in an existing document among a plurality of input character strings. The character string to be stored in the selected character string, based on the character string that cannot be morphologically analyzed, the clauses obtained as a result of the morphological analysis of each character string in the selected character string, and the dictionary By storing character strings that cannot be detected as masking target parts in the above dictionary, a word dictionary used for detecting proper nouns in documents using existing databases in the company or commercially available databases Can be easily constructed and updated.
[0013]
DETAILED DESCRIPTION OF THE INVENTION
Embodiments of the present invention will be described below with reference to the drawings.
[0014]
FIG. 1 is a block diagram showing the configuration of a document masking apparatus to which the document processing apparatus according to this embodiment is applied. The document masking apparatus according to the present embodiment can be realized by a computer that reads a program recorded on a recording medium such as a magnetic disk and whose operation is controlled by this program.
[0015]
The document masking apparatus includes an input unit 101, a control unit 102, an output unit 103, a temporary storage unit 104, a dictionary registration unit 105, a data selection unit 106, a frequency calculation unit 107, and an existing document storage unit 108. The masking target specifying unit 109, the morpheme analyzing unit 110, the masking rule storage unit 111, the word dictionary 112, the masking correcting unit 113, the masking determining unit 114, and the prone character processing unit 115 are configured.
[0016]
An input unit 101 as an input unit takes in (inputs) a document or data to be processed from, for example, a memory, a magnetic disk, or an optical disk. Also, user instructions and key inputs are captured from, for example, a keyboard or a microphone.
[0017]
After receiving and analyzing the information input from the input unit 101, the control unit 102 sends information necessary for the processing to each component necessary for processing the input information. The processing results in each component are returned to the control unit 102 and output via the output unit 103. In the output unit 103, information to be output (output information) is displayed on, for example, a display, or is output by sound from a speaker. Details of the processing operation of the control unit 102 will be described later.
[0018]
The temporary storage unit 104 is a storage area for temporarily storing processing results and the like, and includes, for example, a RAM or a magnetic disk. The temporary storage unit 104 stores a masking result list 104a and a registration candidate list 104b. Information stored in each list will be described later.
[0019]
When the dictionary registration unit 105 receives registration candidate data input from the input unit 101 via the control unit 102, the dictionary registration unit 105 activates the data selection unit 106 and passes the registration candidate data to the data selection unit 106.
[0020]
The data selection unit 106 activates the frequency calculation unit 107, the morpheme analysis unit 110, and the masking target specifying unit 109, and selects valid data (words) as registration candidates in the word dictionary 112 from the received registration candidate data. Sort out. Details of the processing operation of the data selection unit 106 will be described later.
[0021]
The frequency calculation unit 107 refers to the document (existing document) stored in the existing document storage unit 108 and calculates the appearance frequency of the registration candidate data in the existing document.
[0022]
The masking target specifying unit 109 receives the document via the control unit 102, activates the morphological analysis unit 110, and the masking rules stored in the masking rule storage unit 111 and the character string to be masked (replaced by the abbreviation) Alternatively, by referring to the word dictionary 112 that stores a part of it, a masking target portion (a character string to be replaced with a letter) in the document is detected (specified). Details of the processing operation of the masking target specifying unit 109 will be described later.
[0023]
The morpheme analyzer 110 refers to the word dictionary 112 and performs morpheme analysis. Since the processing operation of the morphological analysis unit 110 is widely known, a description thereof will be omitted.
[0024]
The processing result of the masking target specifying unit 109 is stored as a masking result list 104 a via the control unit 102 and displayed on a display or the like via the output unit 103.
[0025]
The masking correction unit 113 receives a user's correction instruction for the masking target portion via the control unit 102, and stores the information in the masking result list 104a.
[0026]
The masking confirmation unit 114 receives a user confirmation instruction for the masking target portion via the control unit 102, and stores the information in the masking result list 104a.
[0027]
The script processing unit 115 receives a user's script processing instruction via the control unit 102, replaces the determined masking target portion with preset characters, symbols, fills, blanks, etc., and outputs the result. For example, it is displayed on a display via the unit 103.
[0028]
Next, details of each of the above-described units will be described using a flowchart.
[0029]
(A) Processing operation of the control unit 102
FIG. 2 is a flowchart showing the processing operation of the control unit 102.
[0030]
First, in step S201, it is determined whether or not the user's instruction specifies a masking target portion. If the masking target part is specified, the process proceeds to step S202. Otherwise, the process proceeds to step S211.
[0031]
In step S202, a document to be processed is acquired via the input unit 101, and the process proceeds to step S203.
[0032]
In step S203, the masking target specifying unit 109 is activated, and the process proceeds to step S204.
[0033]
In step S204, it is determined whether or not the user's instruction is correction of a masking target portion. When it is correction of the masking target part, the process proceeds to step S205. Otherwise, the process proceeds to step S206.
[0034]
In step S205, the masking correction unit 113 is activated, and the process proceeds to step S206. In step S206, it is determined whether or not the user's instruction is to confirm the masking target portion. If it is determined that the masking target portion is determined, the process proceeds to step S207. If not, the process ends.
[0035]
In step S207, the masking determination unit 114 is activated, and the process proceeds to step S208. In step S208, it is determined whether or not the user's instruction is a masking process for a portion to be masked. If it is the masking process for the masking target part, the process proceeds to step S209. Otherwise, the process proceeds to step S210.
[0036]
In step S209, the letter processing unit 115 is activated, and the process proceeds to step S210. In step S210, it is determined whether or not the user's instruction is dictionary registration of a masking target portion. If it is the dictionary registration of the masking target part, the process proceeds to step S213. If not, the process ends.
[0037]
On the other hand, if the process proceeds from step S201 to step S211, it is determined in step S211 whether or not the user instruction is batch dictionary registration. If it is batch dictionary registration, the process proceeds to step S212. Otherwise, the process is terminated.
[0038]
In step S212, registration candidate data is taken in via the input unit 101, and the process proceeds to step S213. Here, the registration candidate data input through the input unit 101 is, for example, a name, an address, and the like included in data such as an electronic telephone book. In the case of batch registration, words such as names and place names included in names, addresses, etc. recorded as a telephone book can be automatically registered in the word dictionary 112 in a batch.
[0039]
In step S213, the dictionary registration unit 105 is activated. The dictionary registration unit 105 activates the data selection unit 106. Since the data selection unit 106 selects valid data (words) as registration candidates for the word dictionary 112 from the input registration candidate data, the dictionary registration unit 105 registers the selected words in the word dictionary 112. Then, the process ends.
[0040]
FIG. 5 shows an example of the data structure of the word dictionary 112. The word dictionary stores information such as notation, reading, part of speech, and attribute of each word. In particular, the attribute of each word stored in the word dictionary 112 is used when (in step S203) the masking target specifying unit 109 determines whether or not a word in the input document is a masking target. .
[0041]
In this manner, the control unit 102 sends information captured by the input unit 101 to a necessary processing unit based on a user instruction, and controls activation of each processing unit.
[0042]
(B) Processing operation of masking target specifying unit
The processing operation of the masking target specifying unit 109 will be described with reference to the flowchart shown in FIG. First, in step S301, the masking target specifying unit 109 reads a document to be processed input from the input unit 101, and proceeds to step S302.
[0043]
In step S302, the number of documents is set in the variable N, and “1” as an initial value is set in the variable i for counting the number of documents, and the process proceeds to step S303. In step S303, it is determined whether i is N or less. If i is N or less, the process proceeds to step S304. If i is greater than N, the process ends.
[0044]
In step S304, the morpheme analysis unit 110 is activated to perform morphological analysis of the i-th read document, that is, the document [i], and the process proceeds to step S305.
[0045]
Here, the morphological analysis will be briefly described with reference to FIG. For example, it is assumed that a sentence as shown in FIG. 7A is a morphological analysis processing target. By dividing this sentence into phrases and words and adding parts of speech for each word, a morphological analysis result as shown in FIG. 7B is obtained. Note that when it cannot be divided into clauses and words, it means that morphological analysis cannot be performed or morphological analysis has failed, and when it can be divided into phrases and words, it means that morphological analysis has succeeded. In FIG. 7B, the symbol “/” is a phrase delimiter, the symbol “+” is a word delimiter in the phrase, and the character string enclosed by the symbol <> indicates the part of speech.
[0046]
In addition, what consists of at least 1 character, such as a clause and a word, is also called a character string simply.
[0047]
Returning to the description of FIG. 3, in step S305, the number of clauses is set in the variable S, and “1” is set as an initial value in the variable k for counting the number of clauses, and the process proceeds to step S306. In step S306, it is determined whether k is S or less. If k is equal to or less than S, the process proceeds to step S307. If k is larger than S, the process proceeds to step S310. In the following description, the kth clause is referred to as clause [k].
[0048]
In step S307, it is determined whether or not the clause [k] and the next clause [k + 1] satisfy the conditions of the masking rules stored in the masking rule storage unit 111. At that time, “attribute” information and the like registered in the word dictionary 112 are referred to. Each word included in each of the two clauses is searched from the word dictionary 112, and when such a word exists in the word dictionary 112, its attribute is read. When the correspondence relationship between the attributes of the words included in the two phrases is stored as a masking rule (when a rule that satisfies the condition exists), the process proceeds to step S308. Otherwise, the process proceeds to step S309.
[0049]
FIG. 6 shows an example of a masking rule stored in the marking rule storage unit 111. Each rule is described by a condition and a result. For example, in the first rule, if the attribute of the clause [k] is a company name, the clause [k] is specified as the company name. When the phrase [k + 1] field is blank, it indicates that the rule is a one-sentence rule. For example, in the fourth rule, if the part of speech of the clause [k] is an unknown word and the attribute of the clause [k + 1] is a personal name co-occurrence word, the clause [k] is specified as a personal name. Here, an example of a rule of one or two clauses is shown, but a rule of three or more clauses may be used. In the case of a rule having three or more clauses, matching of the rules with clauses (phrase [k], clause [k + 1], clause [k + 2],...) Corresponding to the number is performed.
[0050]
Returning to the description of FIG. 3, in step S308, a clause that satisfies the condition (at least one clause that matches the rule) is identified as a masking target portion, stored as a masking result list 104a, and the process proceeds to step S309.
[0051]
FIG. 8 shows an example of the masking result list 104a. In the masking result list 104a, the notation of the clause specified as the masking target portion, that is, the masking target notation, the front three characters of the clause, the subsequent three characters, the start position of the masking target notation, the number of characters, the type, and the confirmation Nine pieces of information including presence / absence, presence / absence processing, and registration / non-registration are stored. At the stage where step S308 has been completed, as shown in FIG.
[0052]
Returning to the description of FIG. 3, in step S309, k is incremented by one, and the process returns to step S306, and the processes in steps S307 to S308 are repeated for all the clauses in the document [i]. In step S310, i is incremented by 1, and the process returns to step S303 to repeat the processes in steps S304 to S309 for all input documents.
[0053]
In this way, the masking target specifying unit 109 uses the morpheme analysis and masking rules in the morpheme analysis unit 110 to specify the masking target part in each document. The identified masking target portion is stored as a masking result list 104a as shown in FIG. 8 and displayed on the display or the like via the output unit 103, for example.
[0054]
(C) Processing operation of the data selection unit
The processing operation of the data selection unit 106 in the dictionary registration process in step S213 in FIG. 2 will be described with reference to the flowchart shown in FIG.
[0055]
First, in step 401, the data selection unit 106 reads data to be processed input to the dictionary registration unit 105, stores it as a registration candidate list 104b, and proceeds to step S402. The data to be processed is registration candidate data read from the input unit 101 in the case of batch dictionary registration, and in the case of dictionary registration of a masking target portion (that is, a word based on the correction result of the masking target described later) In the case of updating the dictionary, among the data stored in the masking result list 104a, data for which a registration instruction has been given (for example, as shown in FIG. ).
[0056]
Here, when the user instruction is batch dictionary registration (step S211 in FIG. 2), the processing operation in step S213 in FIG. 2, that is, based on the registration candidate data read from the input unit 101 in step S212 in FIG. A case where the word dictionary 112 is updated will be described as an example.
[0057]
FIG. 10 is an example of information stored as the registration candidate list 104b. In the registration candidate list 104b, for each registration candidate data read from the input unit 101, the notation, type, appearance frequency, whether morphological analysis is successful, whether masking target identification is successful, and six selection results are displayed. Is described. At the stage where step S401 is completed, fields other than the notation and type of each registration candidate data are not filled in as shown in FIG.
[0058]
Returning to the description of FIG. 4, in step S402, the number of data is set in the variable N, and the initial value “1” is set in the variable i for counting the number of data, and the process proceeds to step S403. In step S403, it is determined whether i is N or less. If i is N or less, the process proceeds to step S404. If i is greater than N, the process ends. Here, the i-th registration candidate data is referred to as data [i].
[0059]
In step S404, it is determined whether the character string length (number of characters) in the notation of data [i] is equal to or greater than a predetermined value α. Here, α is a threshold value set in advance. For example, α is set to “2”. If the character string length is greater than or equal to α, the process proceeds to step S405. If the character string length is less than α, the data [i] proceeds to step S413 so as to be excluded from the registration target in the word dictionary 112.
[0060]
In step S405, it is determined whether or not the character string structure of the data [i] is only hiragana. In the case of hiragana only, the data [i] proceeds to step S413 to be excluded from the registration target in the word dictionary 112. If a character type other than hiragana is included, the process proceeds to step S406.
[0061]
In step S406, first, the frequency calculation unit 107 is activated. The frequency calculation unit 107 calculates the appearance frequency of the notation of the data [i] in the existing document stored in the existing document storage unit 108, and stores the information in the “appearance frequency” column of the registration candidate list 104b. Then, the process proceeds to step S407. In step S407, it is determined whether the appearance frequency is greater than or equal to a predetermined value β. Here, β is a threshold value set in advance. For example, β is set to “3”. When the appearance frequency is β or more, the process proceeds to step S408. If the appearance frequency is lower than β, the data [i] proceeds to step S413 to be excluded from the registration target in the word dictionary 112.
[0062]
In step S408, the morpheme analysis unit 110 is activated to analyze the notation of data [i] whose appearance frequency is β or more, and the process proceeds to step S409. In step S409, it is determined whether the morphological analysis is successful. If successful, the process proceeds to step S410. If unsuccessful, the process advances to step S412 to register the data [i] in the word dictionary 112. As shown in FIG. 11, the result of the morpheme analysis is stored in the “morpheme analysis” column of the registration candidate list 104b (in FIG. 11, when the morpheme analysis is successful, “◯” is marked, and when it is unsuccessful, "X").
[0063]
In step S410, the masking target specifying unit 109 is activated, and as shown in FIG. 3, the masking target part is specified from the notation of the data [i] based on the masking rule, and the process proceeds to step S411.
[0064]
In step S411, it is determined whether or not the notation of the data [i] can be specified as the masking target portion. If it can be specified, the data [i] can be specified as a masking target part from the current word dictionary 112 and does not need to be newly registered in the word dictionary 112 this time. In order to remove it, the process proceeds to step S413. If it cannot be specified, the process proceeds to step S412 to register the data [i] in the word dictionary 112. The determination result here is stored in the “masking target specification” field of the registration candidate list 104b as shown in FIG. 11 (in FIG. 11, when it can be specified, “◯” mark, when it cannot be specified) "X").
[0065]
If the process proceeds from step S409 or step S411 to step S412, in step S412, it is determined that the data [i] is registered in the word dictionary 112, and the result is stored in the registration candidate list 104b (in FIG. "" In the "" column), the process proceeds to step S414.
[0066]
On the other hand, if the process proceeds from step S404, step S405, step S407, or step S411 to step S413, it is determined in step S413 that the data [i] is not registered in the word dictionary 112, and the result is stored in the registration candidate list 104b. (In FIG. 11, “x” is indicated in the “selection result” column), and the process proceeds to step S414.
[0067]
In step S414, i is incremented by one, and the process returns to step S403, and the processing of step S404 to step S413 is repeated for registration candidate data of all data.
[0068]
In this way, the data selection unit 106 uses the frequency calculation unit 107, the morpheme analysis unit 110, and the masking target specifying unit 109 to generate an appearance frequency in an existing document that includes characters other than hiragana and a character string length of α. Among the registration candidate data having a value equal to or larger than β, the one that satisfies any one of the following cannot be specified as a masking target in the current word dictionary 112, which cannot be morphologically analyzed, is registered in the word dictionary 112. Select as a valid string. The result is stored in the “selection result” column of the registration candidate list 104b as shown in FIG.
[0069]
At the stage where the process of step S412 of the data selection unit 106 is completed, the process is as shown in FIG. That is, “Taro Yamada” and “Hayashi Politics” are selected as valid registration candidates in the word dictionary 112. This selected registration candidate data is additionally registered in the word dictionary 112 as shown in FIG. 12 together with attributes and parts of speech obtained from the types. At that time, a word to be newly registered is displayed from the output unit 103 via the dictionary registration unit 105 and the control unit 102, for example, as shown in FIG. 13, and is read to the user, part of speech, attribute, registration. It may be registered in the word dictionary 112 after inquiring about the presence or absence. In addition, only those instructed by the user on this screen may be registered in the word dictionary 112. At that time, “reading”, “part of speech”, and “attribute” input on the screen shown in FIG. 13 are registered in the word dictionary 112.
[0070]
(D) Updating the word dictionary based on the correction result of the masking target
Now, the user is instructed to specify the masking target part to the document masking apparatus shown in FIG. 1, and the process is input from the input unit 101 by the masking target specifying process (see FIG. 3) in step S203 of FIG. It is assumed that a masking result list 104a as shown in FIG. 8 is obtained from the target document.
[0071]
FIG. 14 is a display example of the masking result list 104a and shows an example of a document display screen.
[0072]
Based on the masking result list 104 shown in FIG. 8 obtained from the processing target document, the output unit 103 displays the processing target document input from the input unit 101 as shown in FIG.
[0073]
In FIG. 14, words such as “Company A”, “Communication Research Laboratories”, “Taro Yamada”, and the like, which are required as masking targets, are highlighted or highlighted so that they can be distinguished from other parts in the processing target document. Special indication is given.
[0074]
For example, when the screen as shown in FIG. 14 is displayed by the masking target specifying process in step S203 of FIG. 2, in step S204 of FIG. 2, the user instructs the correction of the masking target portion (for example, the screen). In step S205, the masking correction unit 113 is activated. At this time, for example, the user selects “communication laboratory” using a mouse or the like, and the selection instruction is sent to the masking correction unit 113. The user corrects the masking correction instruction to add at least one character immediately before and immediately after the masking target portion in the processing target document so that the selected character string to be masked becomes “Information and Communication Laboratory”. When input from the unit 113, the masking result list 104a is displayed in the masking result list 104a by the control unit 102 via the masking correction unit 113, as shown in FIG. To “Information and Communication Laboratories”. Based on the masking result list shown in FIG. 9, the screen display is also updated as shown in FIG.
[0075]
Note that the correction of the masking target portion by the user includes correction for deleting at least one character from the character string of the masking target portion in addition to the above correction. This correction is, for example, a case where the first two characters “communication” are deleted from the masking target location “communication laboratory” detected in the document to be processed and corrected to “laboratory”. Also in this case, in the same manner as described above, in the masking result list 104a, the masking target portion modified by the user in the “masking target notation” column is rewritten from “communication laboratory” to “laboratory”. The screen display is also updated based on the rewritten masking result list.
[0076]
In addition to the above two examples, the user may additionally specify a new masking target location as the masking target location correction. For example, on the screen shown in FIG. 14, although not shown in FIG. 14, a person name “New Year's Day” exists in the document to be processed. It is assumed that no detection (specification) is performed by the unit 109. In this case, when the user designates this character string, the masking target portion “New Year's Day” added by the user is added to the “masking target notation” field of the masking result list 104a. Based on the resulting masking result list, the screen display is also updated in the same manner as described above.
[0077]
Next, when the user gives an instruction to confirm the masking target location (for example, by selecting a predetermined button provided on the screen) (step S206 in FIG. 2), the masking confirmation unit 114 is activated (in FIG. 2). In step S207), as shown in FIG. 9, the masking result list 104a records that it has been confirmed in the “determined” column for each masking target (indicated by “◯” in FIG. 9), and is displayed on the screen. As shown in FIG. 16, the display also shows that the highlighted part has been changed to underlined display, and the masking target part including the corrected part has been confirmed.
[0078]
Next, when the user gives an instruction to process a masked portion (for example, a predetermined button provided on the screen is selected) (step S208 in FIG. 2), the script processing unit 115 is activated (FIG. 2). Step S209), in the masking result list 104a, as shown in FIG. 9, the fact that the masking process is instructed is recorded in the “masking process” column of each masking target (in FIG. 9, the masking result list 104a is represented by a mark “◯”). As shown in FIG. 17, each masking target portion is replaced with, for example, the symbol “x”.
[0079]
In step S204 to step S210 in FIG. 2, a masking location correction, a masking confirmation instruction, a prosecution processing instruction, and a registration instruction to the word dictionary 112 can be performed from the screen as shown in FIG. .
[0080]
FIG. 18 shows another display example of the masking result list 104a shown in FIG. 8 obtained in step S203 of FIG.
[0081]
The screen shown in FIG. 18 can be changed by selecting (pressing) one of the “to list list screen” buttons B1 provided on the document display screens shown in FIGS. it can.
[0082]
In the screen display example shown in FIG. 18, the notation “Company A”, “Communication Research Laboratories”, “Taro Yamada”, etc. of the masking target part detected from the document to be processed and stored in the masking result list 104a indicate the context. It is displayed as a list by type, along with the preceding and following character strings. The list list display as shown in FIG. 18 is displayed based on the contents of the masking result list 104a at the time when the display instruction is given. When the correction is made, the correction result is stored in the masking result list 104a, and the masking target portion after the correction is displayed as shown in FIG.
[0083]
In the list list display screen shown in FIG. 18, a “confirmation” instruction region for masking confirmation instruction, a “registration” instruction region to the word dictionary, and a “bend processing” instruction region are provided for each masking target. Is provided.
[0084]
On the screen shown in FIG. 18, the user selects “communication laboratory” detected as a masking target using a mouse or the like, and the selection instruction is sent to the control unit 102 via the input unit 101. The user changes the masking range so that the selected masking target word becomes “Information and Communication Laboratory” (corrected to “Information and Communication Laboratory”), and “Company A” and “Information and Communication Research” Enter “Check” (in this case, “×”) in the “Confirm” instruction area and “Future processing” instruction area of “Taro” and “Taro Yamada”, and enter a check in the “Register” instruction area of “Information and Communication Laboratories” Then, the screen becomes as shown in FIG.
[0085]
In the state of the screen display shown in FIG. 19, when the “confirm execution” button B3 is pressed at the bottom center of the screen, the checked notation is confirmed. That is, the masking correction unit 113 is activated (step S205 in FIG. 2), and the masking result list 104a includes a masking target portion corrected by the user in the “masking target notation” column as shown in FIG. Will be changed to “Information and Communication Laboratories”. Further, the masking confirmation unit 114 is activated (step S207 in FIG. 2), and the masking result list 104a is recorded in the “confirmation” column for each masking target as shown in FIG. 9 (FIG. 9). In this case, it is indicated by “○”.) Here, when the “To Document Screen” button B2 is further pressed, as shown in FIG. 16, on the screen, the highlighted part is changed to an underlined display as described above, and the masked part including the corrected part is masked. This indicates that the target location has been confirmed.
[0086]
Further, in the screen display state shown in FIG. 19, when the “bottom processing” button B5 at the center bottom of the screen is pressed, the checked notation is processed. That is, the prone processing unit 115 is activated (step S209 in FIG. 2), and the masking result list 104a records that the proficiency processing is instructed in the “prone processing” column for each masking target, as shown in FIG. (In FIG. 9, it is represented by “◯” mark). Here, when the “To Document Screen” button B2 is further pressed, as shown in FIG. 17, the masked portion is replaced with, for example, the symbol “x” as described above.
[0087]
Further, in the screen display state shown in FIG. 19, when the “registration execution” button B4 is pressed at the bottom center of the screen (step S210 in FIG. 2), the dictionary registration unit 105 is activated and the word dictionary 112 checks the registration column. A notation including (symbol “×”) is registered. At this time, according to the flowchart shown in FIG. 4, the notation in which the check is input is input to the data selection unit 106 through the dictionary registration unit 105, and among the input words, the character string length, Based on the appearance frequency, the result of morphological analysis, and the masking target specifying result, only those determined as valid registration candidates are registered in the word dictionary 112. It should be noted that the word stored in the “registration” field of the masking result list 104a (recorded with “◯” mark) is not subjected to the processing shown in FIG. It may be registered in the word dictionary 112. Further, all the character strings of the masking target parts for which the confirmation instruction is given without providing the “registration” field for the instruction to register in the word dictionary as shown in FIG. 18 or FIG. The selected character string may be registered in the word dictionary through the processing shown in FIG.
[0088]
The masking result list 104a is updated as shown in FIG. 9 by processing operations of the respective components of the masking correction unit 113, the masking determination unit 114, the concealment processing unit 115, and the dictionary registration unit 105. That is, in the same manner as when operating on the screen shown in FIGS. 14 to 17, the modified masking target notation and the notation before and after it are changed, and the “Confirm” column, “Fuzzy processing” column, “Register” Information corresponding to the user's instruction is entered in the column.
[0089]
As described above, the masking target portion can be confirmed, corrected, confirmed, and processed in the letter form on the document display screens shown in FIGS. 14 to 17 and the list list display screens shown in FIGS. . Further, from the list list display screen, it is possible to instruct dictionary registration reflecting the correction contents to be masked.
[0090]
As described above, when the word dictionary 112 is updated with the correction result of the masking target, for example, in the case of the above example, the notation “communication research institute” may be modified to “information communication research institute”. Similarly, the word “information and communication research institute” may be displayed to the user and inquired about how to read it, the part of speech, the attribute, and the presence / absence of registration. Then, only those instructed by the user on this screen may be registered in the word dictionary 112. At this time, “reading”, “part of speech”, and “attribute” input on the screen shown in FIG. 13 may be registered in the word dictionary 112.
[0091]
It should be noted that when a word is deleted from the word dictionary by correcting the masking portion, the same operation as described above can be performed. For example, among the masking targets displayed on the screen shown in FIG. 15 or FIG. 18, an operation for canceling special display such as reverse display or highlighting of a word to be excluded from the masking target is performed, and then a “confirm execution” button. The fact of deletion is recorded on the masking list result list 104a by operating B3 or operating the “registration execution” button B4. For example, the masking result list 104a may be provided with a “delete” column for this purpose. Words with “o” marks recorded in the “delete” column are then deleted from the word dictionary 112 by the dictionary registration unit 105.
[0092]
Also, on the list list display screen, character strings specified as masking targets are sorted and displayed for each reading method and type, or when there are multiple identical character strings, one of them is displayed. Also good.
[0093]
As described above, according to the above embodiment, the masking target location is determined in the masking target specifying unit 109 from the document input from the input unit 101 based on the word dictionary storing the character string to be masked or a part thereof. The detected masking target location is stored in the masking result list 104a (storage means), and the masking target location stored in the masking result list is displayed on the display screen. When one of the displayed masking target locations is corrected by the user, the masking target location stored in the masking result list is rewritten with the masking target location corrected by the user, and stored in the rewritten masking result list. By masking the masking target part in the document based on the masking target part, it is possible to easily confirm and correct the detected masking target part (such as a proper noun to be replaced with a letter).
[0094]
Further, in the document displayed on the display screen, the new masking target portion designated by the user is stored in the masking result list, and the character string of the new masking target portion stored later in this list is stored in the word dictionary. By storing the information, it is possible to easily update the word dictionary used for detecting proper nouns and the like based on the contents of the correction along with the confirmation and correction of the masking target portion displayed on the display screen.
[0095]
Also, a plurality of character strings are input from the input unit 101, and the data selection unit 106 selects the number of characters of each character string, the character type constituting each character string, and the existing character strings from the plurality of character strings. Based on at least one of the appearance frequency in the document and the result of morphological analysis of each character string, a character string to be stored in the word dictionary is selected, and a character string that cannot be morphologically analyzed among the selected character strings, By storing in the word dictionary a character string that cannot be detected as a masking target location based on each phrase and word dictionary obtained as a result of morphological analysis of each character string in the selected character string, The word dictionary 112 used for the detection can be easily constructed and updated.
[0096]
As described above, according to the above-described embodiment, the masking target portion detected from the input document can be easily confirmed and corrected, and the masking target portion detected from the input document can be detected with high accuracy. Dictionaries can be easily constructed and updated.
[0097]
Therefore, according to the document processing apparatus according to the above-described embodiment, the masking target portion can be detected from the document with high accuracy, and the masking target portion can be concealed by replacing or filling the masking target portion. Prevent infringement of privacy information such as proper nouns in advance and facilitate document sharing and distribution.
[0098]
The method of the present invention described in the embodiment of the present invention is a program that can be executed by a computer, such as a magnetic disk (flexible disk, hard disk, etc.), an optical disk (CD-ROM, DVD, etc.), a semiconductor memory, etc. It can be stored in a medium and distributed.
[0099]
In addition, this invention is not limited to the said embodiment, In the implementation stage, it can change variously in the range which does not deviate from the summary. Furthermore, the above embodiments include inventions at various stages, and various inventions can be extracted by appropriately combining a plurality of disclosed constituent elements. For example, even if some constituent elements are deleted from all the constituent elements shown in the embodiment, the problem (at least one of them) described in the column of problems to be solved by the invention can be solved, and the column of the effect of the invention If at least one of the effects described in (1) is obtained, a configuration in which this configuration requirement is deleted can be extracted as an invention.
[0100]
【The invention's effect】
As described above, according to the present invention, the proper noun part detected from the document can be easily confirmed and corrected.
[0101]
In addition, the word dictionary used for detecting proper noun parts in the document can be easily updated.
[Brief description of the drawings]
FIG. 1 is a diagram showing a configuration example of a document masking apparatus according to an embodiment of the present invention.
FIG. 2 is a flowchart for explaining a processing operation of a control unit.
FIG. 3 is a flowchart for explaining a processing operation of a masking target specifying unit;
FIG. 4 is a flowchart for explaining a processing operation of a data selection unit.
FIG. 5 is a diagram showing an example of a data structure of a word dictionary.
FIG. 6 is a diagram showing an example of a masking rule.
FIG. 7 is a diagram for explaining morphological analysis.
FIG. 8 is a diagram for explaining information stored in a masking result list;
FIG. 9 is a diagram for explaining information stored in a masking result list;
FIG. 10 is a diagram for explaining information stored in a registration candidate list.
FIG. 11 is a diagram for explaining information stored in a registration candidate list.
FIG. 12 is a diagram showing an example of a word dictionary update result.
FIG. 13 is a diagram showing a screen display example for inputting information stored in a word dictionary.
FIG. 14 is a diagram showing an example of a document display screen that displays a portion to be masked.
FIG. 15 is a diagram showing an example of a document display screen that displays a corrected masking target portion.
FIG. 16 is a diagram showing an example on a document display screen when a masking target portion is confirmed.
FIG. 17 is a diagram showing an example of a document display screen when a masking target portion is masked.
FIG. 18 is a diagram showing an example of a list list display screen for displaying masking target portions.
FIG. 19 is a diagram showing an example of a list list display screen that displays a corrected masking target portion and the like.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 101 ... Input part 102 ... Control part 103 ... Output part 104 ... Temporary memory | storage part 105 ... Dictionary registration part 106 ... Data selection part 107 ... Frequency calculation part 108 ... Existing document memory | storage part 109 ... Masking object Specific part 110: Morphological analysis part 111 ... Masking rule storage part 112 ... Word dictionary 113 ... Masking correction part 114 ... Masking confirmation part 115 ... Prone word processing part

Claims (12)

First storage means for storing a dictionary used for morphological analysis, including notation and attributes of each word;
Second storage means for storing an array of word attributes included in each clause in the character string to be masked;
A third storage means;
Document processing means;
A document processing method in a document processing apparatus comprising:
A first registration step in which the document processing means registers dictionary data including the notation of each word and its attributes in the dictionary stored in the first storage means;
An input step in which the document processing means inputs a document;
The document processing means divides the document into phrases and words by the morphological analysis,
A step of detecting, from the document, one or a plurality of consecutive clauses having the attribute array stored in the second storage unit as a masking target portion from the document;
A first storage step in which the document processing means stores the masking target portion detected in the detection step in the third storage means;
The document processing means displaying the document on a display screen and displaying a masking target portion stored in the third storage means in the document; and
(A) When the masking target location in the document displayed on the display screen is corrected by the user, the document processing means displays the masking target location stored in the third storage means. (B) when a new masking target location is designated by the user in the document displayed on the display screen, the document processing means performs the new masking target location. A correction step of storing the target portion in the third storage means;
A masking step in which the document processing means masks a masking target portion stored in the third storage means in the document;
Among the new masking target locations stored in the third storage unit, the document processing unit stores new dictionary data including the notation of the character string of the masking target location designated by the user and its attributes in the dictionary. A second registration step to register;
A document processing method. First storage means for storing a dictionary used for morphological analysis, including notation and attributes of each word;
Second storage means for storing an array of word attributes included in each clause in the character string to be masked;
A third storage means;
Document processing means;
A document processing method in a document processing apparatus comprising:
A first registration step in which the document processing means registers dictionary data including the notation of each word and its attributes in the dictionary stored in the first storage means;
An input step in which the document processing means inputs a document;
The document processing means divides the document into phrases and words by the morphological analysis,
A step of detecting, from the document, one or a plurality of consecutive clauses having the attribute array stored in the second storage unit as a masking target portion from the document;
A first storage step in which the document processing means stores the masking target portion detected in the detection step in the third storage means;
The document processing means displaying the document on a display screen and displaying a masking target portion stored in the third storage means in the document; and
(A) When the masking target location in the document displayed on the display screen is corrected by the user, the document processing means displays the masking target location stored in the third storage means. (B) when a new masking target location is designated by the user in the document displayed on the display screen, the document processing means performs the new masking target location. A correction step of storing the target portion in the third storage means;
A masking step in which the document processing means masks a masking target portion stored in the third storage means in the document;
Among the new masking target locations stored in the third storage means, the document processing means has a character string length equal to or greater than a predetermined value from the character string of the masking target location designated by the user. And a character string that includes characters other than hiragana and has an appearance frequency in an existing document that is equal to or higher than a predetermined value, and that cannot be detected as a masking target location in the current dictionary, A first selection step of selecting as a character string to be stored in the dictionary;
A second registration step of registering in the dictionary new dictionary data including the notation of the character string selected in the first selection step and its attributes;
A document processing method.   The correction to the masking target location displayed on the display screen by the user includes the correction of adding at least one character immediately before and after the masking target location in the document to the masking target location, 3. The document processing method according to claim 1, wherein the document processing method is at least one of corrections for deleting at least one character from a character string at a location. The first registration step includes:
Entering a list containing multiple strings;
Of the plurality of character strings, a character string length is equal to or greater than a predetermined value from a character string group excluding a character string including a phrase having an arrangement stored in the second storage unit , and other than a hiragana character And a character string having an appearance frequency in an existing document that is equal to or higher than a predetermined value and that cannot be morphologically analyzed or that cannot be detected as a masking target location in the current dictionary. A second selection step for selecting as a character string to be stored;
Registering new dictionary data including the notation of the character string selected in the second selection step and its attributes in the dictionary;
The document processing method according to claim 1, further comprising: The first registration step includes:
Entering a list containing multiple strings;
A character string whose character string length is equal to or greater than a predetermined value and includes characters other than hiragana and whose appearance frequency in an existing document is equal to or greater than a predetermined value is selected from the plurality of character strings. 3 selection steps;
Of the character strings selected in the third selection step, registering new dictionary data including the notation of the character string that cannot be morphologically analyzed and its attributes in the dictionary;
Of the character strings selected in the third selection step, the character attribute array included in each clause obtained as a result of morphological analysis of each character string is not stored in the second storage means. Registering new dictionary data including column notation and its attributes in the dictionary;
The document processing method according to claim 1, further comprising: First storage means for storing a dictionary used for morphological analysis, including notation and attributes of each word;
Second storage means for storing an array of word attributes included in each clause in the character string to be masked;
A means of entering a document;
Means for dividing the document into clauses and words by the morphological analysis;
Detecting means for detecting, from the document, one or a plurality of continuous clauses having the attribute array stored in the second storage means as masking target portions;
A third storage means for storing the masking target portion detected by the detection means;
Display means for displaying the document on a display screen and displaying a masking target portion stored in the third storage means in the document;
When the masking target part in the document displayed on the display screen is corrected by the user, the masking target part stored in the third storage unit is rewritten with the corrected masking target part. (B) a correction means for storing the new masking target location in the third storage means when a new masking target location is designated by the user in the document displayed on the display screen; ,
Means for masking a masking target portion stored in the third storage means in the document;
First registration for registering new dictionary data including notation and attribute of a character string of a masking target location designated by a user among new masking target locations stored in the third storage means in the dictionary Means,
A document processing apparatus comprising: First storage means for storing a dictionary used for morphological analysis, including notation and attributes of each word;
Second storage means for storing an array of word attributes included in each clause in the character string to be masked;
A means of entering a document;
Means for dividing the document into clauses and words by the morphological analysis;
Detecting means for detecting, from the document, one or a plurality of continuous clauses having the attribute array stored in the second storage means as masking target portions;
A third storage means for storing the masking target portion detected by the detection means;
Display means for displaying the document on a display screen and displaying a masking target portion stored in the third storage means in the document;
When the masking target part in the document displayed on the display screen is corrected by the user, the masking target part stored in the third storage unit is rewritten with the corrected masking target part. (B) a correction means for storing the new masking target location in the third storage means when a new masking target location is designated by the user in the document displayed on the display screen; ,
Means for masking a masking target portion stored in the third storage means in the document;
Among the new masking target locations stored in the third storage means, characters other than the hiragana character string whose length is longer than a predetermined value from the character string of the masking target location designated by the user A character string that is included in the existing document and that has a frequency of appearance equal to or higher than a predetermined value, and that cannot be detected as a masking target location in the current dictionary, First selection means for selecting as a column;
First registration means for registering new dictionary data including the notation of the character string selected by the first selection means and its attributes in the dictionary;
A document processing apparatus comprising:   The correction to the masking target location displayed on the display screen by the user includes the correction of adding at least one character immediately before and after the masking target location in the document to the masking target location, 8. The document processing apparatus according to claim 6, wherein the document processing apparatus is at least one of corrections for deleting at least one character from a character string at a location. Means for entering a list containing multiple strings;
Of the plurality of character strings, a character string length is equal to or greater than a predetermined value from a character string group excluding a character string including a phrase having an arrangement stored in the second storage unit, and other than a hiragana character And a character string having an appearance frequency in an existing document that is equal to or higher than a predetermined value and that cannot be morphologically analyzed or that cannot be detected as a masking target location in the current dictionary. Second selection means for selecting as a character string to be stored;
Second registration means for registering new dictionary data including the notation of the character string selected by the second selection means and its attributes in the dictionary;
The document processing apparatus according to claim 7, further comprising: Means for entering a list containing multiple strings;
A character string whose character string length is equal to or greater than a predetermined value and includes characters other than hiragana and whose appearance frequency in an existing document is equal to or greater than a predetermined value is selected from the plurality of character strings. 3 selection means;
Of the character strings selected by the third selection means, means for registering new dictionary data including the notation of the character string that cannot be morphologically analyzed and its attributes in the dictionary;
Among the character strings selected by the third selection means, the character attribute array included in each clause obtained as a result of morphological analysis of each character string is not stored in the second storage means. Means for registering in the dictionary new dictionary data including column notation and attributes thereof;
The document processing apparatus according to claim 7, further comprising: Computer
First storage means for storing a dictionary used for morphological analysis, including the notation and attributes of each word;
Second storage means for storing an array of word attributes included in each clause in the character string to be masked;
A means of entering documents,
Means for dividing the document into clauses and words by the morphological analysis;
Detecting means for detecting, from the document, one or a plurality of consecutive clauses having the attribute arrangement stored in the second storage means as masking target portions;
A third storage means for storing the masking target portion detected by the detection means;
Display means for displaying the document on a display screen and displaying a masking target portion stored in the third storage means in the document;
When the masking target part in the document displayed on the display screen is corrected by the user, the masking target part stored in the third storage unit is rewritten with the corrected masking target part. (B) When a new masking target location is instructed by the user in the document displayed on the display screen, a correction unit that stores the new masking target location in the third storage unit,
Means for masking a portion to be masked stored in the third storage means in the document;
Of the new masking target locations stored in the third storage means, registration means for registering new dictionary data including the notation of the character string of the masking target location designated by the user and its attributes in the dictionary,
Program to function as. Computer
First storage means for storing a dictionary used for morphological analysis, including the notation and attributes of each word;
Second storage means for storing an array of word attributes included in each clause in the character string to be masked;
A means of entering documents,
Means for dividing the document into clauses and words by the morphological analysis;
Detecting means for detecting, from the document, one or a plurality of consecutive clauses having the attribute arrangement stored in the second storage means as masking target portions;
A third storage means for storing the masking target portion detected by the detection means;
Display means for displaying the document on a display screen and displaying a masking target portion stored in the third storage means in the document;
When the masking target part in the document displayed on the display screen is corrected by the user, the masking target part stored in the third storage unit is rewritten with the corrected masking target part. (B) When a new masking target location is instructed by the user in the document displayed on the display screen, a correction unit that stores the new masking target location in the third storage unit,
Means for masking a portion to be masked stored in the third storage means in the document;
Among the new masking target locations stored in the third storage means, characters other than the hiragana character string whose length is longer than a predetermined value from the character string of the masking target location designated by the user A character string that is included in the existing document and that has a frequency of appearance equal to or higher than a predetermined value, and that cannot be detected as a masking target location in the current dictionary, A first selection means for selecting as a column;
Registration means for registering new dictionary data including notation of the character string selected by the first selection means and its attributes in the dictionary;
Program to function as.

Images