CN106445963A - Advertisement index keyword automatic generation method and apparatus for APP platform - Google Patents
Advertisement index keyword automatic generation method and apparatus for APP platform Download PDFInfo
- Publication number
- CN106445963A CN106445963A CN201510488047.1A CN201510488047A CN106445963A CN 106445963 A CN106445963 A CN 106445963A CN 201510488047 A CN201510488047 A CN 201510488047A CN 106445963 A CN106445963 A CN 106445963A
- Authority
- CN
- China
- Prior art keywords
- app
- word
- search
- advertisement
- search word
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 44
- 238000000605 extraction Methods 0.000 claims description 14
- 239000000284 extract Substances 0.000 claims description 10
- 230000008569 process Effects 0.000 abstract description 6
- 101100055496 Arabidopsis thaliana APP2 gene Proteins 0.000 description 6
- 101100016250 Saccharomyces cerevisiae (strain ATCC 204508 / S288c) GYL1 gene Proteins 0.000 description 6
- 239000013598 vector Substances 0.000 description 6
- 238000005516 engineering process Methods 0.000 description 5
- 101150053844 APP1 gene Proteins 0.000 description 4
- 101100189105 Homo sapiens PABPC4 gene Proteins 0.000 description 4
- 102100039424 Polyadenylate-binding protein 4 Human genes 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 239000000203 mixture Substances 0.000 description 3
- 230000001737 promoting effect Effects 0.000 description 3
- 230000011218 segmentation Effects 0.000 description 3
- 235000013399 edible fruits Nutrition 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000001035 drying Methods 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000012913 prioritisation Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/335—Filtering based on additional data, e.g. user or group profiles
- G06F16/337—Profile generation, learning or modification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/31—Indexing; Data structures therefor; Storage structures
- G06F16/313—Selection or weighting of terms for indexing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0277—Online advertisement
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Finance (AREA)
- Data Mining & Analysis (AREA)
- Strategic Management (AREA)
- General Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- General Business, Economics & Management (AREA)
- Entrepreneurship & Innovation (AREA)
- Computational Linguistics (AREA)
- Game Theory and Decision Science (AREA)
- Marketing (AREA)
- Economics (AREA)
- Software Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses an advertisement index keyword automatic generation method and apparatus for an APP platform, and relates to the technical field of search. The method comprises the steps of calculating topic distribution of APPs for description information of the APPs; calculating topic distribution of search words according to click relationships between the search words in a search history record and the APPs; for the search words, calculating topic similarity between the topic distribution of the search words and the topic distribution of the advertisement APPs; if the topic similarity is greater than a topic threshold, taking the search words as keywords of the advertisement APPs; and constructing advertisement index keywords corresponding to the advertisement APPs according to corresponding relationships between the keywords and the advertisement APPs. According to the method and the apparatus, the advertisement index keywords can be automatically selected for the advertisement APPs of an APP supplier, so that the process of selecting the advertisement index keywords by the APP supplier is removed, the user experience of the APP platform can be improved, and the income of the APP platform can be increased.
Description
Technical field
The present invention relates to search technique field is and in particular to a kind of keyed advertising key word of APP platform
Automatic generation method and device.
Background technology
With the development of intelligent mobile terminal, increasing user downloads respectively in intelligent mobile terminal
Plant APP (application, application) to use.Based on this kind of situation, APP platform arises at the historic moment, Yong Huke
So that APP platform is accessed by intelligent mobile terminal, such as pass through the APP installing in intelligent mobile terminal
Delivery applications go to access APP platform, such that it is able to download various APP from platform.Wherein, APP
The such as various mobile phone assistant of delivery applications.
And in APP platform, in order to the APP owner for there being popularization demand, such as App supply
The APP of this APP owner forward can be shown, APP has in APP searched page by business
Person can buy for these APP and bid word using as keyed advertising key word.
But, the word of bidding that APP supplier buys may be mismatched with APP itself, makes application platform
Search engine when line retrieval is entered according to the search word of user input, may return actually search with this
The information of the very low APP of rope word degree of association, when leading to user to search the APP with its demand, needs
More operated, such as page turning etc. operates, impact obtains the efficiency of the APP of its demand.And
And, for APP supplier, because it needs oneself to select various words of bidding, complex operation.
Content of the invention
In view of the above problems it is proposed that the present invention is to provide one kind to overcome the problems referred to above or at least portion
The keyed advertising key word automatically generating device of APP platform solving the above problems with dividing is with accordingly
The keyed advertising key word automatic generation method of APP platform.
According to one aspect of the present invention, there is provided a kind of keyed advertising key word of APP platform is automatic
Generation method, including：
For the description information of each APP, calculate the theme distribution of each APP；
According to the click relation of search word in search history record and each APP, calculate the master of each search word
Topic distribution；
For each search word, calculate the theme distribution of described search word and the theme distribution of advertisement APP
Between Topic Similarity；
If described Topic Similarity is more than theme threshold value, using described search word as described advertisement
The key word of APP；
According to the corresponding relation of described key word and advertisement APP, build the advertisement rope of corresponding advertisement APP
Draw key word.
Preferably, for each search word, theme distribution and the advertisement APP of described search word are calculated
Theme distribution between Topic Similarity before also include：
Judge the volumes of searches whether great-than search amount threshold value of described search word；
If the volumes of searches great-than search amount threshold value of described search word, enter and be directed to each search word,
Calculate the Topic Similarity between the theme distribution of described search word and the theme distribution of advertisement APP.
Preferably, the described description information for each APP, calculates the theme distribution of each APP, including：
For the description information of each APP, according to latent Dirichletal location topic model, calculate each APP
Theme distribution.
Preferably, in the corresponding relation according to described key word and advertisement APP, build corresponding advertisement APP
Keyed advertising key word before, also include：
Directly from the title of advertisement APP and/or label, extract the key word of corresponding advertisement APP.
Preferably, in the corresponding relation according to described key word and advertisement APP, build corresponding advertisement APP
Keyed advertising key word before, also include：
For each search word in search Download History, calculate between search word and the title of advertisement APP
Text similarity；
If described text similarity is more than text similar threshold value, obtain described search word as key
Word.
Preferably, in the corresponding relation according to described key word and advertisement APP, build corresponding advertisement APP
Keyed advertising key word before, also include：
Search history record is searched for each search word in Download History, judge described search word
Whether independent access download time is more than independent access threshold value, and the classification of described search word and advertisement
Whether the classification of APP belongs to same classification；
If the independent access download time of described search word is more than independent access threshold value, and described search
The classification of rope word belongs to same classification, then using described search word as key with the classification of advertisement APP
Word.
Preferably, in the corresponding relation according to described key word and advertisement APP, build corresponding advertisement APP
Keyed advertising key word before, also include：
For each one-level class APP now, using the description information of one-level class each APP now, adopt
With grader, each APP is divided into corresponding one-level class two grades of classifications now；
To each search word, according to the click relation of search word in search history record and each APP, one
Two grades of classifications belonging to the described each APP of level, calculate two grades of classifications corresponding to described search word；
Two grades of classifications being located according to advertisement APP, obtain to should each search word of two grades of classifications then make
For key word.
According to another aspect of the present invention, there is provided a kind of keyed advertising key word of APP platform
Automatically generating device, including：
APP theme distribution computing module, is suitable to the description information for each APP, calculates each APP's
Theme distribution；
Search word theme distribution computing module, is suitable to according to search word in search history record and each APP
Click relation, calculate each search word theme distribution；
Theme similarity word extraction module, is suitable to, for each search word, calculate described search word
Topic Similarity between the theme distribution of theme distribution and advertisement APP；
First key word adds module, if being suitable to described Topic Similarity to be more than theme threshold value, will
Described search word is as the key word of described advertisement APP；
Keyed advertising key word builds module, is suitable to according to described key word pass corresponding with advertisement APP
System, builds the keyed advertising key word of corresponding advertisement APP.
Preferably, also included before theme similarity word extraction module：
Volumes of searches judge module, is suitable to judge the volumes of searches whether great-than search amount threshold of described search word
Value；If the volumes of searches great-than search amount threshold value of described search word, enter theme similarity word and carry
Delivery block.
Preferably, described pin APP theme distribution computing module includes：
Latent Dirichletal location topic model computing module, is suitable to the description information for each APP,
According to latent Dirichletal location topic model, calculate the theme distribution of each APP.
Preferably, before keyed advertising key word builds module, also include：
Extracting directly module, is suitable to, directly from the title of advertisement APP and/or label, extract corresponding wide
Accuse the key word of APP.
Preferably, before keyed advertising key word builds module, also include：
Text search word acquisition module, is suitable to for each search word in search Download History, calculating is searched
Text similarity between the title of rope word and advertisement APP；
Second key word adds module, if being suitable to described text similarity to be more than text similar threshold value,
Then obtain described search word as key word.
Preferably, before keyed advertising key word builds module, also include：
Independent access search word extraction module, is suitable to in the search Download History of search history record
Each search word, judge that whether the independent access download time of described search word is more than independent access threshold
It is worth, and whether the classification of described search word belongs to same classification with the classification of advertisement APP；
3rd key word adds module, if the independent access download time being suitable to described search word is more than
Independent access threshold value, and the classification of the classification of described search word and advertisement APP belongs to same class
Mesh, then using described search word as key word.
Preferably, before keyed advertising key word builds module, also include：
APP classification segments module, is suitable to the APP now for each one-level class, using one-level class now
Each APP description information, each APP is divided into by corresponding one-level class two grades of classes now using grader
Mesh；
Search word sort module, is suitable to each search word, according to search word in search history record with
The click relation of each APP, two grades of classifications belonging to each APP described in one-level, calculate described search word institute
Corresponding two grades of classifications；
Class heading search word extraction module, is suitable to two grades of classifications being located according to advertisement APP, obtains and correspond to
Each search word of this two grades of classifications is then as key word.
The keyed advertising key word automatic generation method of the APP platform according to the present invention and device, for
APP supplier needs the advertisement APP promoting, and in APP platform, (is wherein wrapped using each APP
Include advertisement APP) article calculate the theme distribution of each APP, and the point using search word and APP
Hit relation, calculate the theme distribution of each search word, hence for advertisement APP, advertisement can be calculated
Topic Similarity between APP and search word, using Topic Similarity be more than theme threshold value as this advertisement
The key word of APP, such that it is able in index building using this key word as this advertisement APP advertisement
Indexing key words, thus solving APP supplier needs to select keyed advertising to close by loaded down with trivial details operation
Key word problem, and the keyed advertising key word due to selecting is incorrect, leads to its advertisement APP to go out
Problem in the very low Search Results of present and user input search word degree of association, achieving can be automatic
Advertisement APP for APP supplier automatically selects keyed advertising key word, reduces APP supplier to wide
Accuse the selection course of indexing key words, and avoid advertisement APP to occur in the search word with user input
In the very low Search Results of degree of association, improve the Consumer's Experience of application platform, and can be user input
Search word provide theme similar APP, improve the beneficial effect of retrieval width.
Described above is only the general introduction of technical solution of the present invention, in order to better understand the present invention's
Technological means, and being practiced according to the content of description, and in order to allow the above-mentioned of the present invention and
Other objects, features and advantages can become apparent, below especially exemplified by the specific embodiment party of the present invention
Formula.
Brief description
By reading the detailed description of hereafter preferred implementation, various other advantages and benefit for
Those of ordinary skill in the art will be clear from understanding.Accompanying drawing is only used for illustrating the mesh of preferred implementation
, and it is not considered as limitation of the present invention.And in whole accompanying drawing, with identical with reference to symbol
Number represent identical part.In the accompanying drawings：
Fig. 1 shows a kind of keyed advertising key word of APP platform according to an embodiment of the invention certainly
The flow chart of dynamic generation method；
Figure 1A shows a kind of displaying example of advertisement APP according to an embodiment of the invention；
Fig. 2 shows a kind of keyed advertising key word of APP platform according to an embodiment of the invention certainly
The flow chart of dynamic generation method；
Fig. 3 shows a kind of keyed advertising key word of APP platform according to an embodiment of the invention certainly
The structured flowchart of dynamic generating means；
Fig. 4 shows a kind of keyed advertising key word of APP platform according to an embodiment of the invention certainly
The structured flowchart of dynamic generating means.
Specific embodiment
It is more fully described the exemplary embodiment of the disclosure below with reference to accompanying drawings.Although it is aobvious in accompanying drawing
Show the exemplary embodiment of the disclosure it being understood, however, that may be realized in various forms the disclosure
And should not be limited by embodiments set forth here.On the contrary, these embodiments are provided to be able to more
Thoroughly understand the disclosure, and can be by the technology conveying to this area complete for the scope of the present disclosure
Personnel.
With reference to Fig. 1, show the keyed advertising key word of the APP platform according to an embodiment of the present invention
The flow chart of automatic generation method, the method specifically may comprise steps of：
Step 101, for the description information of each APP, calculates the theme distribution of each APP；
The embodiment of the present invention is applied to APP platform, and APP supplier can be registered in APP platform
APP is uploaded in account.Certainly, supplier can specify APP in going account is advertisement APP, and
Pay certain value data for it, then, APP platform would know that this APP is advertisement APP.
In embodiments of the present invention, when calculating the theme distribution of each APP, its article sample is each wide
Accuse the description information of APP and non-advertisement APP.This description information can be understood as being discussed in detail of APP
Information.
Description information based on each APP, it is possible to use topic model is analyzed, by each APP
Theme distribution calculate.The theme distribution of calculated APP includes the theme of advertisement APP
Distribution.Theme distribution can be understood as an article, and the probability distribution of its theme, such as literary composition
Chapter 1, the probability of its theme 1 is 0.6, and the probability of theme 2 is 0.3, obtains a theme vector
(0.6,0.4) it is possible to be interpreted as the theme distribution of article 1.
Step 102, according to the click relation of search word in search history record and each APP, calculates and respectively searches
The theme distribution of rope word；
In the search procedure of user, possible its is clicked in result of page searching and is checked that APP's is detailed
Information and do not download it is also possible to click on download APP, thus form a series of for search word
Click relation.In embodiments of the present invention, described click relation can only include search word and check
Click relation between APP is it is also possible to the click only including between search word and each APP of download is closed
System, it is, of course, also possible to include the click relation between search word and the APP checking and search word
Click relation and each APP downloading between.
Can be according to the click relation of search word and each APP and each in step 101 in the embodiment of the present invention
The theme distribution of individual APP calculates the theme distribution of each search word.I.e. can be according to search word and each APP
Click relation, one search word of statistics is distributed to the click of each APP, and such as search word 1 is clicked on
APP1 accounting 0.8, the accounting clicking on APP2 is 0.2, then the click of search word 1 is distributed as APP2
(0.8,0.2), its two sequentially corresponding A PP1, APP2.
Such as search word 1 clicks on APP1 accounting 0.8, and the accounting clicking on APP2 is 0.2, and APP1
Theme distribution is (0.6,0.4), and the theme distribution of APP2 is (0.7,0.3), then search word
Theme distribution can be ((0.6+.07) * 0.8, (0.4+0.3) * 0.2)；Wherein, theme distribution is every
Item corresponds to theme 1, theme 2 respectively.
Step 103, for each search word, the theme distribution calculating described search word is with advertisement APP's
Topic Similarity between theme distribution；If described Topic Similarity is more than theme threshold value, enter
Step 103；
In embodiments of the present invention, for advertisement APP, its theme distribution can be extracted from step 101
Vector, be then directed to each search word, its theme distribution vector can be extracted.Due to both
Vector, then can calculate the similarity between two vectors.
In embodiments of the present invention, between the theme distribution for search word and the theme distribution of APP
Similarity, can be calculated using KL distance and/or JS distance.Wherein, KL distance is
Kullback Leibler divergence, also known as relative entropy, he is general to two of a discrete random variable
For rate distribution P and Q, their KL divergence is defined as：... formula (1).
When wherein seeking log, with 2 as bottom.
It is Jensen Shannon divergence for JS distance, it is the prioritization scheme of KL distance,
Its formula is：
... formula (2),
Wherein... (formula 3).Wherein, D adopts formula (1) to calculate.
JSD value is between 0 to 1.Bigger represent that two theme distribution are more consistent, similarity is higher.
The theme distribution of the theme distribution of the search word of the present invention and advertisement APP corresponds to P and Q respectively.
Certainly, also the theme distribution of search word and the master of advertisement APP can also be calculated using other distances
Topic Similarity between topic distribution, such as COS distance.
The embodiment of the present invention preferably, adopts COS distance and JS distance simultaneously, then to this two away from
From averaging or carry out weight addition calculation.So, calculated similarity is more accurate.
In embodiments of the present invention, a theme threshold value can be preset according to actual test, if theme phase
Then enter step 104 like degree more than theme threshold value.
Step 104, using described search word as described advertisement APP key word；
Theme threshold search word is more than for the Topic Similarity with an advertisement APP, this can be searched
Rope word adds the keyword set of this advertisement APP.
Step 105, according to the corresponding relation of described key word and advertisement APP, builds corresponding advertisement APP
Keyed advertising key word.
Exist in the embodiment of the present invention, APP platform can arrange two sets of indexes, a set of for common
The index that APP builds, and the another set of index building for advertisement APP.And it is directed to advertisement APP structure
When indexing, then the keyword set of each advertisement APP being obtained using step 101-104, go to construct rope
Draw, the key word in keyword set is then configured to the keyed advertising key word in index.
Certainly, in actual applications, for the different advertisement with same keyed advertising key word
APP, then can be arranged these advertisements APP in index list after keyed advertising key word
Row.
And for APP platform, after receiving the search word of user input, then can be respectively for general
The index of logical APP and scanning for, for the first of common APP in the index of advertisement APP
In Search Results, each advertisement APP of the indexed search for advertisement APP, in the first search knot
Sort forward displaying in fruit.
In embodiments of the present invention, for aforementioned index, can be incited somebody to action by advertisement and identifier in application platform
It is labeled as advertisement APP, then when retrieving APP again, if this APP has advertisement and identifier, can
So that it to be shown in advance.This advertisement and identifier such as " popularization ", " recommending ".Additionally, the embodiment of the present invention
In multiple advertisement and identifiers can be set, different advertisement and identifiers possesses different displaying weights.Such as
The displaying weight of " popularization " is high, and " recommending " shows that weight is less than the displaying weight of " popularization ".As figure
1A, mark " popularization " and " recommending " printed words for advertisement APP, then love is advanced and is managed money matters and favourable net
Manage money matters as advertisement app.Search " financing " keyword represents above-mentioned advertisement app.
In sum, the keyed advertising key word automatic generation method of APP platform and device, for APP
Supplier needs the advertisement APP promoting, in APP platform, using each APP (including wide
Accuse APP) article calculate the theme distribution of each APP, and closed using the click of search word and APP
System, calculates the theme distribution of each search word, hence for advertisement APP, can calculate advertisement APP
Topic Similarity and search word between, Topic Similarity is more than theme threshold value as this advertisement APP
Key word, such that it is able in index building using this key word as this advertisement APP keyed advertising
Key word, thus solving APP supplier needs to select keyed advertising key word by loaded down with trivial details operation
Problem, and due to select keyed advertising key word incorrect, lead to its advertisement APP to occur in
Problem in the Search Results very low with the search word degree of association of user input.The embodiment of the present invention can be certainly
Move the advertisement APP for APP supplier and automatically select keyed advertising key word, reduce APP supplier couple
The selection course of keyed advertising key word, and avoid advertisement APP to occur in the search with user input
In the very low Search Results of word degree of association, improve the Consumer's Experience of application platform.
And, in current technology, search engine according to the retrieval with the text relevant of search word,
The present invention can provide theme similar APP for the search word of user input, improves retrieval width.
With reference to Fig. 2, show the keyed advertising key word of the APP platform according to an embodiment of the present invention
The flow chart of automatic generation method, the method specifically may comprise steps of：
Step 201, for the description information of each APP, according to latent Dirichletal location topic model,
Calculate the theme distribution of each APP；
In embodiments of the present invention, because the description information of APP can essentially be interpreted as a literary composition
Chapter, above-mentioned topic model can (Latent Dirichlet Allocation, potential Di Li Cray divides for LDA
Join theme) model, LDA is a kind of document subject matter generation model.So-called generation model, that is,
Each word thinking an article is by " with certain probability selection certain theme, and from this
With certain word of certain probability selection in theme " such a process obtains.So, if we will
Generate a document, the probability that each word inside it occurs is：
Therefore, the embodiment of the present invention can be analyzed to each article by LDA model, obtains
To the theme distribution of each description information corresponding, the i.e. probability distribution of each theme, the such as probability of theme 1
For 0.6, the probability of theme 2 is 0.3, obtains a vector (0.6,0.4).Then according to description letter
Breath and the corresponding relation of APP, you can obtain the theme distribution of APP.
Certainly, in the embodiment of the present invention, for the calculating of APP theme distribution, can also be adopted other
Topic model carry out, the present invention is not any limitation as to it.
Step 202, according to the click relation of search word in search history record and each APP, calculates and respectively searches
The theme distribution of rope word；
Step 203, judges the volumes of searches whether great-than search amount threshold value of described search word；If described search
The volumes of searches great-than search amount threshold value of rope word, then enter step 204；
In actual applications, some search word volumes of searches some search word volumes of searches little are big, for will push away
For wide APP, the big search word of volumes of searches, because being easier to be retrieved, is easier to corresponding
APP is promoted, thus improving the popularization efficiency of the advertisement APP of APP platform.Thus the present invention is then
The volumes of searches of each search word is counted in search history record, and preset search amount threshold value, if right
In the search word of volumes of searches great-than search amount threshold value, just enter step 205.
Step 204, for each search word, the theme distribution calculating described search word is with advertisement APP's
Topic Similarity between theme distribution；If described Topic Similarity is more than theme threshold value, enter
Step 205；
Step 205, using described search word as described advertisement APP key word；
Preferably, in the corresponding relation according to described key word and advertisement APP, build corresponding advertisement APP
Keyed advertising key word before, also include：
Step A11, directly from the title of advertisement APP and/or label, extracts corresponding advertisement APP's
Key word.
In the embodiment of the present invention, directly can extract character/word from the text message related to advertisement APP
Deng the key word as this advertisement APP.
Specifically, step A11 includes：
Sub-step A111, the title of advertisement APP is carried out participle operation, and word segmentation result is closed as class
Keyword.
In embodiments of the present invention, the identity information of advertisement APP includes title, such as " takes journey travelling ",
So the present invention directly can carry out participle operation to this title, after " taking journey travelling " participle, participle
Result is " taking journey " and " travelling ", then " can take journey " and " travelling " is as advertisement APP
The key word of " taking journey travelling ".
And/or sub-step A112, by the name translation of advertisement APP for pinyin string and/or or by described
Title carries out the word segmentation result that participle obtains and is converted to pinyin string, using described pinyin string as key word；
For the title of advertisement APP, phonetic such as " xiechenglvxing " can be converted it directly to,
Or its word segmentation result is converted to phonetic, such as the phonetic of " taking journey " is " xiecheng ", then these
Phonetic can be used as the key word of this advertisement APP.
And/or sub-step A113, using the label word of advertisement APP as key word.
For the default label word of an advertisement APP, that such as " takes journey to travel " app has people's labour movement
The label word of battalion：" tourism ", " train ticket ", " tourism attack strategy ", " air ticket ", " go out
OK ", " hotel ", then can be using these label words as key word.
Preferably, in the corresponding relation according to described key word and advertisement APP, build corresponding advertisement APP
Keyed advertising key word before, also include：
Step B11, for each search word in search Download History, calculates search word and advertisement APP
Text similarity between title；
In actual applications, user have input search word in the terminal and scans for, and it may be clicked on and download
APP is likely to not download APP, then feelings are downloaded in the search that application platform then can record each search word
Condition, such as user A search for " financing ", downloaded APP1, and user B searches in search results pages
Rope " is managed money matters ", then may download APP2 in search results pages, by under the search to a large number of users
The record of load behavior, then can obtain the search Download History to each search word.
In implementing, this search Download History is stored in application platform in the form of searching for download log
In.
Step B12, if described text similarity is more than text similar threshold value, obtains described search word
As key word.
The embodiment of the present invention can extract the search word that each has used from search download log, calculates
Text similarity between the title of this search word and advertisement APP.As calculated search word text and advertisement
COS distance between APP name text.
The embodiment of the present invention can arrange a text similar threshold value for text similarity, if described literary composition
This similarity is more than text similar threshold value, then obtain described search word as the key word of this advertisement APP.
If described text similarity is less than text similar threshold value, ignore this word.
Preferably, in the corresponding relation according to described key word and advertisement APP, build corresponding advertisement APP
Keyed advertising key word before, also include：
Step B21, searches for each search word in Download History for search history record, judges described
Whether the independent access download time of search word is more than independent access threshold value, and the class of described search word
Whether mesh belongs to same classification with the classification of advertisement APP；
Step B22, if the independent access download time of described search word is more than independent access threshold value, with
And the described classification of search word belongs to same classification, then by described search word with the classification of advertisement APP
As key word.
For a search word in search download log, may there is multiple users this search in terminal display
In the Search Results of word download APP, and its exist the multiple APP of the terminal downloads of same IP or
Same APP has downloaded repeatedly.And download weight to reduce the terminal-pair search word of same IP
Impact, the embodiment of the present invention then counts the independent access download time of each search word, i.e. UV (Unique
Visitor) download, even if that is, the download of the terminal of same IP is repeatedly, its UV download time is also only
Calculate once.Then for a search word, count terminal the searching using this search word of how many IP
Hitch fruit has downloaded APP.
Then, the embodiment of the present invention is provided with the independent access threshold value for UV download time, if sentenced
The UV download time of disconnected search word is more than this independent access threshold value, then can determine whether the classification of described search word
Whether belong to same classification with the classification of advertisement APP, if the now classification of search word and advertisement APP
Classification belong to same classification, then using this search word as this advertisement APP keyword.And for
One search word, its independent access download time is less than or equal to independent access threshold value, and its classification with wide
The classification accusing APP is not belonging to same classification, can ignore this search word.
Certainly, in the embodiment of the present invention, advertisement APP also can be carried out to it as common APP
Classification.For search word it is also possible to classify to it.The specific categorizing process present invention not to itself plus
To limit.It is of course possible to be classified to APP and search word using following steps：
Sub-step B211, for each one-level class APP now, using one-level class each APP's now
Each APP is divided into corresponding one-level class two grades of classifications now using grader by description information；
Default various classification in application platform, this classification from the beginning of first-level class, class of such as having played,
Sport category.And in fact, for an one-level class APP now, can be according to the description of its APP
Information carries out thinner classification.In actual applications, it is possible to use Bayes classifier enters to description information
Row classification, one-level class each APP now is assigned to each two grades of classes now.
Sub-step B212, to each search word, according to search word in search history record with each APP's
Click relation, two grades of classifications belonging to each APP described in one-level, calculate two corresponding to described search word
Level classification.
In the search procedure of user, possible its is clicked in result of page searching and is checked that APP's is detailed
Information and do not download it is also possible to click on lower APP.The embodiment of the present invention can be according to search word and each APP
Click relation, in conjunction with sub-step B211, two grades of classifications APP, each search word is also assigned to
Corresponding two grades of classes are now.Certainly advertisement APP also assists in categorizing process.
The accounting that such as search word 1 clicks on the number of times of APP in two grades of classifications 1 is more than accounting threshold value, then
This search word is grouped under this two grades of classifications 1.
Above-mentioned search word and the click relation of each APP, can check for the click of search word and each APP
Between relation or search word and each APP the relation clicked between downloading, certainly also may be used
Think that the click of search word and each APP is checked and clicked on the total relation between download.
Preferably, in the corresponding relation according to described key word and advertisement APP, build corresponding advertisement
Before the keyed advertising key word of APP, also include：
Step C11, for each one-level class APP now, using the description of one-level class each APP now
Each APP is divided into corresponding one-level class two grades of classifications now using grader by information；
Step C12, to each search word, according to the click of search word in search history record and each APP
Relation, two grades of classifications belonging to each APP described in one-level, calculate two grades of classes corresponding to described search word
Mesh；
Step C11-C12 is similar with aforementioned sub-step B211-B212.Because similar search word 1 clicks on two
The accounting of the number of times of APP in level classification 1 is more than accounting threshold value, then this search word is grouped into this two grades of classes
, there are certain two grades of class now in the situation under mesh 1, the click accounting very little of search word, namely this search
Word is that the probability of this two grades of classifications is little, then can remove it from this two grades of classifications.
After by corresponding for search word two grades of classifications, by should two grades of classifications the little search word of probability
Delete, the search word of remaining two grades of classifications is generated as a word bag, then should in step C13
With.
Step C13, two grades of classifications being located according to advertisement APP, obtain to should respectively the searching of two grades of classifications
Rope word is then as key word.
For advertisement APP, because step C11 calculates two grades of classifications at each APP place, then
Two grades of classifications of advertisement APP also determine, determine the word of the key word of two grades of classifications in step C12
Bag, then can using the word in this word bag as this advertisement APP key word.
In embodiments of the present invention, for an APP, the mode of aforementioned various extraction key words is permissible
Carry out combination in any, the present invention is not any limitation as to it.
In embodiments of the present invention, combine, for by aforementioned various key word acquisition modes, the pass obtaining
Keyword, can be normalized in step 205 first, identical key word is merged, obtains the simplest
Keyword set.
Certainly, the various modes that search history record calculates key word, every two kinds of sides are used for aforementioned
Can be independent to the use of search history record between formula, not affect each other.
Step 206, according to the corresponding relation of described key word and advertisement APP, builds corresponding advertisement APP
Keyed advertising key word.
Exist in the embodiment of the present invention, APP platform can arrange two sets of indexes, a set of for common
The index that APP builds, and the another set of index building for advertisement APP.And it is directed to advertisement APP structure
When indexing, then the keyword set of each advertisement APP being obtained using step 201-205, go to construct rope
Draw, the key word in keyword set is then configured to the keyed advertising key word in index.
In sum, the keyed advertising key word automatic generation method of APP platform and device, for APP
Supplier needs the advertisement APP promoting, in APP platform, using each APP (including wide
Accuse APP) article calculate the theme distribution of each APP, and closed using the click of search word and APP
System, calculates the theme distribution of each search word, hence for advertisement APP, can calculate advertisement APP
Topic Similarity and search word between, Topic Similarity is more than theme threshold value as this advertisement APP
Key word, such that it is able in index building using this key word as this advertisement APP keyed advertising
Key word, thus solving APP supplier needs to select keyed advertising key word by loaded down with trivial details operation
Problem, and due to select keyed advertising key word incorrect, lead to its advertisement APP to occur in
Problem in the Search Results very low with the search word degree of association of user input.The embodiment of the present invention can be certainly
Move the advertisement APP for APP supplier and automatically select keyed advertising key word, reduce APP supplier couple
The selection course of keyed advertising key word, and avoid advertisement APP to occur in the search with user input
In the very low Search Results of word degree of association, improve the Consumer's Experience of application platform.
And, in current technology, search engine according to the retrieval with the text relevant of search word,
The present invention can provide theme similar APP for the search word of user input, improves retrieval width.
Furthermore, the embodiment of the present invention can be entered in conjunction with the acquisition modes of the key word of multiple advertisement APP
One step improves the width of advertisement APP word bag, makes the promoted extension of advertisement APP in retrieving more
Extensively.
Finally, aforesaid way, can improve the recall rate of advertisement APP, improve the income of APP platform.
With reference to Fig. 3, show the keyed advertising key word of the APP platform according to an embodiment of the present invention
The structure chart of automatically generating device, this structure specifically can include with lower module：
APP theme distribution computing module 301, is suitable to the description information for each APP, calculates each APP
Theme distribution；
Search word theme distribution computing module 302, be suitable to according to search word in search history record with each
The click relation of APP, calculates the theme distribution of each search word；
Theme similarity word extraction module 303, is suitable to, for each search word, calculate described search word
Theme distribution and the theme distribution of advertisement APP between Topic Similarity；
Key word adds module 304, if being suitable to described Topic Similarity to be more than theme threshold value, by institute
State search word as the key word of described advertisement APP；
Keyed advertising key word builds module 305, is suitable to corresponding with advertisement APP according to described key word
Relation, builds the keyed advertising key word of corresponding advertisement APP.
With reference to Fig. 4, show the keyed advertising key word of the APP platform according to an embodiment of the present invention
The structure chart of automatically generating device, this structure specifically can include with lower module：
APP theme distribution computing module 401, is suitable to the description information for each APP, calculates each APP
Theme distribution, specifically include：
Latent Dirichletal location topic model computing module 4011, is suitable to the description letter for each APP
Breath, according to latent Dirichletal location topic model, calculates the theme distribution of each APP.
Search word theme distribution computing module 402, be suitable to according to search word in search history record with each
The click relation of APP, calculates the theme distribution of each search word；
Volumes of searches judge module 403, is suitable to judge the volumes of searches whether great-than search amount threshold of described search word
Value；If the volumes of searches great-than search amount threshold value of described search word, enter theme similarity word and carry
Delivery block 404.
Theme similarity word extraction module 404, is suitable to, for each search word, calculate described search word
Theme distribution and the theme distribution of advertisement APP between Topic Similarity；
Key word adds module 405, if being suitable to described Topic Similarity to be more than theme threshold value, by institute
State search word as the key word of described advertisement APP；
Keyed advertising key word builds module 406, is suitable to corresponding with advertisement APP according to described key word
Relation, builds the keyed advertising key word of corresponding advertisement APP.
Preferably, before keyed advertising key word builds module, also include：
Extracting directly module, is suitable to, directly from the title of advertisement APP and/or label, extract corresponding wide
Accuse the key word of APP.
Preferably, before keyed advertising key word builds module, also include：
Text search word acquisition module, is suitable to for each search word in search Download History, calculating is searched
Text similarity between the title of rope word and advertisement APP；
Second key word adds module, if being suitable to described text similarity to be more than text similar threshold value,
Then obtain described search word as key word.
Preferably, before keyed advertising key word builds module, also include：
Independent access search word extraction module, is suitable to in the search Download History of search history record
Each search word, judge that whether the independent access download time of described search word is more than independent access threshold
It is worth, and whether the classification of described search word belongs to same classification with the classification of advertisement APP；
3rd key word adds module, if the independent access download time being suitable to described search word is more than
Independent access threshold value, and the classification of the classification of described search word and advertisement APP belongs to same class
Mesh, then using described search word as key word.
Preferably, before keyed advertising key word builds module, also include：
APP classification segments module, is suitable to the APP now for each one-level class, using one-level class now
Each APP description information, each APP is divided into by corresponding one-level class two grades of classes now using grader
Mesh；
Search word sort module, is suitable to each search word, according to search word in search history record with
The click relation of each APP, two grades of classifications belonging to each APP described in one-level, calculate described search word institute
Corresponding two grades of classifications；
Class heading search word extraction module, is suitable to two grades of classifications being located according to advertisement APP, obtains and correspond to
Each search word of this two grades of classifications is then as key word.
Provided herein algorithm and display not with any certain computer, virtual system or miscellaneous equipment
Inherently related.Various general-purpose systems can also not play use with based on teaching one present invention in this yet.
As described above, the structure constructing required by this kind of system is obvious.Additionally, being directed to
Any certain programmed language.It is understood that, it is possible to use various programming languages realize described here
The content of invention, and the description above language-specific done is the optimal reality in order to disclose the present invention
Apply mode.
In description mentioned herein, illustrate a large amount of details.It is to be appreciated, however, that
Embodiments of the invention can be put into practice in the case of not having these details.In some instances,
Known method, structure and technology are not been shown in detail, so as not to obscure the understanding of this description.
Similarly it will be appreciated that in order to simplify the disclosure and help understand in each inventive aspect
Individual or multiple, in the description to the exemplary embodiment of the present invention above, each feature of the present invention
Sometimes it is grouped together in single embodiment, figure or descriptions thereof.However, should be by
The method of the disclosure is construed to reflect following intention：I.e. the present invention for required protection requires ratio at each
The more feature of feature being expressly recited in claim.More precisely, as following right will
As asking book to be reflected, inventive aspect is all spies less than single embodiment disclosed above
Levy.Therefore, it then follows claims of specific embodiment are thus expressly incorporated in this specific embodiment party
Formula, wherein each claim itself is as the separate embodiments of the present invention.
Those skilled in the art are appreciated that and the module in the equipment in embodiment can be carried out
Adaptively change and they are arranged in one or more equipment different from this embodiment.
Module in embodiment or unit or assembly can be combined into a module or unit or assembly, and
In addition multiple submodule or subelement or sub-component can be divided into.Except such feature and/or
Outside at least some of process or unit exclude each other, can be using any combinations to this explanation
All features disclosed in book (including adjoint claim, summary and accompanying drawing) and so disclosed
All processes of any method or equipment or unit are combined.Unless expressly stated otherwise, this theory
Each feature disclosed in bright book (including adjoint claim, summary and accompanying drawing) can be by offer phase
Same, equivalent or similar purpose alternative features are replacing.
Although additionally, it will be appreciated by those of skill in the art that some embodiments described herein include
Included some features rather than further feature in other embodiments, but the feature of different embodiment
Combination mean to be within the scope of the present invention and formed different embodiments.For example, under
In the claims in face, embodiment required for protection one of arbitrarily can be in any combination
Mode is using.
The all parts embodiment of the present invention can be realized with hardware, or with one or more
The software module run on reason device is realized, or is realized with combinations thereof.Those skilled in the art
It should be appreciated that can be realized using microprocessor or digital signal processor (DSP) in practice
According to embodiments of the present inventionThe keyed advertising key word of APP platform automatically generatesIn equipment some or
The some or all functions of the whole part of person.The present invention is also implemented as being retouched here for execution
Some or all equipment of the method stated or program of device (for example, computer program and calculating
Machine program product).Such program realizing the present invention can store on a computer-readable medium, or
Person can have the form of one or more signal.Such signal can be upper and lower from internet website
Load obtains, or provides on carrier signal, or is provided with any other form.
It should be noted that above-described embodiment the present invention will be described rather than the present invention is limited
Make, and those skilled in the art can design without departing from the scope of the appended claims
Alternative embodiment.In the claims, any reference markss between bracket should not be configured to
Limitations on claims.Word "comprising" do not exclude the presence of element not listed in the claims or
Step.Word "a" or "an" before element does not exclude the presence of multiple such units
Part.The present invention can be by means of the hardware including some different elements and by means of properly programmed
Computer is realizing.If in the unit claim listing equipment for drying, some in these devices
Individual can be to be embodied by same hardware branch.Word first, second, and third
Using not indicating that any order.These words can be construed to title.
The invention discloses a kind of keyed advertising key word automatic generation method of APP platform, including：
For the description information of each APP, calculate the theme distribution of each APP；
According to the click relation of search word in search history record and each APP, calculate the master of each search word
Topic distribution；
For each search word, calculate the theme distribution of described search word and the theme distribution of advertisement APP
Between Topic Similarity；
If described Topic Similarity is more than theme threshold value, using described search word as described advertisement
The key word of APP；
According to the corresponding relation of described key word and advertisement APP, build the advertisement rope of corresponding advertisement APP
Draw key word.
A2, the method according to A1 are it is characterised in that for each search word, calculating is described
Also include before Topic Similarity between the theme distribution of the theme distribution of search word and advertisement APP：
Judge the volumes of searches whether great-than search amount threshold value of described search word；
If the volumes of searches great-than search amount threshold value of described search word, enter and be directed to each search word,
Calculate the Topic Similarity between the theme distribution of described search word and the theme distribution of advertisement APP.
A3, the method according to A1 it is characterised in that the described description information for each APP,
Calculate the theme distribution of each APP, including：
For the description information of each APP, according to latent Dirichletal location topic model, calculate each APP
Theme distribution.
A4, the method according to A1 are it is characterised in that according to described key word and advertisement APP
Corresponding relation, before building the keyed advertising key word of corresponding advertisement APP, also include：
Directly from the title of advertisement APP and/or label, extract the key word of corresponding advertisement APP.
A5, the method according to A1 are it is characterised in that according to described key word and advertisement APP
Corresponding relation, before building the keyed advertising key word of corresponding advertisement APP, also include：
For each search word in search Download History, calculate between search word and the title of advertisement APP
Text similarity；
If described text similarity is more than text similar threshold value, obtain described search word as key
Word.
A6, the method according to A1 are it is characterised in that according to described key word and advertisement APP
Corresponding relation, before building the keyed advertising key word of corresponding advertisement APP, also include：
Search history record is searched for each search word in Download History, judge described search word
Whether independent access download time is more than independent access threshold value, and the classification of described search word and advertisement
Whether the classification of APP belongs to same classification；
If the independent access download time of described search word is more than independent access threshold value, and described search
The classification of rope word belongs to same classification, then using described search word as key with the classification of advertisement APP
Word.
A7, the method according to A1 are it is characterised in that according to described key word and advertisement APP
Corresponding relation, before building the keyed advertising key word of corresponding advertisement APP, also include：
For each one-level class APP now, using the description information of one-level class each APP now, adopt
With grader, each APP is divided into corresponding one-level class two grades of classifications now；
To each search word, according to the click relation of search word in search history record and each APP, one
Two grades of classifications belonging to the described each APP of level, calculate two grades of classifications corresponding to described search word；
Two grades of classifications being located according to advertisement APP, obtain to should each search word of two grades of classifications then make
For key word.
The invention discloses the keyed advertising key word automatically generating device of B8, a kind of APP platform, bag
Include：
APP theme distribution computing module, is suitable to the description information for each APP, calculates each APP's
Theme distribution；
Search word theme distribution computing module, is suitable to according to search word in search history record and each APP
Click relation, calculate each search word theme distribution；
Theme similarity word extraction module, is suitable to, for each search word, calculate described search word
Topic Similarity between the theme distribution of theme distribution and advertisement APP；
First key word adds module, if being suitable to described Topic Similarity to be more than theme threshold value, will
Described search word is as the key word of described advertisement APP；
Keyed advertising key word builds module, is suitable to according to described key word pass corresponding with advertisement APP
System, builds the keyed advertising key word of corresponding advertisement APP.
B9, the device according to B8 it is characterised in that theme similarity word extraction module it
Front also include：
Volumes of searches judge module, is suitable to judge the volumes of searches whether great-than search amount threshold of described search word
Value；If the volumes of searches great-than search amount threshold value of described search word, enter theme similarity word and carry
Delivery block.
B10, the device according to B8 are it is characterised in that described pin APP theme distribution calculates mould
Block includes：
Latent Dirichletal location topic model computing module, is suitable to the description information for each APP,
According to latent Dirichletal location topic model, calculate the theme distribution of each APP.
B11, the device according to B8 it is characterised in that keyed advertising key word build module it
Before, also include：
Extracting directly module, is suitable to, directly from the title of advertisement APP and/or label, extract corresponding wide
Accuse the key word of APP.
B12, the device according to B8 it is characterised in that keyed advertising key word build module it
Before, also include：
Text search word acquisition module, is suitable to for each search word in search Download History, calculating is searched
Text similarity between the title of rope word and advertisement APP；
Second key word adds module, if being suitable to described text similarity to be more than text similar threshold value,
Then obtain described search word as key word.
B13, the device according to B8 it is characterised in that keyed advertising key word build module it
Before, also include：
Independent access search word extraction module, is suitable to in the search Download History of search history record
Each search word, judge that whether the independent access download time of described search word is more than independent access threshold
It is worth, and whether the classification of described search word belongs to same classification with the classification of advertisement APP；
3rd key word adds module, if the independent access download time being suitable to described search word is more than
Independent access threshold value, and the classification of the classification of described search word and advertisement APP belongs to same class
Mesh, then using described search word as key word.
B14, the device according to B8 are it is characterised in that build module in keyed advertising key word
Before, also include：
APP classification segments module, is suitable to the APP now for each one-level class, using one-level class now
Each APP description information, each APP is divided into by corresponding one-level class two grades of classes now using grader
Mesh；
Search word sort module, is suitable to each search word, according to search word in search history record with
The click relation of each APP, two grades of classifications belonging to each APP described in one-level, calculate described search word institute
Corresponding two grades of classifications；
Class heading search word extraction module, is suitable to two grades of classifications being located according to advertisement APP, obtains and correspond to
Each search word of this two grades of classifications is then as key word.
Claims (10)
1. the keyed advertising key word automatic generation method of a kind of APP platform, including：
For the description information of each APP, calculate the theme distribution of each APP；
According to the click relation of search word in search history record and each APP, calculate the master of each search word
Topic distribution；
For each search word, calculate the theme distribution of described search word and the theme distribution of advertisement APP
Between Topic Similarity；
If described Topic Similarity is more than theme threshold value, using described search word as described advertisement
The key word of APP；
According to the corresponding relation of described key word and advertisement APP, build the advertisement rope of corresponding advertisement APP
Draw key word.
2. method according to claim 1 is it is characterised in that for each search word, count
Go back before calculating the Topic Similarity between the theme distribution of described search word and the theme distribution of advertisement APP
Including：
Judge the volumes of searches whether great-than search amount threshold value of described search word；
If the volumes of searches great-than search amount threshold value of described search word, enter and be directed to each search word,
Calculate the Topic Similarity between the theme distribution of described search word and the theme distribution of advertisement APP.
3. method according to claim 1 is it is characterised in that the described description for each APP
Information, calculates the theme distribution of each APP, including：
For the description information of each APP, according to latent Dirichletal location topic model, calculate each APP
Theme distribution.
4. method according to claim 1 it is characterised in that according to described key word with wide
Accuse the corresponding relation of APP, before building the keyed advertising key word of corresponding advertisement APP, also include：
Directly from the title of advertisement APP and/or label, extract the key word of corresponding advertisement APP.
5. method according to claim 1 it is characterised in that according to described key word with wide
Accuse the corresponding relation of APP, before building the keyed advertising key word of corresponding advertisement APP, also include：
For each search word in search Download History, calculate between search word and the title of advertisement APP
Text similarity；
If described text similarity is more than text similar threshold value, obtain described search word as key
Word.
6. method according to claim 1 it is characterised in that according to described key word with wide
Accuse the corresponding relation of APP, before building the keyed advertising key word of corresponding advertisement APP, also include：
Search history record is searched for each search word in Download History, judge described search word
Whether independent access download time is more than independent access threshold value, and the classification of described search word and advertisement
Whether the classification of APP belongs to same classification；
If the independent access download time of described search word is more than independent access threshold value, and described search
The classification of rope word belongs to same classification, then using described search word as key with the classification of advertisement APP
Word.
7. method according to claim 1 it is characterised in that according to described key word with wide
Accuse the corresponding relation of APP, before building the keyed advertising key word of corresponding advertisement APP, also include：
For each one-level class APP now, using the description information of one-level class each APP now, adopt
With grader, each APP is divided into corresponding one-level class two grades of classifications now；
To each search word, according to the click relation of search word in search history record and each APP, one
Two grades of classifications belonging to the described each APP of level, calculate two grades of classifications corresponding to described search word；
Two grades of classifications being located according to advertisement APP, obtain to should each search word of two grades of classifications then make
For key word.
8. the keyed advertising key word automatically generating device of a kind of APP platform, including：
APP theme distribution computing module, is suitable to the description information for each APP, calculates each APP's
Theme distribution；
Search word theme distribution computing module, is suitable to according to search word in search history record and each APP
Click relation, calculate each search word theme distribution；
Theme similarity word extraction module, is suitable to, for each search word, calculate described search word
Topic Similarity between the theme distribution of theme distribution and advertisement APP；
First key word adds module, if being suitable to described Topic Similarity to be more than theme threshold value, will
Described search word is as the key word of described advertisement APP；
Keyed advertising key word builds module, is suitable to according to described key word pass corresponding with advertisement APP
System, builds the keyed advertising key word of corresponding advertisement APP.
9. device according to claim 8 is it is characterised in that extract in theme similarity word
Also include before module：
Volumes of searches judge module, is suitable to judge the volumes of searches whether great-than search amount threshold of described search word
Value；If the volumes of searches great-than search amount threshold value of described search word, enter theme similarity word and carry
Delivery block.
10. device according to claim 8 is it is characterised in that described pin APP theme distribution meter
Calculate module to include：
Latent Dirichletal location topic model computing module, is suitable to the description information for each APP,
According to latent Dirichletal location topic model, calculate the theme distribution of each APP.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510488047.1A CN106445963B (en) | 2015-08-10 | 2015-08-10 | Advertisement index keyword automatic generation method and device of APP platform |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510488047.1A CN106445963B (en) | 2015-08-10 | 2015-08-10 | Advertisement index keyword automatic generation method and device of APP platform |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106445963A true CN106445963A (en) | 2017-02-22 |
| CN106445963B CN106445963B (en) | 2021-11-23 |
Family
ID=58093726
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510488047.1A Active CN106445963B (en) | 2015-08-10 | 2015-08-10 | Advertisement index keyword automatic generation method and device of APP platform |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106445963B (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107861753A (en) * | 2017-06-26 | 2018-03-30 | 平安普惠企业管理有限公司 | APP generations index, search method and system and readable storage medium storing program for executing |
| CN108304453A (en) * | 2017-12-18 | 2018-07-20 | 北京奇艺世纪科技有限公司 | A kind of determination method and device of video relevant search word |
| CN108304457A (en) * | 2017-12-22 | 2018-07-20 | 努比亚技术有限公司 | A kind of application mask method, server and computer readable storage medium |
| WO2018161710A1 (en) * | 2017-03-07 | 2018-09-13 | 广州优视网络科技有限公司 | Method, device, and storage medium for recommending search term on basis of input search term |
| CN110083774A (en) * | 2019-05-10 | 2019-08-02 | 腾讯科技（深圳）有限公司 | Using determination method, apparatus, computer equipment and the storage medium of recommendation list |
| CN110597987A (en) * | 2019-08-21 | 2019-12-20 | 微梦创科网络科技（中国）有限公司 | Search recommendation method and device |
| CN111475729A (en) * | 2020-04-07 | 2020-07-31 | 腾讯科技（深圳）有限公司 | Search content recommendation method and device |
| CN111858684A (en) * | 2020-07-03 | 2020-10-30 | 京华信息科技股份有限公司 | Method and system for improving retrieval accuracy based on user checking and recording |
| CN112132601A (en) * | 2019-06-25 | 2020-12-25 | 百度在线网络技术（北京）有限公司 | Advertisement title rewriting method, device and storage medium |
| CN112765494A (en) * | 2017-06-20 | 2021-05-07 | 创新先进技术有限公司 | Search method and search device |
| CN112765494B (en) * | 2017-06-20 | 2024-11-19 | 创新先进技术有限公司 | Searching method and searching device |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102708115A (en) * | 2004-08-09 | 2012-10-03 | 亚马逊技术股份有限公司 | Method and system for identifying keywords for use in placing keyword-targeted advertisements |
| US20140289227A1 (en) * | 2010-02-24 | 2014-09-25 | A9.Com, Inc. | Fixed phrase detection for search |
| CN104133877A (en) * | 2014-07-25 | 2014-11-05 | 百度在线网络技术（北京）有限公司 | Software label generation method and device |
| CN104424279A (en) * | 2013-08-30 | 2015-03-18 | 腾讯科技（深圳）有限公司 | Text relevance calculating method and device |
-
2015
- 2015-08-10 CN CN201510488047.1A patent/CN106445963B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102708115A (en) * | 2004-08-09 | 2012-10-03 | 亚马逊技术股份有限公司 | Method and system for identifying keywords for use in placing keyword-targeted advertisements |
| US20140289227A1 (en) * | 2010-02-24 | 2014-09-25 | A9.Com, Inc. | Fixed phrase detection for search |
| CN104424279A (en) * | 2013-08-30 | 2015-03-18 | 腾讯科技（深圳）有限公司 | Text relevance calculating method and device |
| CN104133877A (en) * | 2014-07-25 | 2014-11-05 | 百度在线网络技术（北京）有限公司 | Software label generation method and device |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2018161710A1 (en) * | 2017-03-07 | 2018-09-13 | 广州优视网络科技有限公司 | Method, device, and storage medium for recommending search term on basis of input search term |
| CN112765494A (en) * | 2017-06-20 | 2021-05-07 | 创新先进技术有限公司 | Search method and search device |
| CN112765494B (en) * | 2017-06-20 | 2024-11-19 | 创新先进技术有限公司 | Searching method and searching device |
| CN107861753A (en) * | 2017-06-26 | 2018-03-30 | 平安普惠企业管理有限公司 | APP generations index, search method and system and readable storage medium storing program for executing |
| CN108304453A (en) * | 2017-12-18 | 2018-07-20 | 北京奇艺世纪科技有限公司 | A kind of determination method and device of video relevant search word |
| CN108304457A (en) * | 2017-12-22 | 2018-07-20 | 努比亚技术有限公司 | A kind of application mask method, server and computer readable storage medium |
| CN110083774A (en) * | 2019-05-10 | 2019-08-02 | 腾讯科技（深圳）有限公司 | Using determination method, apparatus, computer equipment and the storage medium of recommendation list |
| CN110083774B (en) * | 2019-05-10 | 2023-11-03 | 腾讯科技（深圳）有限公司 | Method and device for determining application recommendation list, computer equipment and storage medium |
| CN112132601A (en) * | 2019-06-25 | 2020-12-25 | 百度在线网络技术（北京）有限公司 | Advertisement title rewriting method, device and storage medium |
| CN112132601B (en) * | 2019-06-25 | 2023-07-25 | 百度在线网络技术（北京）有限公司 | Advertisement title rewriting method, apparatus and storage medium |
| CN110597987A (en) * | 2019-08-21 | 2019-12-20 | 微梦创科网络科技（中国）有限公司 | Search recommendation method and device |
| CN111475729A (en) * | 2020-04-07 | 2020-07-31 | 腾讯科技（深圳）有限公司 | Search content recommendation method and device |
| CN111475729B (en) * | 2020-04-07 | 2023-07-25 | 腾讯科技（深圳）有限公司 | Search content recommendation method and device |
| CN111858684A (en) * | 2020-07-03 | 2020-10-30 | 京华信息科技股份有限公司 | Method and system for improving retrieval accuracy based on user checking and recording |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106445963B (en) | 2021-11-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106445963A (en) | Advertisement index keyword automatic generation method and apparatus for APP platform | |
| CN105808685B (en) | Promotion information pushing method and device | |
| CN106919575B (en) | Application program searching method and device | |
| CN108984675B (en) | Data query method and apparatus based on evaluation | |
| CN106970991B (en) | Similar application identification method and device, application search recommendation method and server | |
| CN109492180A (en) | Resource recommendation method, device, computer equipment and computer readable storage medium | |
| CN108319376B (en) | Input association recommendation method and device for optimizing commercial word promotion | |
| CN111797239B (en) | Application program classification method and device and terminal equipment | |
| CN105718184A (en) | Data processing method and apparatus | |
| CN104951468A (en) | Data searching and processing method and system | |
| CN109885773A (en) | A kind of article personalized recommendation method, system, medium and equipment | |
| CN113570413B (en) | Advertisement keyword generation method and device, storage medium and electronic equipment | |
| CN110968789B (en) | Electronic book pushing method, electronic equipment and computer storage medium | |
| CN108229999B (en) | Method and device for evaluating competitive products | |
| CN106919588A (en) | A kind of application program search system and method | |
| CN110597987A (en) | Search recommendation method and device | |
| CN103902579A (en) | Method and device for acquiring information | |
| CN115659008B (en) | Information pushing system, method, electronic equipment and medium for big data information feedback | |
| CN107977445A (en) | Application program recommends method and device | |
| CN106484698A (en) | A kind of method for pushing of search keyword and device | |
| CN106919576A (en) | Using the method and device of two grades of classes keywords database search for application now | |
| CN109241455B (en) | Recommended object display method and device | |
| CN108446333B (en) | Big data text mining processing system and method thereof | |
| CN112818200A (en) | Data crawling and event analyzing method and system based on static website | |
| CN108664501B (en) | Advertisement auditing method and device and server |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20240115 Address after: 100088 room 112, block D, 28 new street, new street, Xicheng District, Beijing (Desheng Park) Patentee after: BEIJING QIHOO TECHNOLOGY Co.,Ltd. Address before: 100088 room 112, block D, 28 new street, new street, Xicheng District, Beijing (Desheng Park) Patentee before: BEIJING QIHOO TECHNOLOGY Co.,Ltd. Patentee before: Qizhi software (Beijing) Co.,Ltd. |
|
| TR01 | Transfer of patent right |