CN105426514B - Personalized mobile application APP recommended method - Google Patents
Personalized mobile application APP recommended method Download PDFInfo
- Publication number
- CN105426514B CN105426514B CN201510867018.6A CN201510867018A CN105426514B CN 105426514 B CN105426514 B CN 105426514B CN 201510867018 A CN201510867018 A CN 201510867018A CN 105426514 B CN105426514 B CN 105426514B
- Authority
- CN
- China
- Prior art keywords
- app
- user
- information
- model
- mobile application
- 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.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/951—Indexing; Web crawling techniques
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention relates to personalized mobile application APP recommended methods.The present invention obtains the information of user and APP from application market, and this is pre-processed and is come, utilize emotion-aspect-regional model, input pretreated document, user is respectively obtained for emotion-potential preference in aspect-area of APP, prediction user selects the probability value of some APP, APP index file and user index file are converted to after processing, utilize association's related subject model, obtain the recommender score matrix of User-App, the probability value that above-mentioned SAR model obtains linearly is combined with the recommender score that CTM model obtains, weight is distributed, reaches final recommendation.The present invention overcomes defects existing for the conventional recommendation method for only considering individual element.The present invention comprehensively considers aspect, emotion, type and area in comment to find the potential preference of user, more meet the actual demand of user, user is explored to the preference of each attribute of APP, user demand and APP feature is best understood from, overcomes cold start-up problem.
Description
Technical field
The invention proposes a kind of mobile application recommended methods, in particular to personalized mobile application APP recommended method.
Background technique
The development of cell phone application gives user and more facilitates, and facilitates the life of user.But APP's is countless
Some problems also are caused to user with various in style.Research find to find only by browsing and simple inquiry it is useful and
The APP of user preference becomes extremely difficult, and excessive information means poor information to a certain extent, therefore just needs certain
Tool is quickly found required for user and the information of preference carrys out aid decision, prevents user from losing.Then, occur being permitted
More APP recommended methods.
Before the present invention makes, traditional recommended method such as collaborative filtering (CF) are dedicated to by user group
The user for there are similar interests with designated user is found, evaluation of these the comprehensive similar users to a certain information forms system to this
Designated user recommends APP to the fancy grade prediction of this information.However, as personalized business application is extended to user's life
The every aspect of information flow living, personalized recommendation technology are similar to early as collaborative filtering also in continuous development of making rapid progress
Phase technology is no longer satisfied the requirement under new environment, such as in the case where user and commodity more see and increase, the performance of system
Can be lower and lower also or when evaluation of the user to commodity is very sparse, in this way between the obtained user of the evaluation based on user
Similitude may inaccuracy even result in commodity and be not recommended.In addition to this, current most of technologies only consider individual element,
However while user has higher demand to commodity, for APP recommendation, in addition to function, it is also contemplated that its APP belongs to
The influence of property, type, geographical location and user feeling, such as in terms of geographical location, such as Meituan, public comment, Google
Figure in this way based on the software in geographical location, area size involved in them influence whether the recommendation this APP.
Summary of the invention
The purpose of the present invention is overcoming drawbacks described above, a kind of APP recommended method of personalization is developed.
The technical scheme is that
Personalized mobile application APP recommended method, it is characterised in that steps are as follows:
Step 1) data collection: user and mobile application, that is, APP information, including function description are obtained from application market
And comment information；
Step 2) pre-processes the original APP data of acquisition to avoid the occurrence of cold start-up problem；
Step 3) is SAR model using emotion-aspect-area, using the comment information of APP as input document, respectively
To user for the territory of use of the emotion of APP, the aspect of APP and APP, thus obtains user and dive to APP different attribute
In preference and predict that user selects the probability value of APP；
Step 4) is for further processing to the data that step 2) obtains, and is respectively converted into APP index file and user index
File；
Step 5) using association related subject, that is, CTM model, input abovementioned steps 4) two parts of files, obtain User-App
Recommender score matrix；
Step 6) is linearly combined SAR model with the probability value that CTM model respectively obtains with recommender score, is then used
The online proposed algorithm of Top-N is ranked up, and the higher APP of prediction scoring is recommended corresponding user.
In the step 1), in application shop Google Play, user to he the grading of used APP be public
It opens visible, can see all APP that user commented on once the ID for obtaining user, from there through climbing data tool for institute
There is initial data to be retrieved.
The preprocessing process of initial data includes: in the step 2)
A) it removes and writes the user commented on less than 2 and there is no the APP of any comment after filtering user；
B) Tokken: removal punctuation mark, removal number；
C) it removes stop words: removing English stop words, including preposition, pronoun, article；
D) stemmed: to convert each word to its prototype, past tense is converted into prototype, and present progressive tense is converted into original
Type.
The calculation formula for recommending probability is calculated in the step 3):
Indicate the probability that user u likes APP t and grades to it, wherein t, s+, u, r, a, ctAPP is respectively represented,
Positive emotion, user, area, the aspect of APP and the type of APP.
Data processing in the step 4), is divided into following steps:
A) all APP obtained in step 2) are numbered, are followed successively by 0,1,2,3,4 ..., n, each number pair
Answer its APP information, the information that as SAR model is obtained by filtration in step 3)；
B) all users obtained in step 2) are numbered, are followed successively by 0,1,2,3,4 ..., n, each number
Its corresponding user information, the information that as SAR model is obtained by filtration in step 3)；
C) by collected data preparation at a user index input file, call format a: user's of behavior one
Information, the coding+1 of the first as user of row, second is classified as the APP quantity that user graded, and remainder is classified as what user rating was crossed
All APP numbers；
D) by collected data preparation at a APP index input file, call format: the letter of one APP of behavior
Breath, row is first be APP coding+1, second is classified as the number of users graded to this APP, and remainder is classified as all to be commented to APP
The Customs Assigned Number of grade；
E) above two parts of files are inputted into CTM model, obtains a User-App recommender score matrix, recommendation is canonical
Expression can be recommended, and being worth more bigger more worth recommendation conversely, recommendation is negative indicates that this APP is not recommend to be worth to the user
, matrix behavior APP is classified as user.
7. the more attribute recommended methods of APP according to claim 1, it is characterised in that obtain SAR model in step 6)
Probability p (t, the s arrived+| u) and the obtained recommendation r of GTM modelutLinear junction close calculation formula, it is as follows: set two parameters
α, β then merge recommender score Score are as follows:
Score=α p (t, s+|u)+βrut
Wherein, α, β are weighing input parameters.
Aspect, emotion, type and the area that advantages of the present invention and effect are to comprehensively consider in comment reach personalized
The purpose of recommendation.Mainly there are following some advantages:
1. this recommended method comprehensively considers aspect, emotion, type and the regional potential preference to find user in comment,
More meet the actual demand of user.
2. this recommended method can solve the insurmountable cold start-up problem of traditional collaborative filtering i.e. CF.
3. conventional recommendation method such as CF do not use APP content, it be based on the user that selected APP has parallel pattern and
Recommend, and this recommended method is simultaneously based on content and user rating, prediction of result is based on content or user rating
It grades depending on how many user to APP.
4. this recommended method is a kind of personalized recommendation, the application recommended can consider interest, their location of user etc..
5. this recommended method proposes the completely new classification method to APP attribute, by the Attribute transposition of APP obtain in more detail, by
This explores user to the preference of each attribute of APP, is best understood from user demand and APP feature.
Detailed description of the invention
Fig. 1 --- overall procedure schematic diagram of the invention.
Fig. 2 --- the functional circuit information schematic diagram of Google Maps of the present invention.
Fig. 3 --- comment information schematic diagram of the user Sarah of the present invention to Yellow Pages.
Fig. 4 --- comment information schematic diagram of the user Sarah of the present invention to Commander Compass Lite.
Fig. 5 --- the functional circuit information schematic diagram of the Google Maps after the present invention is pretreated.
Fig. 6 --- user Sarah after the present invention is pretreated is to the comment information schematic diagram of Yellow Pages.
Fig. 7 --- user Sarah after the present invention is pretreated is to the comment information of Commander Compass Lite
Schematic diagram.
Fig. 8 --- the preliminary output result schematic diagram of emotion-aspect-regional model of the present invention.
Fig. 9 --- the present invention assists the call format schematic diagram of the user index input file of correlation model.
Figure 10 --- the present invention assists the call format schematic diagram of the APP index input file of correlation model.
Figure 11 --- the output format schematic diagram of present invention association correlation model User-App recommender score matrix.
Specific embodiment
Technical thought of the invention is:
The present invention considers the influence for combining aspect, emotion, type and area to recommended technology, carries out to APP attribute
Relatively classification in detail, such as the ratio between interface, geographical location, function menu, discharging quantity and activation amount, setting, it is more detailed with this
Requirement and preference of the user to APP different attribute carefully are understood, to keep recommendation effect more preferable.Also utilize association's related subject model
This personalized recommendation, to guarantee that this recommended method can be used widely.
Present invention combination CTM model and SAR model model user comment information, find that user's is potential inclined with this
It gets well and is recommended in detail.
Technical solution of the present invention is described in detail with reference to the accompanying drawing:
Step 1)
Data collection: in application shop such as Google Play, user to he the grading of used APP be open
It is visible, all APP that user commented on can be seen once the ID for obtaining user.It is possible thereby to will by climbing data tool
All initial data such as user comment and function description are retrieved.It is exemplified below: assuming that user Sarah wants to look for a navigation
APP, user used Yellow Pages and Commander Compass Lite two to navigate APP before this, and Fig. 2 is
The functional circuit information of Google Maps, this APP are that user Sarah is original, and Fig. 3 is Sarah to Yellow Pages
User comment information, Fig. 4 is Sarah to the user comment information of Commander Compass Lite.
Step 2)
The original APP data of acquisition are pre-processed to avoid the occurrence of cold start-up problem.It writes firstly, removing less than 2
There is no the APP of any comment after the user and filtering user of comment.Secondly, carrying out Tokkenization i.e. tokenization, respectively
Removal punctuation mark such as "@", "? " and removal number such as " 1 ", " 23 ".Then stop words, that is, stopping, such as preposition are removed
" for ", " to "；Pronoun " it ", " he "；Article " a ", " an " " the ".Last stemmed i.e. stemming, each word is turned
Its prototype is turned to, for example " fixed " is past tense, is converted into prototype " fix "；" moving " is present progressive tense, is converted into
Prototype " move ".The algorithm selected when stemmed is Porter stemming algorithm.After the operation of three steps, generation is gone
Except the pretreated document of noise.Result is as shown in Fig. 5/6/7 after processing.
Step 3)
Following letter is respectively obtained using the description of the function of APP and user comment information as input document using SAR model
Breath:
A) the function keyword of Google Maps: nayigate, fast, easy, comprehensive, accurate,
Voice-guided, report, reroute, detailed and shortcoming: not-all-regions, battery-loss
Supplementary explanation: longer phrase need to be converted into compound word.
B) a) emotion, that is, sentiment vocabulary of the user for used Yellow Pages: best, love；
B) aspect is aspect vocabulary: navigation, price (free), view；C) territory of use, that is, region:not-all-
A) the sentiment vocabulary of regions and the user for Commander Compass Lite: well, great；b)
Aspect vocabulary: accurate, tech, price, display, that is, view.
As shown in figure 8, obtaining user from above to APP navigation (navigation), free (free) and view (view)
Etc. potential preference and geographical location influence, finally according to formula obtain user select APP prediction probability value p (t,
s+| it u) is 5.91267063568302.
Step 4)
The data that step 2) obtains are made further to arrange, are respectively converted into APP index file and user index file.It is first
First, all APP obtained are numbered, 0,1,2,3,4 ..., n are followed successively by, each numbers its corresponding APP information, i.e.,
For the information that SAR model is obtained by filtration in step 3), Customs Assigned Number is same as above.Arrange the number of Google Maps is 3, use
Family number is 6.Secondly, as shown in figure 9, by collected data preparation at a user index input file, a behavior one
The information of user, the coding+1 of the first as user of row, second is classified as the APP quantity that user graded, and remainder is classified as user and comments
All APP number that grade is crossed.In addition to this, as shown in Figure 10, collected data preparation is inputted into text at a APP index
Part, the information of one APP of behavior, the first as coding+1 of APP of row, second is classified as the number of users graded to this APP,
Remainder is classified as all Customs Assigned Numbers graded to APP.
Step 5)
As shown in figure 11, above two parts are arranged obtained file and inputs CTM model, obtain a User-App recommended hour
Matrix number, recommendation are that regular representation can be recommended, and be worth more bigger more worth recommendation indicates the APP to this conversely, recommendation is negative
User is not recommend value, and matrix behavior APP is classified as user, by scheme APP 3 is to the recommendation of user 6
5.14723419271134。
Step 6)
Probability value p (t, the s that SAR model and CTM model are respectively obtained+| u) with recommender score rUt is linearIn conjunction with as follows: setting
Fixed two weighing input parameters α, β, weight can voluntarily be distributed according to itself preference of user, for example, if user's more preference
α, β then can be assigned a value of respectively 60%, 40% by the recommendation results of SAR model, thus obtain merging of the Google Maps to user 6
Recommender score Score is 3.477934629648461, and in addition to this, there are also other navigation types APP to the recommender score of user 6,
It is ranked up again with the online proposed algorithm of Top-N, the higher APP of prediction scoring is recommended into user 6, achievees the purpose that recommendation.
It illustrates:
The online proposed algorithm of Top-N is a kind of common means that personalized information push is directly carried out to user, its energy
Time of calculating is enough effectively shortened, to improve computational efficiency.Assuming that there are an array, merging recommender score obtained above
For the element in array, what Top-N recommended to refer to online be exactly from this already existing array, find out it is maximum or the smallest before
N element.Since the purpose of this recommended method is to find the higher APP of recommender score to recommend to user, herein
Top-N is referred to finding maximum preceding n element, is implemented as follows:
A the preceding n element of array, the most rickle that creation length is n) are taken out.
B the surplus element of array) is started the cycle over from n, if the root node of element a ratio most rickle is big, a is arranged to minimum
The root node of heap, and allow heap keep most rickle characteristic.
C after the completion of) recycling, all elements in most rickle are exactly the maximum n element for needing to look for, as recommender score
Highest n APP, thus sequence is effectively recommended.
Claims (6)
1. personalized mobile application APP recommended method, it is characterised in that steps are as follows:
Step 1) data collection: user and mobile application, that is, APP information are obtained from application market, including function is described and commented
By information；
Step 2) pre-processes the original APP data of acquisition to avoid the occurrence of cold start-up problem；
Step 3) is SAR model using emotion-aspect-area, using the comment information of APP as input document, respectively obtains use
Thus family obtains user to the potential inclined of APP different attribute for the territory of use of the emotion of APP, the aspect of APP and APP
It gets well and predicts that user selects the probability value of APP；
Step 4) is for further processing to the data that step 3) obtains, and is respectively converted into APP index file and user index text
Part；
Step 5) using association related subject, that is, CTM model, input abovementioned steps 4) two parts of files, obtain pushing away for User-App
Recommend score matrix；
Step 6) is linearly combined SAR model with the probability value that CTM model respectively obtains with recommender score, is then existed with Top-N
Line proposed algorithm is ranked up, and the higher APP of prediction scoring is recommended corresponding user.
2. the mobile application APP recommended method of personalization according to claim 1, it is characterised in that in step 1), answering
With in the Google Play of shop, user to he used APP grading be disclose it is visible, once obtain the ID of user
It can see all APP that user commented on, be retrieved all initial data from there through data tool is climbed.
3. the mobile application APP recommended method of personalization according to claim 1, it is characterised in that original number in step 2)
According to preprocessing process include:
A) it removes and writes the user commented on less than 2 and there is no the APP of any comment after filtering user；
B) Tokken: removal punctuation mark, removal number；
C) it removes stop words: removing English stop words, including preposition, pronoun, article；
D) stemmed: to convert each word to its prototype, past tense is converted into prototype, and present progressive tense is converted into prototype.
4. the mobile application APP recommended method of personalization according to claim 1, it is characterised in that calculated in step 3)
Recommend the calculation formula of probability:
P (t, s+| u)=rp (r | u) p (ct| u) p (t | r, ct) ap (a | u, ct) p (s+| a, t) indicate that user u likes APP t
And to the probability that it is graded, wherein t, s+, u, r, a, ctRespectively represent APP, positive emotion, user, area, the aspect of APP
And the type of APP.
5. the mobile application APP recommended method of personalization according to claim 1, it is characterised in that the data in step 4)
Processing, is divided into following steps:
A) all APP obtained in step 2) are numbered, are followed successively by 0,1,2,3,4 ..., n, each number corresponds to
Its APP information, the information that as SAR model is obtained by filtration in step 3)；
B) all users obtained in step 2) are numbered, are followed successively by 0,1,2,3,4 ..., n, each number corresponds to
Its user information, the information that as SAR model is obtained by filtration in step 3)；
C) by collected data preparation at a user index input file, call format: the information of one user of behavior,
Row is first be user coding+1, second is classified as the APP quantity that user graded, and what remainder was classified as that user rating crosses owns
APP number；
D) by collected data preparation at a APP index input file, call format: the information of one APP of behavior, row
Head is the coding+1 of APP, and second is classified as the number of users graded to this APP, and remainder is classified as all use graded to APP
Family number；
E) above two parts of files are inputted into CTM model, obtains a User-App recommender score matrix, recommendation is regular representation
It can recommend, be worth more bigger more worth recommendation, conversely, recommendation is negative, indicate that this APP is not recommend value, square to the user
Battle array behavior APP, is classified as user.
6. the mobile application APP recommended method of personalization according to claim 1, it is characterised in that by SAR in step 6)
Probability p (t, the s that model obtains+| u) and the obtained recommendation r of CTM modelutLinear junction close calculation formula, it is as follows: set two
A parameter alpha, β then merge recommender score Score are as follows:
Score=α p (t, s+|u)+βrut
Wherein, α, β are weighing input parameters, t, s+, u respectively represents APP, positive emotion, user.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510867018.6A CN105426514B (en) | 2015-11-30 | 2015-11-30 | Personalized mobile application APP recommended method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510867018.6A CN105426514B (en) | 2015-11-30 | 2015-11-30 | Personalized mobile application APP recommended method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105426514A CN105426514A (en) | 2016-03-23 |
| CN105426514B true CN105426514B (en) | 2019-03-01 |
Family
ID=55504726
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510867018.6A Active CN105426514B (en) | 2015-11-30 | 2015-11-30 | Personalized mobile application APP recommended method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105426514B (en) |
Families Citing this family (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106095776B (en) * | 2016-05-25 | 2019-10-01 | 中山大学 | The method that a kind of couple of user carries out Topics Crawling and application is recommended |
| CN106250146B (en) * | 2016-07-28 | 2021-03-12 | 维沃移动通信有限公司 | Application demand processing method and device |
| CN106569687A (en) * | 2016-10-19 | 2017-04-19 | 北京三快在线科技有限公司 | Icon arrangement method and apparatus for virtual buttons, and terminal |
| TWI647638B (en) | 2016-11-15 | 2019-01-11 | 財團法人工業技術研究院 | Interactive recommendation system and method |
| CN106547922B (en) * | 2016-12-07 | 2020-08-25 | 阿里巴巴（中国）有限公司 | Application program sorting method and device and server |
| CN106846094A (en) * | 2016-12-29 | 2017-06-13 | 广州优视网络科技有限公司 | A kind of method and apparatus for recommending application message based on application has been installed |
| CN106844240A (en) * | 2017-01-25 | 2017-06-13 | 百度在线网络技术（北京）有限公司 | Information displaying method and device |
| CN106952129A (en) * | 2017-02-23 | 2017-07-14 | 广东小天才科技有限公司 | Method and device is recommended in a kind of application in application shop |
| CN107220303A (en) * | 2017-05-10 | 2017-09-29 | 努比亚技术有限公司 | Recommendation method, device and the computer-readable medium of a kind of application |
| CN107861753B (en) * | 2017-06-26 | 2020-12-11 | 平安普惠企业管理有限公司 | APP generation index, retrieval method and system and readable storage medium |
| CN107329770A (en) * | 2017-07-04 | 2017-11-07 | 扬州大学 | The personalized recommendation method repaired for software security BUG |
| CN108307039B (en) * | 2017-12-21 | 2021-02-02 | 维沃移动通信有限公司 | Application information display method and mobile terminal |
| CN108763258B (en) * | 2018-04-03 | 2023-01-10 | 平安科技（深圳）有限公司 | Document theme parameter extraction method, product recommendation method, device and storage medium |
| CN108563758B (en) * | 2018-04-17 | 2021-09-24 | 广州虎牙信息科技有限公司 | User quantity measuring and calculating method, device, equipment and storage medium |
| CN108710680A (en) * | 2018-05-18 | 2018-10-26 | 哈尔滨理工大学 | It is a kind of to carry out the recommendation method of the film based on sentiment analysis using deep learning |
| CN109189887A (en) * | 2018-09-07 | 2019-01-11 | 江苏瑞康安全装备有限公司 | A kind of micro-blog information recommended method of facing moving terminal |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20130059738A (en) * | 2011-11-29 | 2013-06-07 | 에스케이플래닛 주식회사 | System and method for recommending application using contents analysis |
| CN103559262A (en) * | 2013-11-04 | 2014-02-05 | 北京邮电大学 | Community-based author and academic paper recommending system and recommending method |
| CN104461558A (en) * | 2014-12-22 | 2015-03-25 | 扬州大学 | Software-fix-orientated individualized task recommendation method based on co-correlativity topic model |
-
2015
- 2015-11-30 CN CN201510867018.6A patent/CN105426514B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20130059738A (en) * | 2011-11-29 | 2013-06-07 | 에스케이플래닛 주식회사 | System and method for recommending application using contents analysis |
| CN103559262A (en) * | 2013-11-04 | 2014-02-05 | 北京邮电大学 | Community-based author and academic paper recommending system and recommending method |
| CN104461558A (en) * | 2014-12-22 | 2015-03-25 | 扬州大学 | Software-fix-orientated individualized task recommendation method based on co-correlativity topic model |
Non-Patent Citations (1)
| Title |
|---|
| SAR: A sentiment-aspect-region model for user preference analysis in geo-tagged reviews;Kaiqi Zhao 等;《2015 IEEE 31st International Conference on Data Engineering (ICDE)》;20150417;675-686 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105426514A (en) | 2016-03-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105426514B (en) | Personalized mobile application APP recommended method | |
| US11868357B2 (en) | Search result ranking and presentation | |
| US20240004934A1 (en) | Multi-modal virtual experiences of distributed content | |
| AU2010284506B2 (en) | Semantic trading floor | |
| US20180357211A1 (en) | Constructing a Narrative Based on a Collection of Images | |
| CN106227815B (en) | Multi-modal clue personalized application program function recommendation method and system | |
| CN103455487B (en) | The extracting method and device of a kind of search term | |
| CN106202294B (en) | Related news computing method and device based on keyword and topic model fusion | |
| JP2010039710A (en) | Information collection device, travel guiding device, travel guiding system and computer program | |
| CN106126605A (en) | A kind of short text classification method based on user's portrait | |
| CN110781300A (en) | Tourism resource culture characteristic scoring algorithm based on Baidu encyclopedia knowledge graph | |
| CN110321483A (en) | A kind of online course content of platform recommended method, device, system and storage medium based on user's sequence sexual behaviour | |
| Wen et al. | Visual background recommendation for dance performances using deep matrix factorization | |
| Vallet et al. | Effects of usage-based feedback on video retrieval: a simulation-based study | |
| CN102693284A (en) | Extraction method of information in personal address list | |
| CN113657116B (en) | Social media popularity prediction method and device based on visual semantic relationship | |
| Matsumoto et al. | Supporting human recollection of the impressive events using the number of photos | |
| Ribeiro et al. | MEMORIA: A Memory Enhancement and MOment RetrIeval Application for LSC 2023 | |
| JP6103766B2 (en) | Behavioral process extraction method and behavioral process extraction device | |
| KR102279125B1 (en) | Terminal and apparatus for providing recommendation information based on preference filter | |
| KR102497151B1 (en) | Applicant information filling system and method | |
| Wick et al. | Enhancing RSS feeds with extracted geospatial information for further processing and visualization | |
| JP2009129036A (en) | Information retrieval system, information retrieval method, and program | |
| Lou et al. | Recommendation of Tourist Attractions and Route Based on Social Network Data Mining | |
| TWI270793B (en) | Chinese information retrieval system and method based on document projection maps |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |