CN108038133A - Personalized recommendation method - Google Patents
Personalized recommendation method Download PDFInfo
- Publication number
- CN108038133A CN108038133A CN201711155639.7A CN201711155639A CN108038133A CN 108038133 A CN108038133 A CN 108038133A CN 201711155639 A CN201711155639 A CN 201711155639A CN 108038133 A CN108038133 A CN 108038133A
- Authority
- CN
- China
- Prior art keywords
- user
- data
- keyword
- cultural
- activity
- 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.)
- Pending
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
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
-
- 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
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)
- Life Sciences & Earth Sciences (AREA)
- Animal Behavior & Ethology (AREA)
- Computational Linguistics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
A kind of personalized recommendation method, user behavior data is collected first, by the active theme data that the user's behavioral data is divided into the preference pattern data of user and user currently browses, then applicating cooperation filter algorithm handles the preference pattern data, recommends to cultural activity of the user based on collaborative filtering；At the same time, the active theme data currently browsed according to user, carry out the cultural data based on body and recommend.It is an advantage of the invention that：Ontologies technology is dissolved into the retrieval of cultural data, overcomes the limitation that cannot be handled in conventional IR mode semantic relation.By analyzing the semantic information of phrase included in search terms, simple reasoning query expansion is carried out by ontologies, information resources are comprehensively accurately fed back into user.
Description
Technical field
The present invention relates to a kind of personalized recommendation method, is mainly used in Public Culture Service field, Public Culture
Service organization would generally hold some cultural activities, such as theatrical performances, public lecture etc..User is in Public Culture Service website
When browsing cultural activity, website is some cultural activities of the recommendation of user individual, is user when browsing cultural activity details
Recommend some and the relevant cultural data of the active theme.
Background technology
Nowadays, to meet the individual demand of people, personalized recommendation technology be widely applied to e-commerce,
The fields such as social networks, various Video Music websites, such as Amazon, Taobao, bean cotyledon etc., all employ to varying degrees
Personalized recommendation system.Since personalized recommendation technology can be good at meeting users ' individualized requirement, user can be excavated and dived
In interest, therefore, personalized ventilation system is integrated into Public Culture Service platform and is particularly important.Individual character proposed in this paper
Change recommendation service and include two parts：First, for the recommendation of cultural activity, second, in the cultural activity browsed for active user
The recommendation for the cultural data held and carried out.
Outstanding characteristics are focused in recommendation for cultural activity, emphasize user interest preference.Thus platform utilizes user's
Historical viewings behavioral data, gives a mark the corresponding navigation patterns of user, for example user such as thumbs up, collects, comment at the behavior, forms use
Family-activity score matrix, then carries out cultural activity recommendation using collaborative filtering recommending technology to user.
Prominent cultural activity topic relativity is focused in recommendation for cultural data, and it is expected to accomplish is to recommend to live with current
Dynamic theme has semantic relevant cultural data.A kind of conventional way is directly to be closed current culture active title as retrieval
Key word, is gone in cultural data bank to retrieve data of literatures using traditional information retrieval, use is presented to as recommendation results
Family.But there are the following problems for this way：
1. recommendation results are to meet the requirement of user on literal, what the content that actual retrieval goes out may deviate user browses intention
Demand, because this method usually treats the keyword word isolated as one, does not carry out concept extension, can not excavate letter
Inner link between breath, only includes the content containing keyword in the retrieval result of return；So its retrieval result is only anti-
A side of content has been reflected, has not accounted for the semantic information between keyword, can not obtain has semantic congruence with keyword
The retrieval result of other keywords of property, causes the validity of retrieval result to decline comprehensively.
2. lacking the understanding to user view, the keyword A obtained by natural language participle technique is often possible to not
It is that user wants, and is only what user really wanted to know about with the relevant B of keyword A；Simultaneously as whole plateform system
Application scenarios are based on Public Culture Service field, so the information that retrieval is fed back will have field correlation, traditional inspection
Often there are substantial amounts of invalid information, retrieval result in the feedback result of rope mode can not meet user demand well.
The content of the invention
It is an object of the invention to provide a kind of personalized recommendation method, to solve the above problem existing in the prior art.
The technical scheme is that：A kind of personalized recommendation method, it is characterised in that collect user behavior number first
According to by the active theme data that the user's behavioral data is divided into the preference pattern data of user and user currently browses, Ran Houying
The preference pattern data are handled with collaborative filtering, are recommended to cultural activity of the user based on collaborative filtering；
At the same time, the active theme data currently browsed according to user, carry out the cultural data based on body and recommend.
The flow that cultural activity based on collaborative filtering is recommended includes：
（1）User enters in system platform, and system platform can collect the behavior historical data of user, such as participates in cultural activity note
Record, activity thumb up record, activity collection record, these data records indicate the hobby of user；
（2）System generates user-activity score matrix according to pre-set scoring；
（3）Start proposed algorithm flow, the similarity of user is calculated using formula, finds out the highest N number of user of similarity the most
Nearest-neighbors collection, score in predicting is carried out using nearest-neighbors collection, by prediction marking and queuing, obtains recommendation results, result is presented
To user.
The activity that the preference pattern data include user thumbs up data, the data of Collecting, the note of activity
Record data.
The cultural data recommends to comprise the following steps：
（1）Active theme keyword obtains：The active title currently browsed to user carries out word segmentation processing, the theme of extraction activity
Keyword；There can be the insignificant words such as " ", " " and punctuation mark toward contact in word segmentation result, also need to use for this
Stop words is carried out based on the mode for disabling vocabulary to operate；
（2）Query expansion based on body：By above-mentioned processing, the keyword phrase of active theme, this keyword have been obtained
Phrase, which may include noun, may also include adjective, be exactly to put it into ontology library in next step, carry out based on body
Query expansion processing；
（3）Lucene resource retrieval modules：Using by the use of being obtained after Ontology Query extension process as keyword, using based on Java
Full text information retrieval instrument Lucene carry out the retrieval of resource.
The cultural data recommends the active theme keyword acquisition methods to be：
（1）Preprocessing process：Word segmentation processing is carried out using natural language processing technique, a word has been obtained by word segmentation processing
Group, carries out screening and filtering to keyword according to part of speech, retains noun, verb, adjective；
（2）Keyword expansion：At present in information retrieval system, largely it is all based on what keyword was inquired about, obtains first
The keyword that family needs to inquire about is taken, query expansion is then carried out automatically, the result after extension is retrieved, extended mode leads to
It is extended frequently with synonym development method, that is, using synonymicon, near synonym dictionary；
（3）Query expansion：Processing is optimized to it using body processing correlation technique, in order to expand Ontological concept related content
Exhibition, is handled using the hierarchical relationship in body, set membership node is extended query concept, at it
Managing flow is：
A. to each keyword of active theme of acquisition, it is concept, example or the attribute in body to judge it；
If being b. concept or example, the main set membership utilized in body construction is extended；If there are attribute,
Using the relation between attribute and concept, the semantic relation according to present in body and inference mechanism, excavate implicit information；
C. spreading result is preserved, subsequently to carry out the retrieval of cultural resource information.
It is an advantage of the invention that：Ontologies technology is dissolved into the retrieval of cultural data, in this way, tradition can be overcome
The limitation that cannot be handled in information retrieval mode semantic relation, may determine that because Ontology net contains machine
Concept definition, while Ontology net establishes semantic relation between each concept, so that system is to knowledge in field
Concept and concept between contact have unified understanding.System is believed by analyzing the semantic of phrase included in search terms
Breath, carries out simple reasoning query expansion by ontologies, information resources comprehensively accurately is fed back to user.
Brief description of the drawings
Fig. 1 is the general frame and flow chart of the embodiment of the present invention；
Fig. 2 is the flow chart that method is recommended in cultural activity of the present invention based on collaborative filtering；
Fig. 3 is that cultural data of the present invention based on body recommends method flow diagram；
Fig. 4 is the Lucene system architecture diagrams of institute's foundation of the present invention.
Embodiment
Referring to Fig. 1-Fig. 4, personalized ventilation system method of the present invention mainly includes two parts：The recommendation of personalized activity
And recommend with the relevant cultural data of user's current browsing behavior.Wherein, the recommendation of personalized activity uses collaborative filtering skill
Art realizes, first by collecting the behavioral data of user, for example the activity of user thumbs up data, the data of Collecting, participates in
Activity data record etc., forms the preference pattern of user, that is, user-activity score matrix, and then applicating cooperation filtering is calculated
Method carries out activity recommendation to user.
When user browses cultural activity details, system can be carried out according to the active theme that user currently browses using body
Semantic extension, query and search goes out recommends user with the relevant cultural data of current browsing behavior.
The Yii frames based on MVC that the realization of Public Culture Service platform uses, are realized with PHP language developments.And for
The realization of personalized service, due to being related to body processing, it is necessary to provide the query expansion of body using Jena API, thus it is right
Realized in the realization of personalized ventilation system using Java framework exploitation.Personalized ventilation system program is encapsulated after the completion of exploitation
Run into WebService system services, Public Culture Service platform is called by web Service interface.
Collaborative filtering recommending technology is an important component in modern personalized recommendation system.Collaborative filtering recommending
Recommended project mainly is produced for current active user using the viewpoint of colony, is recorded by past user behavior, is calculated
The similarity of interest preference between each user, finds out the neighbor user close with current user interest preference, and pass through these
The similar users group of neighbor user composition is suggested to produce the recommendation to active user.In the recommendation of cultural activity, collaboration
Recommend be that active user finds the user group for having similar interests background or similar cultural demand, pass through similar users
Colony recommends to obtain related culture activity.
Personalized recommendation system based on collaborative filtering is generally made of three modules：User modeling module, recommended
Modeling module, proposed algorithm module.Wherein, user modeling module namely collects the process of user preference information, Data Collection
It is the basis of personalized recommendation system, Data Collection is broadly divided into three modes：It is explicit to obtain, implicit obtain and heuristic obtain
Take.
In the design of this paper, the collection of user preference information mostlys come from the behavioral data of user, such as to culture
Thumbing up, collecting for activity, participates in certain class cultural activity etc..Then according to the marking mechanism set in advance, such as user's collection
One activity, then it is 2 etc. to set corresponding scoring, forms user-project rating matrix, as shown in table 1.
1 user of table-project rating matrix table
Next the cultural activity recommended flowsheet based on collaborative filtering is provided herein, as shown in Figure 2：
Wherein, specific recommendation process step is as follows：
1. user enters in system platform, system platform can collect the behavior historical data of user, for example participate in cultural activity note
Record, activity thumb up record, activity collection record, these data records indicate the hobby of user；
2. system generates user-activity score matrix according to pre-set scoring；
3. starting proposed algorithm flow, the similarity of user is calculated using formula, finds out the highest N number of user of similarity as most
Neighbour occupies collection, and score in predicting is carried out using nearest-neighbors collection, by prediction marking and queuing, obtains recommendation results, result is presented to
User.
Cultural data based on body is recommended：
It is real using information retrieval correlation technique, advantage of the fusion body in terms of semantic understanding herein from the angle of information retrieval
The recommendation of existing culture data.Cultural data recommends the important component as personalized ventilation system, this trifle will provide base
In the critical workflow that the cultural data of body is recommended, and to the key technology wherein used and realize that step provides specifically
It is bright.
First, cultural data recommending module flow
The cultural activity that the cultural data designed herein is currently browsed primarily directed to user come the recommendation of cultural data that carries out,
The module is related to natural language processing（NLP）, the correlation technique such as Ontology query expansion and information retrieval.Provide now
Cultural data recommended flowsheet based on body, and process, which is described in detail introduction, to be realized to processing therein.Based on body
Cultural data recommended flowsheet is as shown in Figure 3.
By upper figure, the cultural data based on body recommends to be generally divided into three parts：Active theme keyword
Acquisition, the query expansion based on body, resource retrieval module.
1. active theme keyword obtains.The module is mainly that the active title currently browsed to user is carried out at participle
Reason, the subject key words of extraction activity.It is insignificant toward contact there are " ", " " and punctuation mark etc. in word segmentation result
Word, also needs to use for this and is carried out stop words based on the mode for disabling vocabulary and operated.
2. the query expansion based on body.By above-mentioned processing, the keyword phrase of active theme is obtained, this is crucial
Word phrase, which may include noun, may also include adjective, be exactly to put it into ontology library in next step, carry out being based on body
Query expansion processing.
3.Lucene resource retrieval modules.Resource retrieval module is the final step that cultural data is recommended, that is, will be used
What is obtained after Ontology Query extension process is used as keyword, carries out full text resource retrieval.There is employed herein the full text based on Java
Information retrieval tool Lucene carries out the retrieval of resource.
Next the realization explanation of each process module will be given herein.
2nd, active theme keyword obtains
In the cultural data based on body is recommended, it is necessary first to a preprocessing process, that is, need to utilize natural language
Treatment technology carries out word segmentation processing.A phrase has been obtained by word segmentation processing, but has been not that each word is keyword,
Wherein usually contain many modal particles, auxiliary word etc., and such word and do not have semantic information, therefore keyword can not be used as.
Meanwhile need to be extended keyword in next step after participle, if all carrying out keyword expansion to all words,
Efficiency will necessarily be influenced, lifts the complexity of processing, especially if active title is long, if carrying out complete keyword expansion
Influence time is long, influences user experience.
Therefore, it is necessary to carry out keyword filtering in some way.In general, typically handled according to part of speech,
The classification of part of speech has very much, specific as shown in table 2.By form it will be seen that there is many words not to be of practical significance, than
Such as " interjection ", " modal particle ", this part word do not include semantic information, if carrying out keyword expansion, can increase retrieval burden,
Improve computation complexity.And only noun, verb, adjective contains more semantic information, therefore use herein according to part of speech
The method screened to keyword.Some punctuation marks are also included in certain word segmentation result, for these punctuation marks also one
And filter out.
The Chinese part of speech classification chart of table 2
In the result of participle, we can be labeled part of speech, and all keywords are filtered according to the part of speech got,
It is main to retain noun, verb, adjective, then obtain a lexical item set.Analysis is found, wherein some nothings still can be included
The lexical item of meaning, therefore herein on the basis of above-mentioned processing, then using the secondary keyword of method progress for disabling dictionary filtering
Filtering, the deactivation dictionary, can be safeguarded by user oneself.
3rd, the query expansion based on body
By the processing of a upper trifle, we have obtained the keyword of activity, next need to carry out according to some way crucial
Word extends.At present in information retrieval system, largely it is all based on what keyword was inquired about, obtaining user first needs to look into
The keyword of inquiry, then carries out query expansion, the result after extension is retrieved, extended mode generally use synonym automatically
Development method, that is, be extended using synonymicon, near synonym dictionary.
, cannot be from although although the query expansion based on keyword largely improves the effect of resource retrieval
Fundamentally solve the problems, such as because traditional query expansion is extended on the level of Symbol matching, it be using query word as
Center, mechanically carries out character string extension, so as to have ignored between query word semantic information and query word and other concepts
Semantic association, therefore user's query intention cannot give full expression to.
Since this physical efficiency is comprehensive, accurately description and the relation between defined notion and concept, there is stronger semanteme
Ability to express, can be better understood from the semantic intention of user.Therefore, to solve the above problems, set forth herein in acquisition activity
After subject key words group, need to optimize it processing using body processing correlation technique in next step, that is, it is described
Query expansion.In order to Ontological concept related content extend, consider first herein using the hierarchical relationship in body come into
Row processing.The relation occurred in inquiry in commonly used body has：
Synonym relation：Expansion concept is the synonym of query concept；
Set membership：Expansion concept and father and son's node that query concept is in body hierarchical structure；
Children tree nodes：Expansion concept is the node in the subtree of query concept；
The brotgher of node：Expansion concept is the brotgher of node of query concept, there is identical father node；
Fraternal children tree nodes：Expansion concept is the node in the fraternal subtree of query concept；
Herein primary concern is that being extended to set membership node to query concept, because the father in body hierarchical structure
Subrelation is the most universal, and the most easily obtain.Its process flow is：
1. pair each keyword of active theme obtained, it is concept, example or the attribute in body to judge it；
2. if be concept or example, the main set membership utilized in body construction is extended；If there are attribute,
Using the relation between attribute and concept, the semantic relation according to present in body and inference mechanism, excavate implicit information；
3. spreading result is preserved, subsequently to carry out the retrieval of cultural resource information；
4th, the realization of cultural resource retrieval module
The realization of the retrieval module of cultural resource employs the Lucene full-text search engine kits based on Java.Lucene makees
For an outstanding full-text search engine, its system structure has strong Object-oriented Features.One is first defined with putting down
The unrelated index file form of platform, is secondly designed to abstract class, specific platform realizes part by the core component of system
The realization of abstract class is designed as, is also encapsulated as class with the such as file storage of platform specific relevant portion in addition.Therefore, Lucene can
To be very easily applied in sorts of systems.The system architecture diagram of Lucene is as shown in Figure 4.
Claims (5)
1. a kind of personalized recommendation method, it is characterised in that collect user behavior data first, the user's behavioral data is divided into
The active theme data that the preference pattern data of user and user currently browse, then applicating cooperation filter algorithm is to described inclined
Good model data is handled, and is recommended to cultural activity of the user based on collaborative filtering；At the same time, currently browsed according to user
Active theme data, carry out cultural data based on body and recommend.
2. personalized recommendation method according to claim 1, it is characterised in that the cultural activity based on collaborative filtering is recommended
Flow include：
(1) user enters in system platform, and system platform can collect the behavior historical data of user, such as participates in cultural activity note
Record, activity thumb up record, activity collection record, these data records indicate the hobby of user；
(2) system generates user-activity score matrix according to pre-set scoring；
(3) start proposed algorithm flow, the similarity of user is calculated using formula, finds out the highest N number of user of similarity the most
Nearest-neighbors collection, score in predicting is carried out using nearest-neighbors collection, by prediction marking and queuing, obtains recommendation results, result is presented
To user.
3. personalized recommendation method according to claim 1, it is characterised in that the preference pattern data include user
Activity thumb up data, the data of Collecting, the record data of activity.
4. personalized recommendation method according to claim 3, it is characterised in that the cultural data is recommended to include following
Step：
(1) active theme keyword obtains：The active title currently browsed to user carries out word segmentation processing, the theme of extraction activity
Keyword；There can be the insignificant words such as " ", " " and punctuation mark toward contact in word segmentation result, also need to use for this
Stop words is carried out based on the mode for disabling vocabulary to operate；
(2) query expansion based on body：By above-mentioned processing, the keyword phrase of active theme, this keyword have been obtained
Phrase, which may include noun, may also include adjective, be exactly to put it into ontology library in next step, carry out based on body
Query expansion processing；
(3) Lucene resource retrievals module：Using by the use of being obtained after Ontology Query extension process as keyword, using based on Java
Full text information retrieval instrument Lucene carry out the retrieval of resource.
5. personalized recommendation method according to claim 1, it is characterised in that the cultural data recommends active theme
Keyword acquisition methods are：
(1) preprocessing process：Word segmentation processing is carried out using natural language processing technique, a word has been obtained by word segmentation processing
Group, carries out screening and filtering to keyword according to part of speech, retains noun, verb, adjective；
(2) keyword expansion：At present in information retrieval system, largely it is all based on what keyword was inquired about, obtains first
The keyword that family needs to inquire about is taken, query expansion is then carried out automatically, the result after extension is retrieved, extended mode leads to
It is extended frequently with synonym development method, that is, using synonymicon, near synonym dictionary；
(3) query expansion：Processing is optimized to it using body processing correlation technique, in order to expand Ontological concept related content
Exhibition, is handled using the hierarchical relationship in body, set membership node is extended query concept, at it
Managing flow is：
A. to each keyword of active theme of acquisition, it is concept, example or the attribute in body to judge it；
If being b. concept or example, the main set membership utilized in body construction is extended；If there are attribute,
Using the relation between attribute and concept, the semantic relation according to present in body and inference mechanism, excavate implicit information；
C. spreading result is preserved, subsequently to carry out the retrieval of cultural resource information.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711155639.7A CN108038133A (en) | 2017-11-20 | 2017-11-20 | Personalized recommendation method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711155639.7A CN108038133A (en) | 2017-11-20 | 2017-11-20 | Personalized recommendation method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108038133A true CN108038133A (en) | 2018-05-15 |
Family
ID=62092607
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711155639.7A Pending CN108038133A (en) | 2017-11-20 | 2017-11-20 | Personalized recommendation method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108038133A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108897887A (en) * | 2018-07-10 | 2018-11-27 | 华南师范大学 | A kind of teaching resource recommended method of knowledge based map and user's similarity |
| CN108921641A (en) * | 2018-05-31 | 2018-11-30 | 江苏乙生态农业科技有限公司 | Method is adopted in a kind of recommendation based on double filtration |
| CN109582852A (en) * | 2018-12-05 | 2019-04-05 | 中国银行股份有限公司 | A kind of sort method and system of full-text search result |
| CN109657129A (en) * | 2018-12-26 | 2019-04-19 | 北京百度网讯科技有限公司 | For obtaining the method and device of information |
| CN109684553A (en) * | 2018-12-26 | 2019-04-26 | 北京百度网讯科技有限公司 | For obtaining the method and device of information |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102663010A (en) * | 2012-03-20 | 2012-09-12 | 复旦大学 | Personalized image browsing and recommending method based on labelling semantics and system thereof |
| CN103425799A (en) * | 2013-09-04 | 2013-12-04 | 北京邮电大学 | Personalized research direction recommending system and method based on themes |
| CN104378659A (en) * | 2014-10-31 | 2015-02-25 | 四川长虹电器股份有限公司 | Personalization recommendation method based on smart television |
| CN105023178A (en) * | 2015-08-12 | 2015-11-04 | 电子科技大学 | Main body-based electronic commercere commendation method |
| CN105843860A (en) * | 2016-03-17 | 2016-08-10 | 山东大学 | Microblog attention recommendation method based on parallel item-based collaborative filtering algorithm |
| CN106209959A (en) * | 2015-05-26 | 2016-12-07 | 徐尚英 | Network service intelligence based on user's request finds method |
| WO2017157090A1 (en) * | 2016-03-15 | 2017-09-21 | 北京京东尚科信息技术有限公司 | Similarity mining method and device |
-
2017
- 2017-11-20 CN CN201711155639.7A patent/CN108038133A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102663010A (en) * | 2012-03-20 | 2012-09-12 | 复旦大学 | Personalized image browsing and recommending method based on labelling semantics and system thereof |
| CN103425799A (en) * | 2013-09-04 | 2013-12-04 | 北京邮电大学 | Personalized research direction recommending system and method based on themes |
| CN104378659A (en) * | 2014-10-31 | 2015-02-25 | 四川长虹电器股份有限公司 | Personalization recommendation method based on smart television |
| CN106209959A (en) * | 2015-05-26 | 2016-12-07 | 徐尚英 | Network service intelligence based on user's request finds method |
| CN105023178A (en) * | 2015-08-12 | 2015-11-04 | 电子科技大学 | Main body-based electronic commercere commendation method |
| WO2017157090A1 (en) * | 2016-03-15 | 2017-09-21 | 北京京东尚科信息技术有限公司 | Similarity mining method and device |
| CN105843860A (en) * | 2016-03-17 | 2016-08-10 | 山东大学 | Microblog attention recommendation method based on parallel item-based collaborative filtering algorithm |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108921641A (en) * | 2018-05-31 | 2018-11-30 | 江苏乙生态农业科技有限公司 | Method is adopted in a kind of recommendation based on double filtration |
| CN108897887A (en) * | 2018-07-10 | 2018-11-27 | 华南师范大学 | A kind of teaching resource recommended method of knowledge based map and user's similarity |
| CN108897887B (en) * | 2018-07-10 | 2020-10-16 | 华南师范大学 | Teaching resource recommendation method based on knowledge graph and user similarity |
| CN109582852A (en) * | 2018-12-05 | 2019-04-05 | 中国银行股份有限公司 | A kind of sort method and system of full-text search result |
| CN109582852B (en) * | 2018-12-05 | 2021-04-09 | 中国银行股份有限公司 | Method and system for sorting full-text retrieval results |
| CN109657129A (en) * | 2018-12-26 | 2019-04-19 | 北京百度网讯科技有限公司 | For obtaining the method and device of information |
| CN109684553A (en) * | 2018-12-26 | 2019-04-26 | 北京百度网讯科技有限公司 | For obtaining the method and device of information |
| CN109657129B (en) * | 2018-12-26 | 2023-04-18 | 北京百度网讯科技有限公司 | Method and device for acquiring information |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108038133A (en) | Personalized recommendation method | |
| Chen et al. | Websrc: A dataset for web-based structural reading comprehension | |
| CN102929873B (en) | Method and device for extracting searching value terms based on context search | |
| Van Zwol et al. | Faceted exploration of image search results | |
| Foley et al. | Learning to extract local events from the web | |
| CN105528437B (en) | A kind of question answering system construction method extracted based on structured text knowledge | |
| CN106095762A (en) | A kind of news based on ontology model storehouse recommends method and device | |
| CN111475625A (en) | News manuscript generation method and system based on knowledge graph | |
| Allisio et al. | Felicittà: Visualizing and estimating happiness in italian cities from geotagged tweets | |
| CN110888991A (en) | Sectional semantic annotation method in weak annotation environment | |
| Nakashole et al. | Real-time population of knowledge bases: opportunities and challenges | |
| Hao et al. | Semantic patterns for user‐interactive question answering | |
| CN103514289A (en) | Method and device for building interest entity base | |
| Jiang et al. | HyOASAM: A hybrid open API selection approach for mashup development | |
| KR101818107B1 (en) | Apparatus and method for supporting storytelling authoring | |
| An et al. | Automatic generation of ontology from the deep web | |
| Samarinas et al. | ProCIS: A Benchmark for Proactive Retrieval in Conversations | |
| Wasim et al. | Extracting and modeling user interests based on social media | |
| Belkaroui et al. | Conversational based method for tweet contextualization | |
| KR101132974B1 (en) | Apparatus and method for modeling ontology of multimodal social network | |
| Zhao et al. | A system to manage and mine microblogging data | |
| Zhang et al. | WiiCluster: A platform for wikipedia infobox generation | |
| KR101526312B1 (en) | System and method for packaging research and development information services on pending national issues | |
| Kawamura et al. | Science graph for characterizing the recent scientific landscape using paragraph vectors | |
| Blandfort et al. | Image captioning in the wild: how people caption images on Flickr |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180515 |