CN110399549A - A kind of collaborative filtering method of user oriented interest reconciliation similarity - Google Patents
A kind of collaborative filtering method of user oriented interest reconciliation similarity Download PDFInfo
- Publication number
- CN110399549A CN110399549A CN201810371429.XA CN201810371429A CN110399549A CN 110399549 A CN110399549 A CN 110399549A CN 201810371429 A CN201810371429 A CN 201810371429A CN 110399549 A CN110399549 A CN 110399549A
- Authority
- CN
- China
- Prior art keywords
- user
- indicate
- similarity
- project
- interest
- 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
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The present invention relates to a kind of collaborative filtering methods of user oriented interest reconciliation similarity, this method comprises: by being decomposed to original rating matrix, time-based user interest degree weighting function is introduced into Pearson came similarity function, and the user's evaluation degree trust factor and user's evaluation deviation trust-factor combine to obtain user interest reconciliation similarity, recommendation results are obtained using original rating matrix and user interest reconciliation similarity, and meet the project of demand to user's recommendation.Not only the probability matrix decomposition method more traditional in the case where user's score data is sparse can significantly improve recommendation precision to method provided by the invention, but also can be recommended in different time sections interested article according to user.
Description
Technical field
The present invention relates to the recommendation fields of data mining, and in particular, to a kind of user oriented interest reconciles similarity
Collaborative filtering method.
Background technique
As user's participation information produces, network information scale is in explosive growth.Massive information provides for information retrieval
It is possible to simultaneously result in information overload.In order to mitigate above-mentioned contradiction, user is helped accurately to be quickly found out in mass data
Its interested information, recommender system are come into being.
Collaborative filtering is common method in recommender system, in order to solve the problems, such as the sparsity in recommendation process, one
As using the collaborative filtering recommending technology decomposed based on probability matrix, common thinking is to calculate user first like degree, is then determined
Neighborhood, finally prediction scoring.As the development of recommended technology starts to occur to obtain higher-quality recommendation precision
The trust of social networks is introduced into recommended technology.The key link of the technology be trust in similarity and social networks because
The fusion of son.But the technology have ignored with time change and the interest of user also the problem of constantly changing, thus
Interests change, the timeliness that user cannot be reacted in time are insufficient, to cause the problem for recommending precision not high.
Therefore, existing recommended technology could be improved and develop.
Summary of the invention
In view of above-mentioned deficiencies of the prior art, the invention proposes a kind of collaborations of user oriented interest reconciliation similarity
Filtering method, specifically, method includes the following steps:
(1) original rating matrix R is obtained using user's score data and probability matrix decomposition is carried out to it, obtain approximate scoring square
Battle array；
(2) original rating matrix R and approximate rating matrix are traversedIt is similar to obtain the Pearson came to score based on user's common interest
DegreeSim fill_t_peirson ；
(3) it traverses original rating matrix R and obtains the user's evaluation degree trust factorN i ；
(4) it traverses original rating matrix R and obtains user's evaluation deviation trust-factor D i ；
(5) amendment user interest similarity is utilizedsim fill_t_pearson , the user's evaluation degree trust factorN i It is inclined with user's evaluation
Poor trust-factor D i , calculate user interest reconciliation similaritysim t ；
(6) recommendation results are calculated using original rating matrix and user interest reconciliation similarity matrix, forms recommendation list and carries out
Recommend.
The collaborative filtering method of a kind of user oriented interest reconciliation similarity provided by the invention, first to original scoring square
Battle array carries out probability matrix and decomposes to obtain approximate rating matrix, and time-based user interest degree weighting function is then introduced into skin
User interest similarity is obtained in your inferior similarity function, user's evaluation grade is then obtained by the trusting relationship of original matrix
Trust-factor and user's evaluation deviation trust-factor further utilize amendment user interest similarity, user's evaluation degree trust
The factor and user's evaluation deviation trust-factor calculate user interest reconciliation similarity, ultimately form recommendation list and recommended.
Compared with traditional collaborative filtering decomposed based on probability matrix, the letter between user is not only being used in recommendation process
The relationship of appointing, and the time-based interest-degree function of user is added in collaborative filtering, it can accomplish to be existed according to user
The different time sections more accurately potential interested article of recommended user.
Detailed description of the invention
Fig. 1 is a kind of collaborative filtering method flow chart of user oriented interest reconciliation similarity.
Specific embodiment
It is as shown in Figure 1 the collaborative filtering method of user interest reconciliation similarity, comprising the following steps:
(1) original rating matrix R is obtained using user's score data and probability matrix decomposition is carried out to it, obtain approximate scoring square
Battle array；
(2) original rating matrix R and approximate rating matrix are traversedIt is similar to obtain the Pearson came to score based on user's common interest
DegreeSim fill_t_peirson ；
(3) it traverses original rating matrix R and obtains the user's evaluation degree trust factorN i ；
(4) it traverses original rating matrix R and obtains user's evaluation deviation trust-factor D i ；
(5) amendment user interest similarity is utilizedsim fill_t_pearson , the user's evaluation degree trust factorN i It is inclined with user's evaluation
Poor trust-factor D i , calculate user interest reconciliation similaritysim t ；
(6) recommendation results are calculated using original rating matrix and user interest reconciliation similarity matrix, forms recommendation list and carries out
Recommend.
The specific implementation details of each step are as follows:
In step (1), more specifically, indicate that consumer bought the matrix of article using matrix R, rating matrix R isiIt is a
User couplejThe scoring of a project, every row indicate that the consumer record of a certain consumer, each column indicate different consumers to the commodity
Scoring；Enable approximate rating matrix= U T V uses mean value for 0, and the Gaussian Profile that variance is 0.1 is initialized；Wherein U ∈
RkWith V ∈ RkRespectively indicate the hidden factor matrix of user and the hidden factor matrix of project, characteristic dimension k；
Step 1.1: establish approximate rating matrix loss function:
Wherein,iIndicate user,jExpression project；NIndicate useriNumber,MExpression projectjNumber；R ij For useriTo item
MeshjScoring；U∈RkWith V ∈ RkRespectively indicate the hidden factor matrix of user and the hidden factor matrix of project, characteristic dimension k；I ij For
UseriWhether to projectjIt scores, is otherwise 0 if scoring is 1；U i ForkIn the hidden factor matrix in Wesy familyiA user
'skPreference vector is tieed up,V j It indicateskIn the hidden factor matrix of dimension projectjA projectkTie up preference vector；U i T ForU i Transposition；λ u
Withλ v For regularization factors；
Step 1.2: for each score dataR ij If loss function variation is greater than given threshold after iteration twice in succession
0.001, it is corresponding that iteration is updated using stochastic gradient descentU i WithV j Training parameter:
Wherein,nIndicate the number of iterations, α indicates Studying factors； U i n WithV j n Indicate thenIt is corresponding after secondary iterationU i WithV j To
Amount,U i n+1 WithV j n+1 Indicate then+It is corresponding after 1 iterationU i WithV j Vector, initializationUWithVFor 0 matrix；Each iteration
By current iterationU i n WithV j n Subtract partial derivative sum product be calculated it is newU i n+1 WithV j n+1 Vector；It is lost after iteration
Function variation is less than threshold value or the number of iterationsnGreater than 10000；
Step 1.3: using the result of last time iteration as finalUWithV, pass through= U T User is calculated in ViTo projectjApproximate rating matrix。
In step (2), more specifically include:
Step 2.1: utilizing original rating matrixRCalculate Pearson came similaritysim pearson :
Whereinsim pearson (i,b)Indicate Pearson came similaritysim pearson Middle different useriAnd userbSimilarity；PTable
Show the set of all users in original matrix R,jExpression project；R i,j Indicate useriTo projectjScore value,R b,j Indicate userbTo projectjScore value；WithRespectively indicate useriScoring mean value and userbScoring mean value；
Step 2.2: decaying exponential function is defined according to forgetting curve and obtains time-based user interest weighting function:
Wherein,w t Indicate time-based user interest weight,t i,j Indicate useriTo projectjThe scoring time,t 0 Indicate mesh
The sampling time of user is marked,TIndicate the time span (time started end time -) of entire data set；
Step 2.3: the time-based user interest weighting function of step 2.2 is introduced into the Pearson came similarity of step 2.1sim pearson In calculation formula, it is calculated and improves Pearson came similarity:
Wherein,sim pearson (i,b)Indicate different useriAnd userbBetween Pearson came similarity；
Step 2.4: utilizing original rating matrixRCalculate useriAnd userbBetween the project set that needs to fill:
Wherein P i∪b Indicate useriScore vector and userbThe union of scoring vector, useriAnd userbIt is indicated in data set
Different users, P i And P b Indicate useriAnd userbScoring vector；Indicate useriOr userbThe project evaluated
User in setiThe project set that do not evaluate,Indicate userbOr useriUser in the project set evaluatedbNot yet
There is the project set of evaluation；
Step 2.5: using user's scoring vector after filling, calculating useriAnd userbCommon interest scoring Pearson came it is similar
Degreesim fill_t_peirson :
Wherein,sim fill_t_peirson (i,b) indicate by the calculated user of institute after fillingiWith userbPearson came similarity；P i∪b Indicate useriAnd userbThe union of scoring item,WithIt respectively indicates towards the user after filling upiTo projectj
Scoring and userbTo projectjScoring.
In step (3), more specifically, according to original rating matrixRCalculate the user's evaluation degree trust factorN i :
Wherein, num(i)Indicate original rating matrixRMiddle useriThe item number evaluated,Indicate original
Rating matrixRIn all user's evaluations cross the mean value of item number.
In step (4), more specifically, according to original rating matrixRCalculate user's evaluation deviation trust-factorD i :
Wherein,d i,j Indicate useriTo projectjEffort analysis, if deviation be greater than ε if be 1, be otherwise 0；r i,j Indicate original
Rating matrixRMiddle useriTo projectjScore value,Indicate useriScoring mean value；ε is deviation threshold, is set as 0.5；d i
Indicate useriOwn in the project set evaluatedd i,j The sum of；D i Indicate useriEvaluation deviation factors,Q i Indicate useriIt comments
The set for all items composition that valence is crossed.
In step (5), more specifically, utilizing the Pearson came similarity based on user interestsim fill_t_pearson , user
Opinion rating trust-factorN i With user's evaluation deviation trust-factorD i Weighted calculation user interest reconciliation similarity:
Wherein,sim t (i,b)For useriAnd userbInterest reconcile similarity,βFor weight regulatory factor, and 0≤β≤1。
In step (6), more specifically include:
Step 6.1: for each target useru 0 To projectjScore in predictingr u0,j , traverse original rating matrixRIt finds to item
MeshjThere is user's set of scoring behaviorU rated ；In conjunction withU rated With user interest reconciliation similarity matrixsim t , rightU rated Into
Row descending arrangement, before takingkA user obtains and useru 0 NeighborhoodU neighbor ；
Step 6.2: according to original rating matrixR, the emerging reconciliation similarity of usersim t And neighborhoodU neighbor Calculate pre- assessment
Point:
Wherein,r u0,j Indicate target useru 0 To projectjPrediction scoring,Indicate target useru 0 Scoring mean value；uTable
Show the user in neighborhood,r u,j Indicate neighboursuTo projectjScore value,Indicate neighbor useruScoring mean value；U neighbor Indicate neighborhood,sim t (u 0, u) indicate target useru 0 And neighbor useruUser interest reconcile similarity；
Step 6.3: finally can form recommendation list according to the height of prediction scoring, recommend the project for meeting demand to user.
The collaborative filtering method of a kind of user oriented interest reconciliation similarity provided by the invention, first to original scoring square
Battle array carries out probability matrix and decomposes to obtain approximate rating matrix, and time-based user interest degree weighting function is then introduced into skin
User interest similarity is obtained in your inferior similarity function, user's evaluation grade is then obtained by the trusting relationship of original matrix
Trust-factor and user's evaluation deviation trust-factor further utilize amendment user interest similarity, user's evaluation degree trust
The factor and user's evaluation deviation trust-factor calculate user interest reconciliation similarity, ultimately form recommendation list and recommended.
Compared with traditional collaborative filtering decomposed based on probability matrix, the letter between user is not only being used in recommendation process
The relationship of appointing, and the time-based interest-degree function of user is added in collaborative filtering, it can accomplish to be existed according to user
The different time sections more accurately potential interested article of recommended user.
Claims (7)
1. a kind of collaborative filtering method of user oriented interest reconciliation similarity, which is characterized in that method includes the following steps:
(1) original rating matrix R is obtained using user's score data and probability matrix decomposition is carried out to it, obtain approximate scoring square
Battle array；
(2) original rating matrix R and approximate rating matrix are traversedIt is similar to obtain the Pearson came to score based on user's common interest
DegreeSim fill_t_peirson ；
(3) it traverses original rating matrix R and obtains the user's evaluation degree trust factorN i ；
(4) it traverses original rating matrix R and obtains user's evaluation deviation trust-factor D i ；
(5) amendment user interest similarity is utilizedsim fill_t_pearson , the user's evaluation degree trust factorN i It is inclined with user's evaluation
Poor trust-factor D i , calculate user interest reconciliation similaritysim t ；
(6) recommendation results are calculated using original rating matrix and user interest reconciliation similarity matrix, forms recommendation list and carries out
Recommend.
2. a kind of collaborative filtering method of user oriented interest reconciliation similarity according to claim 1, which is characterized in that described
Step (1) specifically includes:
Step 1.1: establish approximate rating matrix loss function:
Wherein,iIndicate user,jExpression project；NIndicate useriNumber,MExpression projectjNumber；R ij For useriTo item
MeshjScoring；U∈RkWith V ∈ RkRespectively indicate the hidden factor matrix of user and the hidden factor matrix of project, characteristic dimension k；I ij For
UseriWhether to projectjIt scores, is otherwise 0 if scoring is 1；U i ForkIn the hidden factor matrix in Wesy familyiA user
'skPreference vector is tieed up,V j It indicateskIn the hidden factor matrix of dimension projectjA projectkTie up preference vector；U i T ForU i Transposition；λ u
Withλ v For regularization factors；
Step 1.2: for each score dataR ij If loss function variation is greater than given threshold after iteration twice in succession
0.001, it is corresponding that iteration is updated using stochastic gradient descentU i WithV j Training parameter:
Wherein,nIndicate the number of iterations, α indicates Studying factors； U i n WithV j n Indicate thenIt is corresponding after secondary iterationU i WithV j To
Amount,U i n+1 WithV j n+1 Indicate then+It is corresponding after 1 iterationU i WithV j Vector, initializationUWithVFor 0 matrix；Each iteration
By current iterationU i n WithV j n Subtract partial derivative sum product be calculated it is newU i n+1 WithV j n+1 Vector；It is lost after iteration
Function variation is less than threshold value or the number of iterationsnGreater than 10000；
Step 1.3: using the result of last time iteration as finalUWithV, pass through= U T User is calculated in ViTo projectjApproximate rating matrix。
3. a kind of collaborative filtering method of user oriented interest reconciliation similarity according to claim 1, which is characterized in that described
Step (2) specifically includes:
Step 2.1: utilizing original rating matrixRCalculate Pearson came similaritysim pearson :
Whereinsim pearson (i,b)Indicate Pearson came similaritysim pearson Middle different useriAnd userbSimilarity；PIt indicates
The set of all users in original matrix R,jExpression project；R i,j Indicate useriTo projectjScore value,R b,j Indicate userb
To projectjScore value；WithRespectively indicate useriScoring mean value and userbScoring mean value；
Step 2.2: decaying exponential function is defined according to forgetting curve and obtains time-based user interest weighting function:
Wherein,w t Indicate time-based user interest weight,t i,j Indicate useriTo projectjThe scoring time,t 0 Indicate target
The sampling time of user,TIndicate the time span (time started end time -) of entire data set；
Step 2.3: the time-based user interest weighting function of step 2.2 is introduced into the Pearson came similarity of step 2.1sim pearson In calculation formula, it is calculated and improves Pearson came similarity:
Wherein,sim pearson (i,b)Indicate different useriAnd userbBetween Pearson came similarity；
Step 2.4: utilizing original rating matrixRCalculate useriAnd userbBetween the project set that needs to fill:
Wherein P i∪b Indicate useriScore vector and userbThe union of scoring vector, useriAnd userbIt is indicated in data set
Different users, P i And P b Indicate useriAnd userbScoring vector；Indicate useriOr userbThe project evaluated
User in setiThe project set that do not evaluate,Indicate userbOr useriUser in the project set evaluatedbNot yet
There is the project set of evaluation；
Step 2.5: using user's scoring vector after filling, calculating useriAnd userbCommon interest scoring Pearson came it is similar
Degreesim fill_t_peirson :
Wherein,sim fill_t_peirson (i,b) indicate by the calculated user of institute after fillingiWith userbPearson came similarity；P i∪b Indicate useriAnd userbThe union of scoring item,WithIt respectively indicates towards the user after filling upiTo projectj
Scoring and userbTo projectjScoring.
4. a kind of collaborative filtering method of user oriented interest reconciliation similarity according to claim 1, which is characterized in that in step
Suddenly in (3), according to original rating matrixRCalculate the user's evaluation degree trust factorN i :
Wherein, num(i)Indicate original rating matrixRMiddle useriThe item number evaluated,Indicate original
Rating matrixRIn all user's evaluations cross the mean value of item number.
5. a kind of collaborative filtering method of user oriented interest reconciliation similarity according to claim 1, which is characterized in that in step
Suddenly in (4), according to original rating matrixRCalculate user's evaluation deviation trust-factorD i :
Wherein,d i,j Indicate useriTo projectjEffort analysis, if deviation be greater than ε if be 1, be otherwise 0；r i,j Indicate original
Rating matrixRMiddle useriTo projectjScore value,Indicate useriScoring mean value；ε is deviation threshold, is set as 0.5；d i
Indicate useriOwn in the project set evaluatedd i,j The sum of；D i Indicate useriEvaluation deviation factors,Q i Indicate useriIt comments
The set for all items composition that valence is crossed.
6. a kind of collaborative filtering method of user oriented interest reconciliation similarity according to claim 1, which is characterized in that in step
Suddenly in (5), the Pearson came similarity based on user interest is utilizedsim fill_t_pearson , the user's evaluation degree trust factorN i With with
Evaluate deviation trust-factor in familyD i Weighted calculation user interest reconciliation similarity:
Wherein,sim t (i,b)For useriAnd userbInterest reconcile similarity,βFor weight regulatory factor, and 0≤β≤1。
7. a kind of collaborative filtering method of user oriented interest reconciliation similarity according to claim 1, which is characterized in that in step
Suddenly in (6), according to original rating matrixR, the emerging reconciliation similarity of usersim t And neighborhoodU neighbor Calculate prediction scoring:
Wherein,r u0,j Indicate target useru 0 To projectjPrediction scoring,Indicate target useru 0 Scoring mean value；uIt indicates
User in neighborhood,r u,j Indicate neighboursuTo projectjScore value,Indicate neighbor useruScoring mean value；U neighbor Indicate neighborhood,sim t (u 0, u) indicate target useru 0 And neighbor useruUser interest reconcile similarity.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810371429.XA CN110399549A (en) | 2018-04-24 | 2018-04-24 | A kind of collaborative filtering method of user oriented interest reconciliation similarity |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810371429.XA CN110399549A (en) | 2018-04-24 | 2018-04-24 | A kind of collaborative filtering method of user oriented interest reconciliation similarity |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110399549A true CN110399549A (en) | 2019-11-01 |
Family
ID=68320305
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810371429.XA Pending CN110399549A (en) | 2018-04-24 | 2018-04-24 | A kind of collaborative filtering method of user oriented interest reconciliation similarity |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110399549A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111063227A (en) * | 2019-12-17 | 2020-04-24 | 河南科技学院 | Political thinking answer method and system based on mobile terminal |
| CN111199422A (en) * | 2019-12-25 | 2020-05-26 | 浙江工商大学 | Improved LFM (Linear frequency modulation) collaborative filtering method fusing time attenuation factors |
| CN111400602A (en) * | 2020-03-20 | 2020-07-10 | 珠海读书郎网络教育有限公司 | Online learning system and method based on personalized recommendation |
| CN113221003A (en) * | 2021-05-20 | 2021-08-06 | 北京建筑大学 | Mixed filtering recommendation method and system based on dual theory |
| CN113254777A (en) * | 2021-06-07 | 2021-08-13 | 武汉卓尔数字传媒科技有限公司 | Information recommendation method and device, electronic equipment and storage medium |
| CN113256387A (en) * | 2021-06-15 | 2021-08-13 | 杭州魔点科技有限公司 | Personalized recommendation method for item, electronic equipment and storage medium |
| CN114416246A (en) * | 2021-12-31 | 2022-04-29 | 北京五八信息技术有限公司 | Data processing method and device, electronic equipment and storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090106314A1 (en) * | 2007-10-22 | 2009-04-23 | Samsung Electronics Co., Ltd. | Situation-aware recommendation using correlation |
| CN103729798A (en) * | 2014-01-29 | 2014-04-16 | 河南理工大学 | Coal mine safety evaluation system based on improved k-means clustering |
| CN104881499A (en) * | 2015-06-19 | 2015-09-02 | 中国石油大学(华东) | Collaborative filtering recommendation method based on attribute rating scaling |
| CN106296337A (en) * | 2016-07-13 | 2017-01-04 | 扬州大学 | Dynamic recommendation method based on Non-negative Matrix Factorization |
| CN106570090A (en) * | 2016-10-20 | 2017-04-19 | 杭州电子科技大学 | Method for collaborative filtering recommendation based on interest changes and trust relations |
| US20180001184A1 (en) * | 2016-05-02 | 2018-01-04 | Bao Tran | Smart device |
-
2018
- 2018-04-24 CN CN201810371429.XA patent/CN110399549A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090106314A1 (en) * | 2007-10-22 | 2009-04-23 | Samsung Electronics Co., Ltd. | Situation-aware recommendation using correlation |
| CN103729798A (en) * | 2014-01-29 | 2014-04-16 | 河南理工大学 | Coal mine safety evaluation system based on improved k-means clustering |
| CN104881499A (en) * | 2015-06-19 | 2015-09-02 | 中国石油大学(华东) | Collaborative filtering recommendation method based on attribute rating scaling |
| US20180001184A1 (en) * | 2016-05-02 | 2018-01-04 | Bao Tran | Smart device |
| CN106296337A (en) * | 2016-07-13 | 2017-01-04 | 扬州大学 | Dynamic recommendation method based on Non-negative Matrix Factorization |
| CN106570090A (en) * | 2016-10-20 | 2017-04-19 | 杭州电子科技大学 | Method for collaborative filtering recommendation based on interest changes and trust relations |
Non-Patent Citations (2)
| Title |
|---|
| HUI LI: "Social network recommendation system mixing contex information", 《CAAI TRANSACTIONS ON INTELLIGENT SYSTEMS》 * |
| 王建芳: "基于相似性填充和信任因子的协同过滤算法", 《计算机应用与软件》 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111063227A (en) * | 2019-12-17 | 2020-04-24 | 河南科技学院 | Political thinking answer method and system based on mobile terminal |
| CN111063227B (en) * | 2019-12-17 | 2021-05-07 | 河南科技学院 | Political thinking answer method and system based on mobile terminal |
| CN111199422A (en) * | 2019-12-25 | 2020-05-26 | 浙江工商大学 | Improved LFM (Linear frequency modulation) collaborative filtering method fusing time attenuation factors |
| CN111400602A (en) * | 2020-03-20 | 2020-07-10 | 珠海读书郎网络教育有限公司 | Online learning system and method based on personalized recommendation |
| CN113221003A (en) * | 2021-05-20 | 2021-08-06 | 北京建筑大学 | Mixed filtering recommendation method and system based on dual theory |
| CN113221003B (en) * | 2021-05-20 | 2023-05-02 | 北京建筑大学 | Mixed filtering recommendation method and system based on dual theory |
| CN113254777A (en) * | 2021-06-07 | 2021-08-13 | 武汉卓尔数字传媒科技有限公司 | Information recommendation method and device, electronic equipment and storage medium |
| CN113256387A (en) * | 2021-06-15 | 2021-08-13 | 杭州魔点科技有限公司 | Personalized recommendation method for item, electronic equipment and storage medium |
| CN114416246A (en) * | 2021-12-31 | 2022-04-29 | 北京五八信息技术有限公司 | Data processing method and device, electronic equipment and storage medium |
| CN114416246B (en) * | 2021-12-31 | 2024-03-19 | 北京五八信息技术有限公司 | Data processing method and device, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110399549A (en) | A kind of collaborative filtering method of user oriented interest reconciliation similarity | |
| CN111199343B (en) | Multi-model fusion tobacco market supervision abnormal data mining method | |
| CN105740430B (en) | A kind of personalized recommendation method of mosaic society's information | |
| Guo et al. | High-dimensional and banded vector autoregressions | |
| CN111507796A (en) | Online shopping mall commodity recommendation method based on knowledge graph | |
| CN104899246B (en) | Collaborative filtering recommending method based on blurring mechanism user scoring neighborhood information | |
| CN102982466B (en) | A kind of score in predicting method based on user's liveness | |
| CN104063481A (en) | Film individuation recommendation method based on user real-time interest vectors | |
| CN110503531A (en) | The dynamic social activity scene recommended method of timing perception | |
| CN106651519A (en) | Method and system for personalized recommendation based on label information | |
| CN103399858A (en) | Socialization collaborative filtering recommendation method based on trust | |
| CN106503140A (en) | One kind is based on Hadoop cloud platform web resource personalized recommendation system and method | |
| CN105913290A (en) | Commodity matching recommending method and recommending system | |
| Eliyas et al. | Recommendation systems: Content-based filtering vs collaborative filtering | |
| CN107025311A (en) | A kind of Bayes's personalized recommendation method and device based on k nearest neighbor | |
| CN108038746A (en) | Method is recommended based on the bigraph (bipartite graph) of key user and time context | |
| CN107301583A (en) | It is a kind of that method is recommended based on user preference and the cold start-up trusted | |
| CN106919699A (en) | A kind of recommendation method for personalized information towards large-scale consumer | |
| CN106791964A (en) | Broadcast TV program commending system and method | |
| CN105678590A (en) | topN recommendation method for social network based on cloud model | |
| CN112231583A (en) | E-commerce recommendation method based on dynamic interest group identification and generation of countermeasure network | |
| CN109034981A (en) | A kind of electric business collaborative filtering recommending method | |
| CN110362755A (en) | A kind of recommended method of the hybrid algorithm based on article collaborative filtering and correlation rule | |
| CN108109058A (en) | A kind of single classification collaborative filtering method for merging personal traits and article tag | |
| Yu et al. | Spectrum-enhanced pairwise learning to rank |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20191101 |