CN102495864A - Collaborative filtering recommending method and system based on grading - Google Patents

Collaborative filtering recommending method and system based on grading Download PDF

Info

Publication number
CN102495864A
CN102495864A CN2011103820780A CN201110382078A CN102495864A CN 102495864 A CN102495864 A CN 102495864A CN 2011103820780 A CN2011103820780 A CN 2011103820780A CN 201110382078 A CN201110382078 A CN 201110382078A CN 102495864 A CN102495864 A CN 102495864A
Authority
CN
China
Prior art keywords
user
scoring
overbar
rightarrow
score data
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
Application number
CN2011103820780A
Other languages
Chinese (zh)
Inventor
谢峰
陈震
曹军威
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tsinghua University
Original Assignee
Tsinghua University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Tsinghua University filed Critical Tsinghua University
Priority to CN2011103820780A priority Critical patent/CN102495864A/en
Publication of CN102495864A publication Critical patent/CN102495864A/en
Pending legal-status Critical Current

Links

Images

Abstract

The invention discloses a collaborative filtering recommending method and system based on grading, which relates to the technical field of collaborative filtering recommendation. In the method, the similarity among users is calculated by using the statistical characteristics of history grading data of the users, and a project which is not evaluated by a current user is calculated through other users which have high similarity with the current user, so that the problem of incapability of calculating similarity or inaccurate similarity caused by difficulty in finding a common grading item among the users is solved, and accurate and rapid project recommendation can be realized.

Description

Collaborative filtering recommending method and system based on scoring
Technical field
The present invention relates to the collaborative filtering recommending technical field, particularly a kind of collaborative filtering recommending method and system based on scoring.
Background technology
The epoch of information explosion have been brought us in the fast development of Internet technology; Appear in the time of magnanimity information; Not only make the user be difficult to therefrom find to control oneself interested content, and make a large amount of information of knowing for the people less become " the dark information " in the network, can't be obtained by the general user.Commending system utilizes historical selection information of user or the potential hobby of similarity relation excavation user, and then recommends through setting up the binary relation between user and the project (for example: product, film, music, program etc.).
Had many classical commending systems at present, the collaborative filtering recommending system is the commending system that is suggested the earliest and is used widely.The score data that its core concept just is based on the similar nearest-neighbors of scoring produces recommendation to the targeted customer.Because nearest-neighbors is closely similar to scoring and the targeted customer of project, so the targeted customer can approach through the weighted mean value that nearest-neighbors is marked to this project the scoring of scoring item not.Typestry is the commending system based on collaborative filtering that puts forward the earliest, and the targeted customer it may be noted that and relatively more similar other users of own hobby.GroupLens is based on the automation collaborative filtered recommendation system of user's scoring, is used for film and news and recommends.Other systems that utilize collaborative filtering method to recommend also have the books commending system of Amazon.com, joke commending system of Jester or the like.
Compare with general commending system, the collaborative filtering recommending system has two big advantages: the one, there is not special requirement to recommending object, and can handle music, film etc. and be difficult to carry out the object that text structureization is represented; The 2nd, have the ability of recommending fresh information, can find the user potential but the interest preference oneself do not discovered as yet.
Traditional collaborative filtering recommending system utilizes common scoring item calculating similarity between different user, and the similarity calculation method of main flow comprises: cosine similarity method and relevant similarity method; The targeted customer predicts the weighted mean value of the scoring of project through the bigger neighbours of similarity the scoring of scoring item not.Can find out that the recommendation precision of collaborative filtering recommending system depends on the accuracy that similarity is calculated between the user.Yet in the huge network system of user and the number of entry, under the extremely sparse situation of user's score data, be difficult to find common scoring item between the user, thereby cause between the user similarity result of calculation inaccurate even can't calculate similarity.When the collaborative filtering recommending system has obtained widespread use, also be faced with a lot of problems, for example how new user being recommended or how to recommend new product is the cold start-up problem to the user, the sparse property problem of marking, algorithm scalability problem etc.In addition, traditional collaborative filtering recommending algorithm is along with number of users increases, the linear increasing of calculated amount, real-time performance worse and worse, response speed is also more and more slower simultaneously.
Summary of the invention
The technical matters that (one) will solve
The technical matters that the present invention will solve is: how in the huge collaborative filtering recommending system of user and the number of entry under the extremely sparse situation of user's score data, solve and be difficult to find between the user common scoring and cause calculating similarity or the inaccurate problem of similarity.
(2) technical scheme
For solving the problems of the technologies described above, the invention provides a kind of collaborative filtering recommending method based on scoring, may further comprise the steps:
S1: travel through all users in the current network system, obtain all users' historical score data;
S2: the statistical nature according to all users historical score data is separately confirmed the similarity degree between each user;
S3: select k other user the highest with active user's similarity degree, according to said k other user to active user's historical score data of scoring item not, come to the active user not scoring item predict;
S4: predicting the outcome of each user screened, produce the recommended project to each user.
Preferably, the similarity degree among the step S2 between each user calculates through formula,
sim ( u , v ) = cos ( u → , v → ) = u → · v → | | u → | | * | | v → | | = r ‾ u r ‾ v + S u S v + R u R v r ‾ u 2 + S u 2 + R u 2 · r ‾ v 2 + S v 2 + R v 2
Wherein, sim (u v) is the similarity degree between user u and the user v,
Figure BDA0000112545350000032
Figure BDA0000112545350000033
Figure BDA0000112545350000034
With
Figure BDA0000112545350000035
Be the tri-vector of respective user u and user v successively,
Figure BDA0000112545350000036
S uAnd R uBe average, variance and the extreme difference of the historical score data of respective user u successively,
Figure BDA0000112545350000037
S vAnd R vBe average, variance and the extreme difference of the historical score data of respective user v successively.
Preferably, among the step S3 through following formula come to the active user not scoring item predict,
P u , i = r ‾ u = 1 Σ x ∈ S ( u ) sim ( u , x ) Σ x ∈ S ( u ) sim ( u , x ) ( r x , i - r ‾ x )
Wherein, P U, iBe the prediction mark of the not scoring item i of active user u, S (u) was for carrying out user's set of scoring, r to the not scoring item i of active user u among k other user X, iBe the scoring of user x to the not scoring item i of active user u, user x is certain element among the S (u).
Preferably, when among the step S4 predicting the outcome of each user being screened, compare through average, if greater than average, then as the recommended project with the prediction mark of active user's not scoring item and active user's historical score data.
The invention also discloses a kind of collaborative filtering recommending system, comprising based on scoring:
The historical data computing module is used for traveling through all users of current network system, obtains all users' historical score data;
The similarity degree computing module is used for confirming the similarity degree between each user according to the statistical nature of all users historical score data separately;
Prediction module is used to select k other user the highest with active user's similarity degree, according to said k other user to active user's historical score data of scoring item not, come to the active user not scoring item predict;
Screening module is used for predicting the outcome of each user screened, and produces the recommended project to each user.
Preferably, the similarity degree in the similarity degree computing module between each user calculates through formula,
sim ( u , v ) = cos ( u → , v → ) = u → · v → | | u → | | * | | v → | | = r ‾ u r ‾ v + S u S v + R u R v r ‾ u 2 + S u 2 + R u 2 · r ‾ v 2 + S v 2 + R v 2
Wherein, sim (u v) is the similarity degree between user u and the user v,
Figure BDA0000112545350000043
With Be the tri-vector of respective user u and user v successively,
Figure BDA0000112545350000046
S uAnd R uBe average, variance and the extreme difference of the historical score data of respective user u successively,
Figure BDA0000112545350000047
S vAnd R vBe average, variance and the extreme difference of the historical score data of respective user v successively.
Preferably, in the prediction module through following formula to the active user not scoring item predict,
P u , i = r ‾ u = 1 Σ x ∈ S ( u ) sim ( u , x ) Σ x ∈ S ( u ) sim ( u , x ) ( r x , i - r ‾ x )
Wherein, P U, iBe the prediction mark of the not scoring item i of active user u, S (u) was for carrying out user's set of scoring, r to the not scoring item i of active user u among k other user X, iBe the scoring of user x to the not scoring item i of active user u, user x is certain element among the S (u).
Preferably, when in the screening module predicting the outcome of each user being screened, compare through average, if greater than average, then as the recommended project with the prediction mark of active user's not scoring item and active user's historical score data.
(3) beneficial effect
The present invention utilizes the statistical nature of the historical score data of user to calculate the similarity degree between each user; Through calculating the unvalued project of active user with other higher users of active user's similarity degree; Realized in the huge collaborative filtering recommending system of user and the number of entry under the extremely sparse situation of user's score data; Solved and be difficult to find between the user common scoring and cause to calculate similarity or the inaccurate problem of similarity, can realize accurately and project recommendation fast.
Description of drawings
Fig. 1 is the process flow diagram based on the collaborative filtering recommending method of marking according to one embodiment of the present invention;
Fig. 2 only comprises 4 users in the current network system in an embodiment of the present invention, and these 4 users are to the synoptic diagram of the score data of 10 projects;
Fig. 3 is the synoptic diagram that process is calculated the statistic sign matrix of back acquisition in an embodiment of the present invention;
Fig. 4 is the synoptic diagram of similarity degree between each user in an embodiment of the present invention;
Fig. 5 is user's prediction mark of scoring item not in an embodiment of the present invention, and the project recommendation result that obtains of each user;
Fig. 6 be in an embodiment of the present invention with " MovieLens 100K " as data set; Select 80% at random as training set; Remaining 20% as test set, adopts the result of the mean absolute error of method of the present invention and traditional collaborative filtering method with said k other number of users variation respectively;
Fig. 7 is as data set with " MovieLens 100K "; Select 80% at random as training set; Remaining 20% as test set, adopts the result of the root mean square absolute error of method of the present invention and traditional collaborative filtering method with said k other number of users variation respectively.
Embodiment
Below in conjunction with accompanying drawing and embodiment, specific embodiments of the invention describes in further detail.Following examples are used to explain the present invention, but are not used for limiting scope of the present invention.
Fig. 1 is the process flow diagram based on the collaborative filtering recommending method of marking according to one embodiment of the present invention, and with reference to Fig. 1, the method for this embodiment may further comprise the steps:
S1: travel through all users in the current network system, obtain all users' historical score data;
S2: the statistical nature according to all users historical score data is separately confirmed the similarity degree between each user; In this embodiment, make up each user's tri-vector earlier, calculate the similarity degree between each user according to each user's tri-vector through the included angle cosine value again; With the included angle cosine value between each user characteristics vector as each other similarity degree.Two vector angles are more little, and cosine value is big more, show that the user is similar more;
S3: select k other user the highest with active user's similarity degree, according to said k other user to active user's historical score data of scoring item not, come to the active user not scoring item predict;
S4: predicting the outcome of each user screened, produce the recommended project to each user.
Preferably, the similarity degree among the step S2 between each user calculates through formula,
sim ( u , v ) = cos ( u → , v → ) = u → · v → | | u → | | * | | v → | | = r ‾ u r ‾ v + S u S v + R u R v r ‾ u 2 + S u 2 + R u 2 · r ‾ v 2 + S v 2 + R v 2
Wherein, sim (u v) is the similarity degree between user u and the user v,
Figure BDA0000112545350000062
Figure BDA0000112545350000063
With
Figure BDA0000112545350000065
Be the tri-vector of respective user u and user v successively, S uAnd R uBe average, variance and the extreme difference of the historical score data of respective user u successively,
Figure BDA0000112545350000067
S vAnd R vBe average, variance and the extreme difference of the historical score data of respective user v successively.For example, (
Figure BDA0000112545350000068
S u, R u) statistical nature of historical score data of expression user u, mean value is represented the average score hobby of user to project, and variance is represented fluctuation or the deviation of user to the project scoring, and extreme difference is represented the scope hobby of user to scoring.Explain that to a certain extent these three values will influence the user to the not scoring in future of scoring item.
Wherein, mean value, variance and the extreme difference of all users historical score data separately calculates through formula,
r ‾ u = Σ i ∈ I ( u ) r u , i n
S u = 1 size ( I ( u ) ) Σ i ∈ I ( u ) ( r u , i - r ‾ u ) 2
R u=max(r u,i)-min(r u,i)
Wherein,
Figure BDA00001125453500000611
S u, R uAverage, variance and the extreme difference of representing the historical score data of user u respectively, I (u) are all scoring item set of user u, and size (I (u)) is a user u scoring item sum, r U, iBe the scoring of user u to project i, max (r U, i) and min (r U, i) represent user u scoring maximal value and minimum value respectively.
Preferably, among the step S3 through following formula come to the active user not scoring item predict,
P u , i = r ‾ u = 1 Σ x ∈ S ( u ) sim ( u , x ) Σ x ∈ S ( u ) sim ( u , x ) ( r x , i - r ‾ x )
Wherein, P U, iBe the prediction mark of the not scoring item i of active user u, S (u) was for carrying out user's set of scoring, r to the not scoring item i of active user u among k other user X, iBe the scoring of user x to the not scoring item i of active user u, user x is certain element among the S (u).
Preferably, when among the step S4 predicting the outcome of each user being screened, compare through average, if greater than average, then as the recommended project with the prediction mark of active user's not scoring item and active user's historical score data.
Embodiment 1
With a simple example above-mentioned algorithm recommendation process is described below.With reference to Fig. 2, this example comprises the scoring of 4 users to 10 projects, each user 5 projects of marking wherein, and we need be to 5 projects that the targeted customer does not mark prediction of marking, and makes corresponding recommendation.Step-by-step procedures is following:
1, calculates the statistical nature of the historical score data of each user,, obtain statistic as shown in Figure 3 and characterize matrix, promptly describe user characteristics with tri-vector like average, variance and extreme difference.
2, utilize the cosine similarity method to calculate the cosine angle value between each vector, obtain user's similarity matrix, with reference to Fig. 4, the value of similarity degree is big more, explains that similarity degree is high more between the user, and it is similar more to mark, like user u in the example 1With user u 2Similarity degree is the highest, and user u 4With user u 3Similarity degree is the highest.
3, owing to number of users in the example is less; Here we select the nearest-neighbors of 3 users that similarity degree is the highest as the targeted customer; And to the targeted customer not scoring item carry out the similarity weighting, obtain scoring prediction matrix shown in Figure 5, according to prediction score value size scoring item is not sorted; Select the prediction score value to produce recommendation greater than the project of targeted customer's average score, the project of Fig. 5 mid-score overstriking is final recommendation results.
Fig. 6 be in an embodiment of the present invention with " MovieLens 100K " as data set; Select 80% at random as training set; Remaining 20% as test set, adopts the result of the mean absolute error of method of the present invention and traditional collaborative filtering method with said k other number of users variation respectively; Fig. 7 is as data set with " MovieLens 100K "; Select 80% at random as training set; Remaining 20% as test set, adopts the result of the root mean square absolute error of method of the present invention and traditional collaborative filtering method with said k other number of users variation respectively.。
Collaborative filtering recommending method based on scoring proposed by the invention, its advantage is following:
(1) utilizes the statistic of user's score information to calculate similarity between the user, can effectively overcome and be difficult to find between the user common scoring under the sparse situation of scoring and cause to calculate similarity or similarity is calculated inaccurate shortcoming;
(2) to each user, all corresponding unique one group of statistic (average, variance, extreme difference), the similarity between the different user are calculated and all in three dimensions, are carried out, and have avoided between different user common scoring item to count the dimension difference that difference causes;
Faster than the searching of common scoring item, this algorithm advisory speed is faster more than 10 times than traditional collaborative filtering recommending algorithm far away in the calculating of (3) data statistics amount; Improved about 10% recommendation precision simultaneously, made the present invention in actual e-commerce system is used, obtain more performance.
The invention also discloses a kind of collaborative filtering recommending system, it is characterized in that, comprising based on scoring:
The historical data statistical module is used for traveling through all users of current network system, obtains all users' historical score data;
The similarity degree computing module is used for confirming the similarity degree between each user according to the statistical nature of all users historical score data separately;
Prediction module is used to select k other user the highest with active user's similarity degree, according to said k other user to active user's historical score data of scoring item not, come to the active user not scoring item predict;
Screening module is used for predicting the outcome of each user screened, and produces the recommended project to each user.
Preferably, the similarity degree in the similarity degree computing module between each user calculates through formula,
sim ( u , v ) = cos ( u → , v → ) = u → · v → | | u → | | * | | v → | | = r ‾ u r ‾ v + S u S v + R u R v r ‾ u 2 + S u 2 + R u 2 · r ‾ v 2 + S v 2 + R v 2
Wherein, sim (u v) is the similarity degree between user u and the user v,
Figure BDA0000112545350000092
Figure BDA0000112545350000093
Figure BDA0000112545350000094
With
Figure BDA0000112545350000095
Be the tri-vector of respective user u and user v successively, S uAnd R uBe average, variance and the extreme difference of the historical score data of respective user u successively,
Figure BDA0000112545350000097
S vAnd R vBe average, variance and the extreme difference of the historical score data of respective user v successively.
Preferably, in the prediction module through following formula to the active user not scoring item predict,
P u , i = r ‾ u = 1 Σ x ∈ S ( u ) sim ( u , x ) Σ x ∈ S ( u ) sim ( u , x ) ( r x , i - r ‾ x )
Wherein, P U, iBe the prediction mark of the not scoring item i of active user u, S (u) was for carrying out user's set of scoring, r to the not scoring item i of active user u among k other user X, iBe the scoring of user x to the not scoring item i of active user u, user x is certain element among the S (u).
Preferably, when in the screening module predicting the outcome of each user being screened, compare through average, if greater than average, then as the recommended project with the prediction mark of active user's not scoring item and active user's historical score data.
Above embodiment only is used to explain the present invention; And be not limitation of the present invention; The those of ordinary skill in relevant technologies field under the situation that does not break away from the spirit and scope of the present invention, can also be made various variations and modification; Therefore all technical schemes that are equal to also belong to category of the present invention, and scope of patent protection of the present invention should be defined by the claims.

Claims (8)

1. the collaborative filtering recommending method based on scoring is characterized in that, may further comprise the steps:
S1: travel through all users in the current network system, obtain all users' historical score data;
S2: the statistical nature according to all users historical score data is separately confirmed the similarity degree between each user;
S3: select k other user the highest with active user's similarity degree, according to said k other user to active user's historical score data of scoring item not, come to the active user not scoring item predict;
S4: predicting the outcome of each user screened, produce the recommended project to each user.
2. the method for claim 1 is characterized in that, the similarity degree among the step S2 between each user calculates through formula,
sim ( u , v ) = cos ( u → , v → ) = u → · v → | | u → | | * | | v → | | = r ‾ u r ‾ v + S u S v + R u R v r ‾ u 2 + S u 2 + R u 2 · r ‾ v 2 + S v 2 + R v 2
Wherein, sim (u v) is the similarity degree between user u and the user v,
Figure FDA0000112545340000012
Figure FDA0000112545340000013
Figure FDA0000112545340000014
With
Figure FDA0000112545340000015
Be the tri-vector of respective user u and user v successively,
Figure FDA0000112545340000016
S uAnd R uBe average, variance and the extreme difference of the historical score data of respective user u successively, S vAnd R vBe average, variance and the extreme difference of the historical score data of respective user v successively.
3. method as claimed in claim 2 is characterized in that, among the step S3 through following formula come to the active user not scoring item predict,
P u , i = r ‾ u = 1 Σ x ∈ S ( u ) sim ( u , x ) Σ x ∈ S ( u ) sim ( u , x ) ( r x , i - r ‾ x )
Wherein, P U, iBe the prediction mark of the not scoring item i of active user u, S (u) was for carrying out user's set of scoring, r to the not scoring item i of active user u among k other user X, iBe the scoring of user x to the not scoring item i of active user u, user x is certain element among the S (u).
4. like each described method in the claim 1~3; It is characterized in that; When among the step S4 predicting the outcome of each user being screened; Average through with the prediction mark of active user's not scoring item and active user's historical score data compares, if greater than average, then as the recommended project.
5. the collaborative filtering recommending system based on scoring is characterized in that, comprising:
The historical data statistical module is used for traveling through all users of current network system, obtains all users' historical score data;
The similarity degree computing module is used for confirming the similarity degree between each user according to the statistical nature of all users historical score data separately;
Prediction module is used to select k other user the highest with active user's similarity degree, according to said k other user to active user's historical score data of scoring item not, come to the active user not scoring item predict;
Screening module is used for predicting the outcome of each user screened, and produces the recommended project to each user.
6. system as claimed in claim 5 is characterized in that the similarity degree in the similarity degree computing module between each user calculates through formula,
sim ( u , v ) = cos ( u → , v → ) = u → · v → | | u → | | * | | v → | | = r ‾ u r ‾ v + S u S v + R u R v r ‾ u 2 + S u 2 + R u 2 · r ‾ v 2 + S v 2 + R v 2
Wherein, sim (u v) is the similarity degree between user u and the user v,
Figure FDA0000112545340000022
Figure FDA0000112545340000023
With Be the tri-vector of respective user u and user v successively,
Figure FDA0000112545340000026
S uAnd R uBe average, variance and the extreme difference of the historical score data of respective user u successively,
Figure FDA0000112545340000027
S vAnd R vBe average, variance and the extreme difference of the historical score data of respective user v successively.
7. system as claimed in claim 6 is characterized in that, in the prediction module through following formula to the active user not scoring item predict,
P u , i = r ‾ u = 1 Σ x ∈ S ( u ) sim ( u , x ) Σ x ∈ S ( u ) sim ( u , x ) ( r x , i - r ‾ x )
Wherein, P U, iBe the prediction mark of the not scoring item i of active user u, S (u) was for carrying out user's set of scoring, r to the not scoring item i of active user u among k other user X, iBe the scoring of user x to the not scoring item i of active user u, user x is certain element among the S (u).
8. like each described system in the claim 5~7; It is characterized in that; When in the screening module predicting the outcome of each user being screened; Average through with the prediction mark of active user's not scoring item and active user's historical score data compares, if greater than average, then as the recommended project.
CN2011103820780A 2011-11-25 2011-11-25 Collaborative filtering recommending method and system based on grading Pending CN102495864A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN2011103820780A CN102495864A (en) 2011-11-25 2011-11-25 Collaborative filtering recommending method and system based on grading

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN2011103820780A CN102495864A (en) 2011-11-25 2011-11-25 Collaborative filtering recommending method and system based on grading

Publications (1)

Publication Number Publication Date
CN102495864A true CN102495864A (en) 2012-06-13

Family

ID=46187689

Family Applications (1)

Application Number Title Priority Date Filing Date
CN2011103820780A Pending CN102495864A (en) 2011-11-25 2011-11-25 Collaborative filtering recommending method and system based on grading

Country Status (1)

Country Link
CN (1) CN102495864A (en)

Cited By (25)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102841929A (en) * 2012-07-19 2012-12-26 南京邮电大学 Recommending method integrating user and project rating and characteristic factors
CN102968506A (en) * 2012-12-14 2013-03-13 北京理工大学 Personalized collaborative filtering recommendation method based on extension characteristic vectors
CN103279677A (en) * 2013-06-07 2013-09-04 华东师范大学 Predicted value correction method based on collaborative filtering
CN103383702A (en) * 2013-07-17 2013-11-06 中国科学院深圳先进技术研究院 Method and system for recommending personalized news based on ranking of votes of users
CN103514255A (en) * 2013-07-11 2014-01-15 江苏谐云智能科技有限公司 Method for collaborative filtering recommendation based on item level types
CN103530416A (en) * 2013-10-28 2014-01-22 海南大学 Project data forecasting grading library generating and project data pushing method and project data forecasting grading library generating and project data pushing system
CN103544625A (en) * 2012-07-10 2014-01-29 百度在线网络技术(北京)有限公司 Method and system for judging application similarity according to massive data
CN103745100A (en) * 2013-12-27 2014-04-23 浙江大学 Item-based explicit and implicit feedback mixing collaborative filtering recommendation algorithm
CN103793505A (en) * 2014-01-27 2014-05-14 西安理工大学 Network service collaborative filtering method based on user-service characteristics
CN104063481A (en) * 2014-07-02 2014-09-24 山东大学 Film individuation recommendation method based on user real-time interest vectors
CN104077351A (en) * 2014-05-26 2014-10-01 东北师范大学 Heterogeneous information network based content providing method and system
CN104111938A (en) * 2013-04-18 2014-10-22 腾讯科技(深圳)有限公司 Information recommendation method and device
CN104166732A (en) * 2014-08-29 2014-11-26 合肥工业大学 Project collaboration filtering recommendation method based on global scoring information
CN104239390A (en) * 2014-06-11 2014-12-24 杭州联汇数字科技有限公司 Audio recommending method on basis of improved collaborative filtering algorithm
CN104462597A (en) * 2014-12-31 2015-03-25 湖南大学 Comprehensive user positive and negative grading and grading preference factor collaborative filtering algorithm
CN104899236A (en) * 2014-11-13 2015-09-09 深圳市腾讯计算机系统有限公司 Comment information display method, comment information display device and comment information display system
CN105260460A (en) * 2015-10-16 2016-01-20 桂林电子科技大学 Diversity-oriented recommendation method
CN105408929A (en) * 2013-03-08 2016-03-16 Seeon有限公司 Method and apparatus for recommending affiliated store by using reverse auction
CN105894254A (en) * 2016-06-28 2016-08-24 上海理工大学 Social recommendation system based on degree distribution and user rating
CN107368540A (en) * 2017-06-26 2017-11-21 北京理工大学 The film that multi-model based on user's self-similarity is combined recommends method
CN108415926A (en) * 2018-01-15 2018-08-17 大连理工大学 A kind of collaborative filtering recommending method for eliminating original score data scoring noise
CN108830460A (en) * 2018-05-23 2018-11-16 重庆邮电大学 A method of recommender system Deta sparseness is alleviated based on substep dynamic filling
CN109670121A (en) * 2018-12-18 2019-04-23 辽宁工程技术大学 Project level and feature level depth Collaborative Filtering Recommendation Algorithm based on attention mechanism
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
WO2021213546A1 (en) * 2020-05-29 2021-10-28 青岛海尔电冰箱有限公司 Coal gas concentration prediction method based on collaborative filtering, and device and refrigerator

Cited By (41)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103544625A (en) * 2012-07-10 2014-01-29 百度在线网络技术(北京)有限公司 Method and system for judging application similarity according to massive data
CN102841929A (en) * 2012-07-19 2012-12-26 南京邮电大学 Recommending method integrating user and project rating and characteristic factors
CN102968506A (en) * 2012-12-14 2013-03-13 北京理工大学 Personalized collaborative filtering recommendation method based on extension characteristic vectors
CN105408929A (en) * 2013-03-08 2016-03-16 Seeon有限公司 Method and apparatus for recommending affiliated store by using reverse auction
CN104111938B (en) * 2013-04-18 2018-09-18 腾讯科技(深圳)有限公司 A kind of method and device of information recommendation
CN104111938A (en) * 2013-04-18 2014-10-22 腾讯科技(深圳)有限公司 Information recommendation method and device
CN103279677A (en) * 2013-06-07 2013-09-04 华东师范大学 Predicted value correction method based on collaborative filtering
CN103279677B (en) * 2013-06-07 2016-08-10 华东师范大学 A kind of predictive value bearing calibration based on collaborative filtering
CN103514255B (en) * 2013-07-11 2017-04-05 江苏谐云智能科技有限公司 A kind of collaborative filtering recommending method based on project stratigraphic classification
CN103514255A (en) * 2013-07-11 2014-01-15 江苏谐云智能科技有限公司 Method for collaborative filtering recommendation based on item level types
CN103383702A (en) * 2013-07-17 2013-11-06 中国科学院深圳先进技术研究院 Method and system for recommending personalized news based on ranking of votes of users
CN103530416A (en) * 2013-10-28 2014-01-22 海南大学 Project data forecasting grading library generating and project data pushing method and project data forecasting grading library generating and project data pushing system
CN103530416B (en) * 2013-10-28 2017-01-18 海南大学 Project data forecasting grading library generating and project data pushing method and project data forecasting grading library generating and project data pushing system
CN103745100A (en) * 2013-12-27 2014-04-23 浙江大学 Item-based explicit and implicit feedback mixing collaborative filtering recommendation algorithm
CN103745100B (en) * 2013-12-27 2017-01-04 浙江大学 A kind of method of the collaborative filtering recommending of the dominant explicit feedback of project-based mixing
CN103793505A (en) * 2014-01-27 2014-05-14 西安理工大学 Network service collaborative filtering method based on user-service characteristics
CN103793505B (en) * 2014-01-27 2017-02-15 西安理工大学 Network service collaborative filtering method based on user-service characteristics
CN104077351A (en) * 2014-05-26 2014-10-01 东北师范大学 Heterogeneous information network based content providing method and system
CN104077351B (en) * 2014-05-26 2017-01-25 东北师范大学 Heterogeneous information network based content providing method and system
CN104239390B (en) * 2014-06-11 2017-12-29 杭州联汇科技股份有限公司 A kind of audio based on modified collaborative filtering recommends method
CN104239390A (en) * 2014-06-11 2014-12-24 杭州联汇数字科技有限公司 Audio recommending method on basis of improved collaborative filtering algorithm
CN104063481A (en) * 2014-07-02 2014-09-24 山东大学 Film individuation recommendation method based on user real-time interest vectors
CN104063481B (en) * 2014-07-02 2017-11-14 山东大学 A kind of film personalized recommendation method based on the real-time interest vector of user
CN104166732A (en) * 2014-08-29 2014-11-26 合肥工业大学 Project collaboration filtering recommendation method based on global scoring information
CN104166732B (en) * 2014-08-29 2017-04-12 合肥工业大学 Project collaboration filtering recommendation method based on global scoring information
CN104899236A (en) * 2014-11-13 2015-09-09 深圳市腾讯计算机系统有限公司 Comment information display method, comment information display device and comment information display system
CN104899236B (en) * 2014-11-13 2019-01-29 深圳市腾讯计算机系统有限公司 A kind of comment information display methods, apparatus and system
CN104462597A (en) * 2014-12-31 2015-03-25 湖南大学 Comprehensive user positive and negative grading and grading preference factor collaborative filtering algorithm
CN104462597B (en) * 2014-12-31 2018-04-03 湖南大学 A kind of positive negativity of synthetic user scores and the collaborative filtering method of scoring preference heterogeneity
CN105260460B (en) * 2015-10-16 2018-08-14 桂林电子科技大学 One kind is towards multifarious recommendation method
CN105260460A (en) * 2015-10-16 2016-01-20 桂林电子科技大学 Diversity-oriented recommendation method
CN105894254A (en) * 2016-06-28 2016-08-24 上海理工大学 Social recommendation system based on degree distribution and user rating
CN107368540A (en) * 2017-06-26 2017-11-21 北京理工大学 The film that multi-model based on user's self-similarity is combined recommends method
CN108415926A (en) * 2018-01-15 2018-08-17 大连理工大学 A kind of collaborative filtering recommending method for eliminating original score data scoring noise
CN108415926B (en) * 2018-01-15 2021-08-10 大连理工大学 Collaborative filtering recommendation method for eliminating scoring noise of original scoring data
CN108830460A (en) * 2018-05-23 2018-11-16 重庆邮电大学 A method of recommender system Deta sparseness is alleviated based on substep dynamic filling
CN108830460B (en) * 2018-05-23 2021-12-17 重庆邮电大学 Method for relieving data sparsity of recommendation system based on step-by-step dynamic filling
CN109670121A (en) * 2018-12-18 2019-04-23 辽宁工程技术大学 Project level and feature level depth Collaborative Filtering Recommendation Algorithm based on attention mechanism
WO2021213546A1 (en) * 2020-05-29 2021-10-28 青岛海尔电冰箱有限公司 Coal gas concentration prediction method based on collaborative filtering, and device and refrigerator
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
CN113254777B (en) * 2021-06-07 2021-09-24 武汉卓尔数字传媒科技有限公司 Information recommendation method and device, electronic equipment and storage medium

Similar Documents

Publication Publication Date Title
CN102495864A (en) Collaborative filtering recommending method and system based on grading
US20220292527A1 (en) Methods of assessing long-term indicators of sentiment
Kim et al. Mobile application service networks: Apple’s App Store
CN105404700A (en) Collaborative filtering-based video program recommendation system and recommendation method
CN103995839A (en) Commodity recommendation optimizing method and system based on collaborative filtering
CN106251174A (en) Information recommendation method and device
CN104063481A (en) Film individuation recommendation method based on user real-time interest vectors
CN103473128A (en) Collaborative filtering method for mashup application recommendation
CN105893609A (en) Mobile APP recommendation method based on weighted mixing
CN103678431A (en) Recommendation method based on standard labels and item grades
CN103309967A (en) Collaborative filtering method and system based on similarity propagation
CN103399858A (en) Socialization collaborative filtering recommendation method based on trust
CN102982107A (en) Recommendation system optimization method with information of user and item and context attribute integrated
CN101482884A (en) Cooperation recommending system based on user predilection grade distribution
Weiß et al. Binomial AR (1) processes: moments, cumulants, and estimation
CN107194430A (en) A kind of screening sample method and device, electronic equipment
Liu et al. Information filtering via weighted heat conduction algorithm
CN106168980A (en) Multimedia resource recommends sort method and device
CN104331459A (en) Online learning-based network resource recommendation method and device
CN105260390A (en) Group-oriented project recommendation method based on joint probability matrix decomposition
Wang et al. Functional bid landscape forecasting for display advertising
CN105447193A (en) Music recommending system based on machine learning and collaborative filtering
CN103530416A (en) Project data forecasting grading library generating and project data pushing method and project data forecasting grading library generating and project data pushing system
CN104361062A (en) Associated information recommendation method and device
CN104166732A (en) Project collaboration filtering recommendation method based on global scoring information

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C12 Rejection of a patent application after its publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20120613