CN106951462B - Movie recommendation method based on Time-Trust similarity - Google Patents

Movie recommendation method based on Time-Trust similarity Download PDF

Info

Publication number
CN106951462B
CN106951462B CN201710106152.3A CN201710106152A CN106951462B CN 106951462 B CN106951462 B CN 106951462B CN 201710106152 A CN201710106152 A CN 201710106152A CN 106951462 B CN106951462 B CN 106951462B
Authority
CN
China
Prior art keywords
movie
movies
similarity
user
recommendation method
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.)
Expired - Fee Related
Application number
CN201710106152.3A
Other languages
Chinese (zh)
Other versions
CN106951462A (en
Inventor
钟俊
李德新
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.)
Sichuan University
Original Assignee
Sichuan 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 Sichuan University filed Critical Sichuan University
Priority to CN201710106152.3A priority Critical patent/CN106951462B/en
Publication of CN106951462A publication Critical patent/CN106951462A/en
Application granted granted Critical
Publication of CN106951462B publication Critical patent/CN106951462B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/95—Retrieval from the web
    • G06F16/953—Querying, e.g. by the use of web search engines
    • G06F16/9535—Search customisation based on user profiles and personalisation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/22—Matching criteria, e.g. proximity measures

Abstract

The invention provides a movie recommendation method based on Time-Trust similarity, which utilizes the scoring information of movies, integrates the characteristics of forgetting rule, score among movies, interest tendency of a user to the movies and the like, calculates the similarity degree among the movies, thereby carrying out scoring prediction on the movies and carrying out movie recommendation. The method and the system can realize accurate prediction of movie evaluation and recommend movies to the user according to the user interests.

Description

Movie recommendation method based on Time-Trust similarity
Technical Field
The invention relates to a movie recommendation method, in particular to a movie recommendation method based on Time-Trust similarity.
Background
The similarity measurement method based on the film mainly adopts a cosine similarity method, a Jaccard similarity method and the like. The conventional similarity measurement method has some drawbacks. The cosine similarity method does not take into account the number of users that are commonly scored for the movie. The criteria for measuring similarity is generally that the closer the two movies score the same user, the greater the number of common scores, and the higher the similarity. However, if there are only a few or even only 1 common scoring user among movies, the similarity obtained by using cosine similarity is quite high, which often results in a violation of the conventional rule. The Jaccard similarity considers the problem of the number of users scoring together, however, the similarity is measured by simply using the intersection union ratio of two objects, the actual scoring value cannot be considered, and the defect is still obvious.
Disclosure of Invention
Aiming at the defects of the prior art, the invention aims to provide a movie recommendation method based on Time-Trust similarity, which is characterized by comprising the following steps:
(1) acquiring movie data; the movie data comprises the evaluation time difference delta t of the movie i and the movie j and the respective scores r of the movie i and the movie j of a single useriAnd rjAverage scores for movie i and movie j for all users
Figure BDA0001233093380000011
And
Figure BDA0001233093380000012
the highest score max in the two movies of movie I and movie j, the minimum score min in the two movies of movie I and movie j, and the respective score user sets I of movie I and movie jiAnd IjAnd a common user union I between two moviesi∪Ij；
The movie i is any movie, and the movie j is any movie different from the movie i;
(2) the similarity is calculated according to the following formula:
Figure BDA0001233093380000021
wherein the content of the first and second substances,
Figure BDA0001233093380000022
Figure BDA0001233093380000023
Figure BDA0001233093380000024
Figure BDA0001233093380000025
Figure BDA0001233093380000026
wherein the content of the first and second substances,
rememory＝0.01*31.8×(Δt)-0.125；
(3) storing the result data obtained in the step (2);
(4) and acquiring the historical movies of the user, outputting the movies with the highest similarity to the favorite movies, and recommending the movies to the user.
The similarity result obtained by the method has high accuracy, and can recommend high-quality movies to users. As shown in the embodiments of the present invention, the present invention has better accuracy in performing similarity-based recommendations than other methods of the prior art.
To further illustrate the steps and principles of the present invention, the above-described steps of the present invention will now be described in detail:
A. and calculating the memory residual amount rememory after the delta t by using the principle of Ebinghaos forgetting law and taking the evaluation time difference delta t between the films as one of the measurement indexes of the similarity.
rememory＝0.01*31.8×(Δt)-0.125
B. Combining the similarity of the movie evaluation time difference set as sim1(i, j), using the memory residual rememory obtained by the method in step a, and combining the sigmod function curve, a calculation expression corresponding to sim1(i, j) is obtained as follows:
Figure BDA0001233093380000031
C. the score similarity is presented using the user's scores for the two movies. Let a user's score for movie i be riScore of r for movie jjThe maximum score of the movies is max, the minimum score is min, and by combining a sigmod function derivative decay function curve, the score similarity sim2(i, j) between the movies can be obtained, wherein the formula is as follows:
Figure BDA0001233093380000041
D. and (4) providing the movie tendency similarity by using the scores of the two movies by the user. Let the user's score for movie i be riThe average of all users' scores for movie i is
Figure BDA0001233093380000042
Score of movie j as rjThe average of all users' scores for movie j is
Figure BDA0001233093380000043
The tendency similarity sim3(i, j) between available movies is given by:
Figure BDA0001233093380000044
E. a common user set weight factor is presented using the user's scores for the two movies. Let the scoring user set of movie I be IiThe set of scoring users for movie j is IjA union of common users I between two movies is availablei∪IjThe common user set weight factor sim4(i, j) takes the Jaccard similarity as a trust factor, and the formula is as follows:
Figure BDA0001233093380000045
F. common user set I using movie I and movie ji∩jAnd considering the common user set into the calculation of the new similarity to obtain a formula:
Figure BDA0001233093380000046
G. by integrating A, B, C, D, E, F, the overall movie-based similarity metric formula is:
Figure BDA0001233093380000051
the invention has the beneficial effects that:
the method and the system can realize accurate prediction of movie evaluation and recommend movies to the user according to the user interests.
Drawings
FIG. 1 is a diagram of a model of a motion picture based recommendation algorithm; the arrow connecting line between the user and the movie in the graph represents the association relationship of the user to the movie, and the value of the arrow connecting line represents the interest degree of the user in the movie; the arrow connecting line between the movies represents the association relationship between the movies, and the value of the arrow connecting line represents the similarity between the two movies; calculating the interest degree of the user to the unscored movies according to the interest degree of the user to the scored movies and the similarity between the movies, and sorting the interest degree of the user to the unscored movies in a descending order according to the interest degree value to generate a recommendation list;
FIG. 2 is a comparison chart of movie recommendation accuracy;
FIG. 3 shows a comparison of movie recommendation recall.
Detailed Description
The present invention is described in detail below by way of examples, and it should be noted that the following examples are only for illustrating the present invention and should not be construed as limiting the scope of the present invention.
Example 1
In this embodiment, the selected movies are Huobi, Feng-speeches, Pearl harbors, spirit anger, Zeus's, magic rings, and war horses. In the following steps, movie i and movie j are independently selected from the above-mentioned movies. The scores of the above movies were obtained from the bean net.
(1) Acquiring movie data; the movie data comprises the evaluation time difference delta t of the movie i and the movie j and the respective scores r of the movie i and the movie j of a single useriAnd rjAverage scores for movie i and movie j for all users
Figure BDA0001233093380000061
And
Figure BDA0001233093380000062
the highest score max in the two movies of movie I and movie j, the minimum score min in the two movies of movie I and movie j, and the respective score user sets I of movie I and movie jiAnd IjAnd a common user union I between two moviesi∪Ij；
The movie i is any movie, and the movie j is any movie different from the movie i;
(2) the similarity is calculated according to the following formula:
Figure BDA0001233093380000063
wherein the content of the first and second substances,
Figure BDA0001233093380000064
Figure BDA0001233093380000065
Figure BDA0001233093380000071
Figure BDA0001233093380000072
Figure BDA0001233093380000073
wherein the content of the first and second substances,
rememory＝0.01*31.8×(Δt)-0.125；
(3) storing the result data obtained in the step (2);
(4) and acquiring the historical movies of the user, outputting the movies with the highest similarity to the favorite movies, and recommending the movies to the user.
Defining the obtained similarity as wijAnd calculating the accuracy and the recall rate. The formula is as follows:
Figure BDA0001233093380000074
Figure BDA0001233093380000075
i refers to the movie set of the system, wijIs the similarity between the movies. R (i) is a recommendation list generated from the action of the movie on the training set, and T (i) is a list of actions of the movie on the test set.
As can be seen from FIG. 1, the user has the highest interest in Huobi people, and better recommendation effect can be obtained when the Hoobi people are recommended to the user. The interest in Yu Si Zi is the lowest.
The same calculation of accuracy and recall was performed using the conventional cosine similarity method and the Jaccard method as a comparison, and the results are shown in fig. 2 and 3.

Claims (3)

1. A movie recommendation method based on Time-Trust similarity is characterized by comprising the following steps:
(1) acquiring movie data; the movie data comprises the evaluation time difference delta t of the movie i and the movie j and the respective scores r of the movie i and the movie j of a single useriAnd rjAverage scores for movie i and movie j for all users
Figure FDA0002988681970000011
And
Figure FDA0002988681970000012
the highest score max in the two movies of movie I and movie j, the minimum score min in the two movies of movie I and movie j, and the respective score user sets I of movie I and movie jiAnd IjAnd a common user union I between two moviesi∪Ij；
The movie i is any movie, and the movie j is any movie different from the movie i;
(2) the similarity is calculated according to the following formula:
Figure FDA0002988681970000013
wherein the content of the first and second substances,
Figure FDA0002988681970000014
Figure FDA0002988681970000015
Figure FDA0002988681970000016
Figure FDA0002988681970000017
Figure FDA0002988681970000018
wherein the content of the first and second substances,
rememory＝0.01*31.8×(△t)-0.125(ii) a sim1(i, j) is the similarity of the time difference in combination with movie evaluation; sim2(i, j) is the fractional similarity between movies; sim3(i, j) is the tendency similarity degree between movies; sim4(i, j) is the common user set weight factor between movies; sim5(i, j) is a set of similarity that takes into account a common set of users;
(3) storing the result data obtained in the step (2);
(4) and acquiring the historical movies of the user, outputting the movies with the highest similarity to the favorite movies, and recommending the movies to the user.
2. The movie recommendation method according to claim 1, wherein said movie data is obtained from a movie website.
3. The movie recommendation method according to claim 1, wherein the movie preferred by the user is a movie with a high rating by the user.
CN201710106152.3A 2017-02-27 2017-02-27 Movie recommendation method based on Time-Trust similarity Expired - Fee Related CN106951462B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710106152.3A CN106951462B (en) 2017-02-27 2017-02-27 Movie recommendation method based on Time-Trust similarity

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710106152.3A CN106951462B (en) 2017-02-27 2017-02-27 Movie recommendation method based on Time-Trust similarity

Publications (2)

Publication Number Publication Date
CN106951462A CN106951462A (en) 2017-07-14
CN106951462B true CN106951462B (en) 2021-06-25

Family

ID=59467000

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710106152.3A Expired - Fee Related CN106951462B (en) 2017-02-27 2017-02-27 Movie recommendation method based on Time-Trust similarity

Country Status (1)

Country Link
CN (1) CN106951462B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108710648B (en) * 2018-04-28 2021-08-31 东华大学 Collaborative filtering recommendation method based on S-type improved similarity
CN108733784B (en) * 2018-05-09 2020-12-29 深圳市领点科技有限公司 Teaching courseware recommendation method, device and equipment

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104063481A (en) * 2014-07-02 2014-09-24 山东大学 Film individuation recommendation method based on user real-time interest vectors
CN104156472A (en) * 2014-08-25 2014-11-19 四达时代通讯网络技术有限公司 Video recommendation method and system
CN105718551A (en) * 2016-01-19 2016-06-29 浙江工业大学 Hadoop based social recommendation method

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9942609B2 (en) * 2014-11-13 2018-04-10 Comcast Cable Communications, Llc Personalized content recommendations based on consumption periodicity

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104063481A (en) * 2014-07-02 2014-09-24 山东大学 Film individuation recommendation method based on user real-time interest vectors
CN104156472A (en) * 2014-08-25 2014-11-19 四达时代通讯网络技术有限公司 Video recommendation method and system
CN105718551A (en) * 2016-01-19 2016-06-29 浙江工业大学 Hadoop based social recommendation method

Also Published As

Publication number Publication date
CN106951462A (en) 2017-07-14

Similar Documents

Publication Publication Date Title
US9075882B1 (en) Recommending content items
CN108829808B (en) Page personalized sorting method and device and electronic equipment
JP5746658B2 (en) Information processing apparatus, method and program, information communication terminal, control method thereof and control program thereof
CN112074857A (en) Combining machine learning and social data to generate personalized recommendations
US20130283303A1 (en) Apparatus and method for recommending content based on user's emotion
WO2021119119A1 (en) System and method for a personalized search and discovery engine
JP6261547B2 (en) Determination device, determination method, and determination program
KR20050043917A (en) Statistical personalized recommendation system
WO2014169776A1 (en) Cluster method and apparatus based on user interest
CN107301247B (en) Method and device for establishing click rate estimation model, terminal and storage medium
CN104615741B (en) Cold-start project recommendation method and device based on cloud computing
JP2019113943A (en) Information providing apparatus, information providing method, and program
US20180293312A1 (en) Computerized Method and System for Organizing Video Files
CN106951462B (en) Movie recommendation method based on Time-Trust similarity
KR101708254B1 (en) Story-based recommendation system and method for movies by character-net and collaborative filtering
KR100781399B1 (en) Apparatus and method for providing weights to recommendation engines according to situation of user and computer readable medium processing the method
CN110990717B (en) Interest point recommendation method based on cross-domain association
JP2007519326A (en) Content recommendation method and apparatus
CA3111094C (en) Noise contrastive estimation for collaborative filtering
CN107133811A (en) The recognition methods of targeted customer a kind of and device
CN108335131B (en) Method and device for estimating age bracket of user and electronic equipment
KR20170036874A (en) Method and apparatus for recommendation of social event based on users preference
CN106096029B (en) Recommendation method based on user bidirectional relationship
US20150170035A1 (en) Real time personalization and categorization of entities
CN107038169B (en) Object recommendation method and object recommendation device

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
GR01 Patent grant
GR01 Patent grant
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20210625

CF01 Termination of patent right due to non-payment of annual fee