CN104462385B - A kind of film personalization similarity calculating method based on user interest model - Google Patents
A kind of film personalization similarity calculating method based on user interest model Download PDFInfo
- Publication number
- CN104462385B CN104462385B CN201410753644.8A CN201410753644A CN104462385B CN 104462385 B CN104462385 B CN 104462385B CN 201410753644 A CN201410753644 A CN 201410753644A CN 104462385 B CN104462385 B CN 104462385B
- Authority
- CN
- China
- Prior art keywords
- user
- dimension
- movie
- similarity
- vector
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 21
- 230000006399 behavior Effects 0.000 claims abstract description 33
- 238000004458 analytical method Methods 0.000 claims abstract description 9
- 238000000605 extraction Methods 0.000 claims abstract description 5
- 238000004364 calculation method Methods 0.000 claims description 27
- 238000004422 calculation algorithm Methods 0.000 claims description 20
- 238000007619 statistical method Methods 0.000 claims description 7
- 238000011156 evaluation Methods 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 3
- 238000007418 data mining Methods 0.000 claims description 2
- 238000013507 mapping Methods 0.000 claims 2
- 230000000694 effects Effects 0.000 abstract description 2
- 239000000126 substance Substances 0.000 abstract 1
- 238000010586 diagram Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000005065 mining Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
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
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a kind of film personalization similarity calculating methods based on user interest model, this method is according to the historical behavior of user, i.e. by personalized film commending system platform, user is excavated to the various search behaviors in movie resource library and viewing and collection behavior；It fully excavates and different preference degrees of the analysis different user in the performer of film, director, type, area, time, brief introduction this six essential attributes represents to get to the first layer sextuple space vector of user model；According to the above-mentioned behavior of user, by keyword extraction or semantic analysis, weight of the analysis different user in above-mentioned six dimensions shared by each characteristic value represents to get to the second layer sextuple space vector of user model；The interest model of user is represented with the hyperspace vector of two layers, the substance feature of interest model and film based on user generates different film similarity lists for different user, so as to improve the effect of recommendation.
Description
Technical Field
The invention relates to a movie personalized similarity calculation method based on an old user interest model and a movie personalized similarity calculation method aiming at a new user.
Background
With the rapid development of the internet, in the face of increasingly updated massive movie resources, personalized recommendation applications are randomly generated.
The current major popular recommendation algorithms are: association rule based recommendations, knowledge based recommendations, content based recommendations, collaborative filtering recommendations, combination recommendations, and the like. The above recommendation algorithms all involve a key technology, namely: and calculating the similarity between the articles, and finding the nearest neighbor of the article according to the similarity. The traditional methods for calculating the similarity of the movies only relate to the content characteristics of the movies, and the influence of interest and preference of different users on the movies is not considered. According to the existing computing method, the characteristic of individuation can not be embodied, so that the user experience is difficult to satisfy.
In a real personalized recommendation system, a user interest model is a basis, and a personalized recommendation algorithm is a core. Personalized movie recommendations should take into account three sources of information: 1) a user information base, namely basic information of a user, including age, gender, occupation and the like; 2) a commodity information base, namely attribute information of the movie, including actors, director, genre, content introduction, region, release time and the like; 3) the user history information base, i.e. the history of the user's use, includes different preference degrees for actors, director, genre, region, release time, content introduction, etc. The information sources are very critical, and better recommendation can be achieved only by fully utilizing the information base to establish the user interest model.
Disclosure of Invention
Aiming at the technical problems in the prior art, the invention provides a movie personalized similarity calculation method based on an old user interest model.
In order to achieve the purpose, the invention adopts the following technical scheme:
a movie personalized similarity calculation method based on an old user interest model comprises the following steps:
s1, establishing a user dynamic behavior information base based on the user behavior data in a certain period of time T and the N movies with the highest evaluation in the film watching records in the period of time;
s2, performing data mining on the user dynamic behavior information base to obtain the preference of the user to each dimension of the movie and the preference of the user to the characteristic value of each dimension in the movie, and constructing a user interest model; wherein,
(1) the preference of the user for each dimension of the movie can be expressed as a six-tuple by using a space vector, namely:
V＝{V1:w1,V2:w2,V3:w3,V4:w4,V5:w5，,V6:w6} (1.1)
Vidimension, w, representing a movieiRepresenting the user's weight for each dimension of the movie, i ∈ [1,6 ]]And is and
by means of statistical analysis of user's search behavior and condition inquiry behavior and characteristic extraction of N high-grade movies from the user in T period, six-dimensional preference degrees of the user in actor, director, genre, region, time and content introduction are mined, and each dimensional weight w is obtainedi；
(2) User preference list for feature values in various dimensions in a movieShown as follows: vi＝{Tij:Wij} (1.2)
In the formula, TijIs the jth eigenvalue, W, in the ith dimensionijIs the weight of the characteristic value j in the ith dimension, and
the actor and director information which are interested by the user are mined through the searching behavior of the user, and the characteristic value information which is interested by the user in the three dimensions is mined through screening and checking of the user under different conditions of type, time and area; extracting characteristic value information of each dimension of the film by analyzing N high-scoring film watching records and collection records of a user;
combining the above analysis, calculating the weight of each feature value of each dimension, namely:
Vij＝wi*Wij(1.3)
in the formula, VijExpressing the interestingness of the jth eigenvalue in the ith dimension, i ∈ [1,6 ]]，j∈[1,n)；
s3, analyzing the six dimensions of the movie through the user interest model, and calculating the similarity of each dimension between the movie A and the movie B according to a formula (1.4) to generate a personalized movie similarity table;
in which i is ∈ [1,4 ]]Respectively representing four dimensions of actors, director, genre and region of the film, j representing a characteristic value in the dimension, VijRepresenting the weight of the jth characteristic value of the movie in the ith dimension;
the similarity calculation formula of movie a and movie B in the time dimension is:
where Date denotes the current time, DAShowing the release time of movie A, DBRepresents the release time of movie B, min () represents taking the minimum value, max () represents taking the maximum value;
film MAAnd movie MBThe similarity calculation formula in the dimension of the content profile is sim (M)6 A,M6 B) The calculation is carried out by using a Simhash algorithm;
s4, calculating the film similarity by using the formula (1.6) according to the dimension weights and the dimension similarities obtained in s2 and s 3:
in the formula, wiRepresents the weight, sim (M) for each dimensioni A,Mi B) Representing the degree of similarity for each dimension, i ∈ [1,6 ]]。
In addition, the invention also provides a movie personalized similarity calculation method for the new user, and the movie similarity is calculated by adopting a method of weighted summation of all dimensions based on the content characteristics of the movie according to the registration information of the user, including explicit information of age, gender and the like and the interest preference of the public with the same age and gender.
In order to achieve the purpose, the invention adopts the following technical scheme:
a movie personalized similarity calculation method for a new user comprises the following steps:
s1, extracting information of actors, director, genre, time and introduction of contents of each movie to form six-dimensional vector space;
s2, classifying the user based on the user dominance information, finding the most similar cluster group to the user, analyzing the average preference of the crowd in six dimensions of actor, director, type, region, release time and content introduction by using a statistical method;
s3, based on the content characteristics of the movie, calculating the similarity of the movie by adopting a method of weighted summation of all dimensions, namely:
Sim＝w1x1+w2x2+w3x3+w4x4+w5x5+w6x6(2.1)
in the formula, wiRepresenting the weight for each dimension, i ∈ [1,6 ]]This value is calculated by the method in step s 2; x is the number ofiRespectively representing the similarity of five dimensions of an actor, a director, a type, a region and release time, i belongs to [1,5 ]]And calculating by using a cosine similarity formula (2.2) to obtain:
similarity value x in content profile dimension6And calculating by using a Simhash algorithm.
The invention has the following advantages:
the method is based on six basic attributes of actors, director, type, region, time and content introduction of the movie, and a six-dimensional user interest model of two corresponding layers is established; when the movie similarity is calculated, a user is taken as a center, a user interest model is established from the user interest angle, and an individualized movie similarity recommendation list corresponding to the user one by one is formed by combining the content characteristics of the movie; for a new user, firstly, a classification technology is adopted to find the most suitable cluster of the user, a statistical method is applied to analyze the average interest preference of the user group to be used as the initial preference of the user, and a corresponding film similarity initial list is obtained; for old users, fully mining clicking, searching, condition inquiring, watching and collecting behaviors of the users, establishing a two-layer six-dimensional user interest model, and calculating a personalized film similarity list according to different preference degrees of the users to all dimensions and all characteristic values of the dimensions; and updating the movie similarity list at regular time according to the dynamic interest change of the user. The method of the invention truly takes the user as the center, can improve the recommendation effect and enables the user to experience better personalized recommendation service.
Drawings
Fig. 1 is a flowchart of a movie personalized similarity calculation method based on an old user interest model in embodiment 1 of the present invention;
fig. 2 is a schematic structural diagram of a user interest model established in embodiment 1 of the present invention;
fig. 3 is a flowchart of a movie personalized similarity calculation method based on a new user in embodiment 2 of the present invention.
Detailed Description
The invention is described in further detail below with reference to the following figures and detailed description:
example 1
Referring to fig. 1, a movie personalized similarity calculation method based on an old user interest model includes the steps:
s1, gathering of user interests
Selecting user behavior data in a certain period of time T and N movies with the highest evaluation in the film watching records in the period of time, and establishing a user dynamic behavior information base;
here, the user behavior data is mainly dynamic interests of the user, including click, search, view, and collection behaviors, and the like.
s2, formalized representation of user interest model
Through the analysis of the special medium of the movie, the invention adopts a two-layer six-dimensional space vector to represent the user interest model, and the two-layer six-dimensional space vector is shown in figure 2.
The user movie interest model contains two layers, namely: the preference of the user to each dimension of the movie and the preference of the user to the characteristic value of each dimension in the movie;
(1) the preference of the user for each dimension of the movie can be expressed as a six-tuple by using a space vector, namely:
V＝{V1:w1,V2:w2,V3:w3,V4:w4,V5:w5，,V6:w6} (1.1)
Vidimension, w, representing a movieiRepresenting the user's weight for each dimension of the movie, i ∈ [1,6 ]]And is and
by means of statistical analysis of user's search behavior and condition inquiry behavior and characteristic extraction of N high-grade movies from the user in T period, six-dimensional preference degrees of the user in actor, director, genre, region, time and content introduction are mined, and each dimensional weight w is obtainedi；
(2) User preference for feature values in various dimensions in a movie
In this layer, it is mainly reflected which feature values in this dimension are of interest to the user. Such as: the user particularly likes to dragon in the actor dimension, then the weight of this feature value is higher.
Vi＝{Tij:Wij} (1.2)
In the formula, TijIs the jth eigenvalue, W, in the ith dimensionijIs the weight of the characteristic value j in the ith dimension, and
the actor and director information which are interested by the user are mined through the searching behavior of the user, and the characteristic value information which is interested by the user in the three dimensions is mined through screening and checking of the user under different conditions of type, time and area; extracting characteristic value information of each dimension of the film by analyzing N high-scoring film watching records and collection records of a user;
and (4) combining the analysis, and calculating the weight of each characteristic value of each dimension.
The user's interest in a feature value is to know which dimension the feature value belongs to first, and know the interest weight value of the user for the dimension, and then under the dimension, the calculation formula of the interest weight value of the user for the feature value is as shown in (1.1):
Vij＝wi*Wij(1.3)
in the formula, VijExpressing the interestingness of the jth eigenvalue in the ith dimension, i ∈ [1,6 ]]，j∈[1,n)；
s3, calculation of movie M according to equation (1.4) by analysis of movie in six dimensions through user interest modelAAnd movie MBGenerating an individualized movie similarity table according to the similarity of all dimensions;
where i represents the dimensions of the actor, director, genre and region of the movie, j represents the feature value in the dimension, and VijRepresenting the weight of the jth characteristic value of the movie in the ith dimension;
film MAAnd movie MBIn the time dimensionThe similarity calculation formula is as follows:
where Date denotes the current time, DAShowing the release time of movie A, DBRepresents the release time of movie B, min () represents taking the minimum value, max () represents taking the maximum value;
film MAAnd movie MBThe similarity calculation formula in the dimension of the content profile is sim (M)6 A,M6 B) The calculation is carried out by using a Simhash algorithm;
s4, calculating the film similarity by using the formula (1.6) according to the dimension weights and the dimension similarities obtained in s2 and s 3:
in the formula, wiRepresents the weight, sim (M) for each dimensioni A,Mi B) Representing the degree of similarity for each dimension, i ∈ [1,6 ]]
s5, checking whether the user dynamic behavior information base changes every day, and updating the film similarity table off-line if the user dynamic behavior information base changes.
The main idea of the Simhash algorithm is to reduce dimensions, map a high-dimensional feature vector into an f-bit fingerprint (finger print), and determine whether an article is repeated or highly similar by comparing the Hamming Distance of the f-bit fingerprints of two documents. The Simhash algorithm is delicate but very easy to understand and implement, and the specific Simhash process is as follows:
1) the articles are first converted into a vector of weighted eigenvalues based on the traditional IR method.
2) An f-dimensional vector V is initialized, where each element has an initial value of 0.
3) For each feature in the feature vector set of the article, the following calculation is made:
a conventional hash algorithm is used to map to an f-bit signature. For the f-bit signature, if the ith bit of the signature is 1, the weight of the feature is added to the ith dimension in the vector V, otherwise, the weight of the feature is subtracted from the ith dimension of the vector.
4) After the operation is iterated on the whole feature vector set, the value of the generated f-bit fingerprint is determined according to the sign of each dimension vector in V, if the ith dimension of V is a positive number, the ith dimension of the generated f-bit fingerprint is 1, otherwise, the ith dimension of the generated f-bit fingerprint is 0.
In summary, example 1 has the following features:
1. according to the historical behaviors of the user, namely through a personalized movie recommendation system platform, various search behaviors of the user on a movie resource library and movie watching and collecting behaviors are mined; fully mining and analyzing different preference degrees of different users on six basic attributes of actors, directors, types, regions, time and content introduction of the film, namely obtaining a first-layer six-dimensional space vector representation of a user model; 2. according to the behaviors of the users, analyzing the weight of each characteristic value of different users in the six dimensions through keyword extraction or semantic analysis to obtain a second-layer six-dimensional space vector representation of the user model; 3. and expressing the interest model of the user by using a two-layer multi-dimensional space vector, and generating different movie similarity lists aiming at different users based on the interest model of the user and the basic content characteristics of the movie.
Example 2
In this embodiment 2, for a new user, since the user has no historical behavior, the registration information of the user at this time includes explicit information such as age and gender, and interest preferences of the public of the same age and gender. Based on the content characteristics of the movie, the movie similarity is calculated by adopting a method of weighted summation of all dimensions, and a specific flow diagram is shown in fig. 3.
A movie personalized similarity calculation method for a new user comprises the following steps:
s1 extracting information of actor, director, genre, time and brief introduction of content from each film to form six-dimensional vector space, and calculating similarity x of each dimension off-linei；
s2, classifying the user based on the user dominance information, finding the most similar cluster group to the user, analyzing the average preference of the crowd in six dimensions of actor, director, type, region, release time and content introduction by using a statistical method;
s3, based on the content characteristics of the movie, calculating the similarity of the movie by adopting a method of weighted summation of all dimensions, namely:
Sim＝w1x1+w2x2+w3x3+w4x4+w5x5+w6x6(2.1)
in the formula, wiRepresenting the weight for each dimension, i ∈ [1,6 ]]This value is calculated by the method in step s 2;
in the formula, xiRespectively representing the similarity of five dimensions of an actor, a director, a type, a region and release time, i belongs to [1,5 ]]And calculating by using a cosine similarity formula (2.2) to obtain:
similarity value x in content profile dimension6And calculating by using a Simhash algorithm. The Simhash algorithm described in this embodiment 2 can be referred to as the Simhash algorithm in embodiment 1.
It should be understood, however, that the description herein of specific embodiments is not intended to limit the invention to the particular forms disclosed, but on the contrary, the intention is to cover all modifications, equivalents, and alternatives falling within the spirit and scope of the invention as defined by the appended claims.
Claims (2)
1. A movie personalized similarity calculation method based on an old user interest model is characterized by comprising the following steps:
s1, establishing a user dynamic behavior information base based on the user behavior data in a certain period of time T and the N movies with the highest evaluation in the film watching records in the period of time;
the user behavior data is dynamic interests of the user and comprises clicking, searching, watching and collecting behaviors;
s2, performing data mining on the user dynamic behavior information base to obtain the preference of the user to each dimension of the movie and the preference of the user to the characteristic value of each dimension in the movie, and constructing a user interest model; wherein,
(1) the preference of the user for each dimension of the movie can be expressed as a six-tuple by using a space vector, namely:
V＝{V1:w1,V2:w2,V3:w3,V4:w4,V5:w5,V6:w6} (1.1)
Vidimension, w, representing a movieiRepresenting the user's weight for each dimension of the movie, i ∈ [1,6 ]]And is and
by means of statistical analysis of user's search behavior and condition inquiry behavior and characteristic extraction of N high-grade movies from the user in T period, six-dimensional preference degrees of the user in actor, director, genre, region, time and content introduction are mined, and each dimensional weight w is obtainedi；
(2) The user's preference for feature values in each dimension in a movie can be expressed as: vi＝{Tij:Wij} (1.2)
In the formula, TijIs the jth eigenvalue, W, in the ith dimensionijIs the weight of the characteristic value j in the ith dimension, and
the actor and director information which are interested by the user are mined through the searching behavior of the user, and the characteristic value information which is interested by the user in the three dimensions is mined through screening and checking of the user under different conditions of type, time and area; extracting characteristic value information of each dimension of the film by analyzing N high-scoring film watching records and collection records of a user;
combining the above analysis, calculating the weight of each feature value of each dimension, namely:
Vij＝wi*Wij(1.3)
in the formula, VijExpressing the interestingness of the jth eigenvalue in the ith dimension, i ∈ [1,6 ]]，j∈[1,n)；
s3, analyzing the six dimensions of the movie through the user interest model, and calculating the similarity of each dimension between the movie A and the movie B according to a formula (1.4) to generate a personalized movie similarity table;
in which i is ∈ [1,4 ]]Respectively representing four dimensions of actors, director, genre and region of the film, j representing a characteristic value in the dimension, VijRepresenting the weight of the jth characteristic value of the movie in the ith dimension;
the similarity calculation formula of movie a and movie B in the time dimension is:
where Date denotes the current time, DAShowing the release time of movie A, DBRepresents the release time of movie B, min () represents taking the minimum value, max () represents taking the maximum value;
the similarity calculation formula of movie A and movie B in the dimension of content profile is sim (M)6 A,M6 B) The calculation is carried out by using a Simhash algorithm; the Simhash algorithm maps the high-dimensional feature vector into an f-bit fingerprint, and the Hamming distance of the f-bit fingerprints of the two documents is compared to determine whether the articles are repeated or highly approximate;
the Simhash algorithm process is as follows:
1) firstly, converting an article into a vector formed by a group of weighted characteristic values based on a traditional IR method;
2) initializing a vector V of f dimensions, wherein each element has an initial value of 0;
3) for each feature in the feature vector set of the article, the following calculation is made:
mapping to an f-bit signature by using a traditional hash algorithm;
for the f-bit signature, if the ith bit of the signature is 1, adding the weight of the feature to the ith dimension in the vector V, and otherwise, subtracting the weight of the feature from the ith dimension of the vector;
4) after iterating the operation on the whole feature vector set, determining the value of the generated f-bit fingerprint according to the sign of each dimension vector in V, wherein if the ith dimension of V is a positive number, the ith dimension of the generated f-bit fingerprint is 1, otherwise, the ith dimension is 0;
s4, calculating the film similarity by using the formula (1.6) according to the dimension weights and the dimension similarities obtained in s2 and s 3:
in the formula, wiRepresents the weight, sim (M) for each dimensioni A,Mi B) Representing the degree of similarity for each dimension, i ∈ [1,6 ]]。
2. A method for calculating personalized similarity of a movie to a new user is characterized by comprising the following steps:
s1, extracting information of actors, director, genre, time and introduction of contents of each movie to form six-dimensional vector space;
s2, classifying the user based on the user dominance information, finding the most similar cluster group to the user, analyzing the average preference of the crowd in six dimensions of actor, director, type, region, release time and content introduction by using a statistical method;
s3, based on the content characteristics of the movie, calculating the similarity of the movie by adopting a method of weighted summation of all dimensions, namely:
Sim＝w1x1+w2x2+w3x3+w4x4+w5x5+w6x6(2.1)
in the formula, wiRepresenting the weight for each dimension, i ∈ [1,6 ]]This value is passed through the method in step s2Calculating to obtain; x is the number ofiRespectively representing the similarity of five dimensions of an actor, a director, a type, a region and release time, i belongs to [1,5 ]]And calculating by using a cosine similarity formula (2.2) to obtain:
similarity value x in content profile dimension6Calculating by using a Simhash algorithm;
the Simhash algorithm maps the high-dimensional feature vector into an f-bit fingerprint, and the Hamming distance of the f-bit fingerprints of the two documents is compared to determine whether the articles are repeated or highly approximate;
the Simhash algorithm process is as follows:
1) firstly, converting an article into a vector formed by a group of weighted characteristic values based on a traditional IR method;
2) initializing a vector V of f dimensions, wherein each element has an initial value of 0;
3) for each feature in the feature vector set of the article, the following calculation is made:
mapping to an f-bit signature by using a traditional hash algorithm;
for the f-bit signature, if the ith bit of the signature is 1, adding the weight of the feature to the ith dimension in the vector V, and otherwise, subtracting the weight of the feature from the ith dimension of the vector;
4) after the operation is iterated on the whole feature vector set, the value of the generated f-bit fingerprint is determined according to the sign of each dimension vector in V, if the ith dimension of V is a positive number, the ith dimension of the generated f-bit fingerprint is 1, otherwise, the ith dimension of the generated f-bit fingerprint is 0.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410753644.8A CN104462385B (en) | 2014-12-10 | 2014-12-10 | A kind of film personalization similarity calculating method based on user interest model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410753644.8A CN104462385B (en) | 2014-12-10 | 2014-12-10 | A kind of film personalization similarity calculating method based on user interest model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104462385A CN104462385A (en) | 2015-03-25 |
| CN104462385B true CN104462385B (en) | 2018-07-03 |
Family
ID=52908420
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410753644.8A Active CN104462385B (en) | 2014-12-10 | 2014-12-10 | A kind of film personalization similarity calculating method based on user interest model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104462385B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106021485A (en) * | 2016-05-19 | 2016-10-12 | 中国传媒大学 | Multi-element attribute movie data visualization system |
Families Citing this family (29)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106445942B (en) * | 2015-08-05 | 2020-07-03 | 腾讯科技（北京）有限公司 | User cross-screen identification method and device |
| CN105389329B (en) * | 2015-09-21 | 2019-02-12 | 中国人民解放军国防科学技术大学 | A kind of open source software recommended method based on community review |
| CN106708883B (en) * | 2015-11-17 | 2020-09-29 | 阿里巴巴集团控股有限公司 | Recommendation method and device |
| CN105718565B (en) * | 2016-01-20 | 2019-07-02 | 北京京东尚科信息技术有限公司 | The construction method and construction device of data warehouse model |
| CN106055657B (en) * | 2016-06-03 | 2019-10-01 | 网智天元科技集团股份有限公司 | Viewing index assessment system for specific crowd |
| CN106372235A (en) * | 2016-09-12 | 2017-02-01 | 中国联合网络通信集团有限公司 | Movie recommendation method and system |
| CN108230059B (en) * | 2016-12-13 | 2022-04-15 | 中国联合网络通信集团有限公司 | Movie recommendation method and device |
| CN106997381B (en) * | 2017-03-21 | 2021-03-09 | 海信集团有限公司 | Method and device for recommending movies to target user |
| CN106991173A (en) * | 2017-04-05 | 2017-07-28 | 合肥工业大学 | Collaborative filtering recommending method based on user preference |
| CN107122423A (en) * | 2017-04-06 | 2017-09-01 | 深圳Tcl数字技术有限公司 | Video display promotion method and device |
| CN108932645A (en) * | 2017-05-26 | 2018-12-04 | 车伯乐（北京）信息科技有限公司 | A kind of user selects vehicle system |
| 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 |
| CN108038730A (en) * | 2017-12-22 | 2018-05-15 | 联想(北京)有限公司 | Product similarity determination methods, device and server cluster |
| CN108419100B (en) * | 2018-01-29 | 2020-10-02 | 山东云缦智能科技有限公司 | Method and system for acquiring similarity of movie playing behaviors of users |
| CN108460125A (en) * | 2018-02-26 | 2018-08-28 | 影核（北京）网络科技有限公司 | A method of carrying out displaying labeling classification for movie theatre user |
| CN108664558B (en) * | 2018-04-04 | 2020-05-05 | 山东科技大学 | Network television personalized recommendation service method for large-scale users |
| CN108804506A (en) * | 2018-04-13 | 2018-11-13 | 北京猫眼文化传媒有限公司 | A kind of the recommendation method, apparatus and electronic equipment of information |
| CN108846055B (en) * | 2018-05-31 | 2021-08-24 | 北京五八信息技术有限公司 | Search processing method and device, electronic equipment and readable storage medium |
| CN109189899A (en) * | 2018-08-01 | 2019-01-11 | 优视科技新加坡有限公司 | Content interest acquisition and content-data recommended method, device, equipment/terminal/server and storage medium |
| CN109118284A (en) * | 2018-08-15 | 2019-01-01 | 深圳快购科技有限公司 | The management method and system of movie ticket sale |
| CN109299375A (en) * | 2018-10-24 | 2019-02-01 | 中国平安人寿保险股份有限公司 | Information personalized push method, device, electronic equipment and storage medium |
| CN109933648B (en) * | 2019-02-28 | 2022-07-05 | 北京学之途网络科技有限公司 | Real user comment distinguishing method and device |
| CN110083775B (en) * | 2019-05-13 | 2021-06-25 | 北京顺丰同城科技有限公司 | Configuration method and configuration device for recommended resources |
| CN110210765B (en) * | 2019-06-06 | 2021-06-11 | 中国传媒大学 | Movie and television resource matching method and system |
| CN110569374B (en) * | 2019-08-20 | 2022-03-18 | 成都信息工程大学 | Movie recommendation method based on improved collaborative filtering algorithm |
| CN110674344B (en) * | 2019-08-28 | 2022-03-08 | 北京爱奇艺智慧娱乐科技有限公司 | Method, device, equipment and storage medium for generating model and recommending film |
| US11727312B2 (en) | 2019-09-03 | 2023-08-15 | International Business Machines Corporation | Generating personalized recommendations to address a target problem |
| CN110866181B (en) * | 2019-10-12 | 2022-04-22 | 平安国际智慧城市科技股份有限公司 | Resource recommendation method, device and storage medium |
| CN114547459B (en) * | 2022-02-23 | 2023-10-31 | 深圳环金科技有限公司 | Cross-border electronic commerce data processing method and system |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20110049674A (en) * | 2009-11-05 | 2011-05-12 | 삼성전자주식회사 | Service providing apparatus and method for recommending service thereof |
| CN102411596A (en) * | 2010-09-21 | 2012-04-11 | 阿里巴巴集团控股有限公司 | Information recommendation method and system |
| CN102722532B (en) * | 2012-05-18 | 2014-04-02 | 山东大学 | Music recommendation algorithm based on content and user history |
| CN103294795A (en) * | 2013-05-24 | 2013-09-11 | 华东师范大学 | Method for adjusting film recommending diversity by utilizing users' characters |
-
2014
- 2014-12-10 CN CN201410753644.8A patent/CN104462385B/en active Active
Non-Patent Citations (3)
| Title |
|---|
| simhash算法原理和代码实现;新浪博客;《http://blog.sina.com.cn/s/blog_81e6c30b0101cpvu.html》;20121129;第1页 * |
| 一种基于数据迁移的冷启动解决算法;马远坤 等;《计算机工程》;20140415;第40卷(第4期);274页 * |
| 基于多维度权重动态更新的用户兴趣模型;任保宁 等;《计算机工程》;20140915;第40卷(第9期);43-44页 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106021485A (en) * | 2016-05-19 | 2016-10-12 | 中国传媒大学 | Multi-element attribute movie data visualization system |
| CN106021485B (en) * | 2016-05-19 | 2019-05-14 | 中国传媒大学 | A kind of polynary attribute cinematic data visualization system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104462385A (en) | 2015-03-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104462385B (en) | A kind of film personalization similarity calculating method based on user interest model | |
| US11074434B2 (en) | Detection of near-duplicate images in profiles for detection of fake-profile accounts | |
| CN109947987B (en) | Cross collaborative filtering recommendation method | |
| Guo et al. | User relationship strength modeling for friend recommendation on Instagram | |
| Vivekanandan et al. | Hybrid convolutional neural network (CNN) and long-short term memory (LSTM) based deep learning model for detecting shilling attack in the social-aware network | |
| Gras et al. | Identifying grey sheep users in collaborative filtering: a distribution-based technique | |
| Li et al. | A novel time-aware hybrid recommendation scheme combining user feedback and collaborative filtering | |
| Chung et al. | Categorization for grouping associative items using data mining in item-based collaborative filtering | |
| Yadav et al. | A new approach for movie recommender system using K-means Clustering and PCA | |
| Cheung et al. | Characterizing user connections in social media through user-shared images | |
| Wang et al. | Link prediction in heterogeneous collaboration networks | |
| Elahi et al. | Recommender systems: Challenges and opportunities in the age of big data and artificial intelligence | |
| Cao et al. | Web API recommendation via combining graph attention representation and deep factorization machines quality prediction | |
| CN114817715A (en) | Privacy-protecting Internet of things entity service searching method | |
| Cheung et al. | Evaluating the privacy risk of user-shared images | |
| Feng et al. | Recommendations based on comprehensively exploiting the latent factors hidden in items’ ratings and content | |
| Basha et al. | A proficient video recommendation framework using hybrid fuzzy C means clustering and Kullback-Leibler divergence algorithms | |
| Solainayagi et al. | Trustworthy media news content retrieval from web using truth content discovery algorithm | |
| Wang et al. | Recommender systems and their security concerns | |
| CN109543094B (en) | Privacy protection content recommendation method based on matrix decomposition | |
| Kannout et al. | Clustering-based Frequent Pattern Mining Framework for Solving Cold-Start Problem in Recommender Systems | |
| Duan et al. | A hybrid recommendation system based on fuzzy c-means clustering and supervised learning | |
| Sun et al. | Enabling 5G: sentimental image dominant graph topic model for cross-modality topic detection | |
| Venkat et al. | Clustering of huge data with fuzzy c-means and applying gravitational search algorithm for optimization | |
| CN104641386A (en) | Method and apparatus for obfuscating user demographics |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |