CN103714130B - Video recommendation system and method - Google Patents
Video recommendation system and method Download PDFInfo
- Publication number
- CN103714130B CN103714130B CN201310684807.7A CN201310684807A CN103714130B CN 103714130 B CN103714130 B CN 103714130B CN 201310684807 A CN201310684807 A CN 201310684807A CN 103714130 B CN103714130 B CN 103714130B
- Authority
- CN
- China
- Prior art keywords
- user
- video
- friend
- data
- module
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/70—Information retrieval; Database structures therefor; File system structures therefor of video data
- G06F16/78—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
-
- 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)
- Library & Information Science (AREA)
- Multimedia (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
Video recommendation system and method that the present invention is provided, being obtained by data obtaining module includes userspersonal information, the multi-source data of the user social contact network information and video classification information, the multi-source data is pre-processed by data preprocessing module, the user data library storage multi-source data after pretreatment built again through data memory module, the microblog data sent again by user's characteristic analysis module according to user obtains the good friend's clustering of personality and mood and social network analysis module to user of user, obtain the friend circle of the user, video recommendations module is according to the personality of the user, mood and friend circle, choose the video of the user preferences and the video chosen is shown to the user by front end display module.
Description
Technical field
The present invention relates to areas of information technology, more particularly to a kind of video recommendation system and method.
Background technology
Current existing video recommendation method and system mainly have two kinds.A kind of is the video class that user actively selects to like
Not, then system recommends the video of identical category according to user's selection;It is another, it is the viewing historical record according to user,
Recommend other videos of video identical category watched with user.
The common trait of both approaches is all the personal information using user and lays particular emphasis on only its viewing information of utilization
To carry out video recommendations, have ignored user's sex, age, occupation, nationality, area etc., other portray the important letter of user characteristics
Breath.Which results in a certain class film that its recommendation results concentrations was watched in user, it is impossible to reflects the true interest of user,
The degree of accuracy of recommendation, spend comprehensively it is relatively low.
And the member of the social networks such as relatives, the friend of user can directly affect the viewing custom of user, such as user
Good friend Xiang Qi recommend a certain film, or the multidigit good friend of the user have viewed same portion's film, then the user
Like the possibility of the film very big.These information have very big reference value for video recommendations, and method mentioned above is but
These social informations are not used completely.The social feature of user can not be reacted completely which results in its recommendation results, it is impossible to predict
The potential video classification that user may like, without adaptability, flexibility is poor.
Therefore prior art also existing defects, need to be improved and develop.
The content of the invention
The present invention provides a kind of video recommendation system, and it is accurate, comprehensive that the video recommendation system can be supplied to user to recommend
Video.
A kind of video recommendation system, including:
Data obtaining module, includes userspersonal information, the user social contact network information and video classification information for obtaining
Multi-source data, the userspersonal information include the age, sex, occupation, nationality, location, the video type liked, see
The list of videos seen, content of microblog that the friend information of the user social contact network information including user, user push and
Send when and where, the list of videos that user watched, the video scoring information of user of microblogging, the video classification information
Including title, director, age, type, video scoring and video tab;
Data preprocessing module, for being pre-processed to the multi-source data, the data preprocessing module includes:Number
According to cleaning submodule, for rejecting incomplete data in the multi-source data, hough transformation submodule, for difference will to be come from
The form of the multi-source data of platform is unified, and data integration submodule, for by after stipulations from disparate databases
Data integration is into identical database;
Data memory module, builds customer data base, and multi-source data after pretreatment is stored in into the database
In;
User's characteristic analysis module, the microblog data for being sent according to user obtains the personality and mood of user;
Social network analysis module, for good friend's clustering to user, obtains the friend circle of the user;
Video recommendations module, for the personality according to the user, mood and friend circle, chooses regarding for the user preferences
Frequently;
Front end display module, for showing the video chosen to the user;
The social network analysis module is analyzed for user social contact network data, including:
The good friend of user is expressed as to the set of series of features vector, the set includes region, age, sex, duty
Industry, the video type liked and the list of videos watched;The similarity between good friend and good friend is calculated, and according to similarity
If the good friend of user is divided into dry type;
The intimate degree of user and its good friend are calculated, its close friend's circle is obtained;
The propagating influence of user is calculated according to the microblogging quantity forwarded and forwarding quantity of user good friend:P=0.2S+
0.8F, wherein P are the propagating influences of good friend, and S is microblogging quantity forwarded, and F is the number of times that microblogging is forwarded.
Further, described information acquisition module includes:
Userspersonal information's acquisition submodule, by the user registration module of webpage version, obtains the personal information of user;
Social network information acquisition submodule, the user social contact network information is obtained by intercommunion platform;And
Video classification acquisition of information submodule, video classification information is obtained by video website.
Further, the data memory module is used to userspersonal information storage arriving Sqlserver or Mysql
Relevant database in and by the user social contact network information store MongoDB figure relational database in.
Further, user's characteristic analysis module includes:
User's character analysis submodule, the microblog data for being sent according to user in the past analyzes the personality of user;And
User emotion analyzes submodule, and the microblog data for being sent according to user in the past analyzes the mood of user.
Further, the video recommendations module includes:
Userspersonal information's recommending module, the video film of user preferences is chosen for the personality according to user and mood;
And
Social network information recommending module, the video film for choosing user preferences according to the friend circle of user.
Further, in addition to user profile real-time capture module, for the social network information of user in real, and
It is stored in the database.
Further, in addition to result optimizing module, for the video recommendations module according to regarding that friend circle is chosen
Frequency assigns weights, and is ranked up generation recommendation list according to the weights.
Further, the front end display module includes form web page, television terminal or mobile phone terminal.
In addition, present invention also offers a kind of video recommendation method, comprising the steps:
Acquisition includes the multi-source data of userspersonal information, the user social contact network information and video classification information, the use
Family personal information includes age, sex, occupation, nationality, location, the video type liked, the list of videos watched, institute
State friend information of the user social contact network information including user, the content of microblog that user pushes and time and the ground for sending microblogging
List of videos that point, user watched, the video scoring information of user, the video classification packet include title, director, year
Generation, type, video scoring and video tab;
The multi-source data is pre-processed, including:Incomplete data in the multi-source data are proposed, will be from not
Form with the multi-source data of platform is unified, by the data integration from disparate databases after stipulations to identical data
In storehouse;
Customer data base is built, and multi-source data after pretreatment is stored in the database;
The microblog data sent according to user obtains the personality and mood of user;
To good friend's clustering of user, the friend circle of the user is obtained, the friend circle includes close friend's circle, good friend
Circle and maximum propagation influence power good friend;
According to the personality, mood and friend circle of the user, the video of the user preferences is chosen;And
For showing the video chosen to the user;
To good friend's clustering of user, the friend circle of the user is obtained, is comprised the steps:
The good friend of user is expressed as to the set of series of features vector, the set includes region, age, sex, duty
Industry, the video type liked and the list of videos watched;
The distance between Companion Vector is calculated to characterize the similarity between good friend and good friend, and according to similarity using poly-
If good friend's automatic cluster of user is dry type by class algorithm KMeans;
Choose higher preceding 20 good friends of similarity and constitute its close friend's circle;
The propagating influence of user good friend, the meter are calculated according to the microblogging quantity forwarded and forwarding quantity of user good friend
Calculation formula is P=0.2S+0.8F, and wherein P is the propagating influence of good friend, and S is microblogging quantity forwarded, and F is that microblogging is forwarded
Number of times.
Further, the microblog data sent according to user obtains the personality and mood of user, comprises the steps:
Mathematical abstractions are the vector space of a N-dimensional, and each vector correspondence personality/mood fundamental, and it is neutral
Lattice essentiality is:Introversion, flare, stably, it is unstable, mood fundamental is:Happy, indignation, it is sad, frightened, detest, it is frightened
Very, the vector space of the N-dimensional is Q=[x1、x2、……];
Participle is carried out to the microblog data, semantic feature is obtained, is defined as C=[c1、c2、……];
Set up Function Mapping relation, Q=f (C), wherein, Q is basic and key element the set of personality/mood, and C is microblogging number
According to the semantic feature set obtained after participle, f is corresponding mapping function;
Microblog data C is collected from the microblogging, the score of each fundamental is judged, so as to obtain personality/mood base
Essentiality Q, composing training data set;
Carry out learning to obtain model of fit for training dataset using neural network algorithm, then according to obtained model
Predict user's personality/mood.
Further, according to the personality, mood and friend circle of the user, the video of the user preferences is chosen, including
Following step:
The video film of user preferences is chosen according to the personality of user and mood;And
The video film of user preferences is chosen according to the friend circle of user.
Further, the video film of user preferences is chosen according to the personality of user and mood, is comprised the steps:
Every film is described with region, age, protagonist, four attributes of type;
Preference of the user for the film with some attribute is obtained according to the personality and mood of user, and assigned
Weights;
For any one film, the preference that the user corresponds to the film native is summed up, and obtains institute
State fancy grade of the user to the film;
All films are calculated, and choose 10 maximum films of fancy grade and recommend user.
Further, the video film of user preferences is chosen according to the friend circle of user, is comprised the steps:
Calculate every good friend in user's friend circle and watch more video jointly, extract preceding 10 films and tied as recommendation
Really;
More video is watched in the close friend's circle for finding user jointly, preceding 10 films is extracted and is used as recommendation results;
According to the propagating influence of user good friend, preceding 10 films are extracted as recommendation results.
Further, also comprise the steps:The video that the video recommendations module is chosen according to friend circle is assigned and weighed
Value, and it is ranked up generation recommendation list according to the weights.
Further, weights are assigned to the video that the video recommendations module is chosen according to friend circle, and according to the power
Value is ranked up generation recommendation list, comprises the steps:
Recommendation results, the recommendation results of the good friend of propagating influence maximum and the common circle drawn respectively for close friend's circle
The recommendation results of son assign weights;
Calculate the weights sum of all films appeared in recommendation results;
And generation recommendation list is ranked up according to the weights, it is used as consequently recommended result.
Video recommendation system and method that the present invention is provided, by data obtaining module obtain include userspersonal information,
The multi-source data of the user social contact network information and video classification information, is carried out pre- by data preprocessing module to the multi-source data
Processing, then the user data library storage multi-source data after pretreatment built through data memory module, then pass through user's feature
The microblog data that analysis module is sent according to user obtains the personality and mood and social network analysis module of user to user's
Good friend's clustering, obtains the friend circle of the user, and video recommendations module is according to the personality, mood and good friend of the user
Circle, chooses the video of the user preferences and shows the video chosen to the user by front end display module.
Brief description of the drawings
The composition schematic diagram for the video recommendation system that Fig. 1 provides for the present invention;
The composition schematic diagram for the data obtaining module that Fig. 2 provides for the present invention;
The composition schematic diagram for the data preprocessing module that Fig. 3 provides for the present invention;
The composition schematic diagram for user's characteristic analysis module that Fig. 4 provides for the present invention;
The step flow chart for the video recommendation method that Fig. 5 provides for the present invention;
Fig. 6 obtains the personality of user and the step flow of mood for the microblog data sent according to user that the present invention is provided
Figure;
Good friend's clustering to user that Fig. 7 provides for the present invention, obtains the step flow of the friend circle of the user
Figure;
The step flow for the video film that user preferences are chosen according to the personality and mood of user that Fig. 8 provides for the present invention
Figure;
The step flow chart for the video film that user preferences are chosen according to the friend circle of user that Fig. 9 provides for the present invention.
Embodiment
In order that the objects, technical solutions and advantages of the present invention become apparent from, below in conjunction with drawings and Examples, to this hair
It is bright to be further elaborated.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, and without
It is of the invention in limiting.
Referring to Fig. 1, the composition schematic diagram for the video recommendation system 100 that Fig. 1 provides for the present invention, including:Acquisition of information
Module 110, data preprocessing module 120, data memory module 130, user's characteristic analysis module 140, social network analysis mould
Block 150, video recommendations module 160 and front end display module 170.
Referring to Fig. 2, data obtaining module 110 be used for obtain include userspersonal information, the user social contact network information and
The multi-source data of video classification information.Preferably, data obtaining module 110 includes:Userspersonal information's acquisition submodule 111,
Social network information acquisition submodule 112 and video classification acquisition of information submodule 113.Userspersonal information's acquisition submodule
111 by webpage version user registration module, obtain user personal information, wherein, personal information include age, sex, duty
Industry, nationality, location, the video type liked, the list of videos watched;Social network information acquisition submodule 112 is led to
Cross intercommunion platform and obtain the user social contact network information, for example, it is possible to use Sina weibo, Tengxun's microblogging, Tencent QQ, Yoqoo etc.
The opening API interface that platform is provided obtains user profile, wherein, the user social contact network information include user friend information,
The content of microblog and when and where, the list of videos that user watched, the video of user of transmission microblogging that user pushes are beaten
Divide information;Video classification acquisition of information submodule 113 by video website obtain video classification information, for example can from bean cotyledon,
Required information is extracted in the major video such as youku.com website, wherein, video classification packet includes title, director, age, class
Type, video scoring, video tab etc..
Referring to Fig. 3, data preprocessing module 120 is used to pre-process multi-source data.Preferably, data prediction
Module 120 includes:Data cleansing submodule 121, hough transformation submodule 122 and data integration submodule 123.Data cleansing
Module 121 is used to reject incomplete data in the multi-source data, such as user does not fill in any in addition to name
The data record of people's information will not be analyzed recommending module and use;Hough transformation submodule 122 is used to that different platform will to be come from
The form of multi-source data unified, for example, put forward by personal register information of user and from websites such as microblogging, QQ
People's information unification is following form:User name, age, sex, occupation, nationality, location, the video type liked, viewing
The list of videos crossed, the video scoring information of user;Social network information from microblogging, QQ is unified for following form:With
Name in an account book, user good friend 1, user good friend 2, user good friend 3 etc.;By the video information stipulations from networks such as bean cotyledon, youku.coms for such as
Lower form:Video name, video age, director, protagonist, region, type, scoring, label;Data integration submodule 123 is used for
By the data integration from disparate databases after stipulations into identical database.
Data memory module 130 builds customer data base, and multi-source data after pretreatment is stored in database.
Preferably, data memory module 130 is used for the relevant database stored userspersonal information to Sqlserver or Mysql
In and by the user social contact network information store MongoDB figure relational database in.It is appreciated that Sqlserver or Mysql or
MongoDB databases are a kind of preferred embodiment therein, and in practice can also be using other databases.
Referring to Fig. 4, user's characteristic analysis module 140 is used for the personality of the microblog data acquisition user sent according to user
And mood.Preferably, user's characteristic analysis module 140 includes user's character analysis submodule 141 and user emotion analysis submodule
Block 142, user's character analysis submodule 141 is used for the personality of the microblog data analysis user sent in the past according to user, user
Mood analysis submodule 142 is used for the mood of the microblog data analysis user sent in the past according to user.
Specifically, the microblog data that user's character analysis submodule 141 is used to be sent in the past according to user analyzes user's
Personality, the specific analysis using following step to user's personality:
Mathematical abstractions are the vector space of a N-dimensional, and each vectorial correspondence lattice essentiality, and the vector of N-dimensional is empty
Between be Q=[x1、x2、……];
Participle is carried out to the microblog data, semantic feature is obtained, is defined as C=[c1、c2、……];
Set up Function Mapping relation, Q=f (C), wherein, Q is basic and key element the set of personality, and C is microblog data participle
The semantic feature set obtained afterwards, f is corresponding mapping function;
Microblog data C is collected from the microblogging, its personality/emotion trait of manual analysis resolves to personality fundamental
Q, training dataset is obtained by artificial labeled data;
Carry out learning to obtain model of fit for training dataset using neural network algorithm, then according to obtained model
Predict user's personality.
And use user emotion analyze the microblog data that submodule 142 sent according to user in the past analyze the mood of user with
Above-mentioned user's character analysis submodule 141 is realized completely the same to user's character analysis, will not be repeated here.
Social network analysis module 150 is used for good friend's clustering to user, obtains the friend circle of the user.It is social
The module of nework analysis module 150 carries out a variety of analyses for user social contact network data, mainly including following analysis:
One, the good friend to user are clustered, and its good friend is divided into some classifications according to feature.By good friend's table of user
It is shown as the set of series of features vector:Region, the age, sex, occupation, the video type liked, the list of videos watched,
And the similarity between good friend and good friend is calculated, and if the good friend of user is divided into dry type according to similarity.
Secondly, calculate the intimate degree of user and its good friend, obtain its close friend's circle.By calculating between user and good friend
Similarity, the selection several good friends of similarity highest enter close friend's circle of access customer, weight are increased when recommending.
Thirdly, calculate the propagating influence of its good friend, and then determine its influence power size for the user.According to user
The microblogging quantity forwarded of good friend calculates the propagating influence of user with forwarding quantity:
P=0.2S+0.8F
Wherein P is the propagating influence of good friend, and S is microblogging quantity forwarded, and F is the number of times that microblogging is forwarded.
And its shared proportion in analysis recommending module is determined according to the propagating influence of its good friend, influence power is got over
The proportion that big good friend accounts in analysis recommending module is bigger.
Video recommendations module 160 is used for personality, mood and friend circle according to the user, chooses regarding for user preferences
Frequently;The module according to it is obtained as above to various data provide preliminary recommendation results, mainly including following recommendation results:
First, showing that the algorithm of recommendation results is as follows according to userspersonal information:
A) every video is described with region, age, protagonist, four attributes of type.
B) preference of the user for the film with some attribute is obtained from the personal information of user, also with one
Individual weights are represented.
C) for a certain portion's film, the preference value that the user corresponds to this film native, which is summed up, can be used
Fancy grade of the family to this film.
D) all films are calculated, and chooses 10 larger films of fancy grade and recommend user.
Second, the social network information according to user recommends video information, its algorithm is as follows:
A) good friend of user is divided into some classifications in (6) social network analysis module, calculates the personal letter of user
Similarity between breath and good friend's classification, so that it is determined which kind of user is more likely to belong to.
B) calculate every good friend in the category and watch more video jointly, extract first 10 and be used as recommendation results.
C) more video is watched in the close friend's circle for finding user jointly, first 10 is extracted and is used as recommendation results.
D) more influential good friend is found, first 10 are extracted as recommendation results.
Front end display module 170 is used to show the video chosen to the user.Front end display module include form web page,
Television terminal or mobile phone terminal.
In addition, above-mentioned video recommendation system 100 also includes user profile real-time capture module 180, used for obtaining in real time
The social network information at family, and be stored in the database.It is appreciated that after user is by terminal logs ins such as webpages, system
The history for showing the user in terminal watches list, and records the video information that user watches in real time, and then updates the data storehouse.
The social information in user's nearly a period of time is then followed the trail of by data acquisition module simultaneously for users such as microblogging, QQ, such as pushed
Microblogging, the renewal of personal good friend's data etc., and be stored in database.
In addition, above-mentioned video recommendation system 100 also includes result optimizing module 190, for the video recommendations module
The video chosen according to friend circle assigns weights, and is ranked up generation recommendation list according to the weights.It is appreciated that for
Each result that video recommendations module 160 is drawn draws final result after optimizing by result optimizing module 190, specifically includes:
A) weights 1, the recommendation drawn for the larger good friend of influence power are assigned for the recommendation results that close friend's circle is drawn
As a result weights 1.5 are assigned, weights 0.8 are assigned for the recommendation results that common circle is drawn;
B) the weights sum of all films appeared in recommendation results is calculated;
C) maximum preceding 20 videos of circle are selected after sorting, consequently recommended result is used as.
Referring to Fig. 5, the video recommendation method 200 provided for the present invention, comprises the steps:
Step S210:Acquisition includes the multi-source number of userspersonal information, the user social contact network information and video classification information
According to;
It is appreciated that the personal information of user can be obtained, personal information includes by the user registration module of webpage version
Age, sex, occupation, nationality, location, the video type liked, the list of videos watched;Obtained by intercommunion platform
The user social contact network information, for example, it is possible to use the opening that the platform such as Sina weibo, Tengxun's microblogging, Tencent QQ, Yoqoo is provided
Api interface obtains user profile, wherein, the microblogging that the user social contact network information includes the friend information of user, user pushes
Content and when and where, the list of videos that user watched, the video scoring information of user for sending microblogging;Pass through video
Website obtains video classification information, such as can from bean cotyledon, youku.com required information is extracted in major video website, its
In, video classification packet includes title, director, age, type etc., video scoring, video tab.
Step S220:Multi-source data is pre-processed;
Specifically include:Incomplete data in the multi-source data are rejected, such as user does not fill in addition to name
The data record of any personal information will not be analyzed recommending module and use;By the form of the multi-source data from different platform
Unified, for example, the personal information unification put forward by personal register information of user and from websites such as microblogging, QQ is as follows
Form:User name, age, sex, occupation, nationality, location, the video type liked, the list of videos watched, user
Video scoring information;Social network information from microblogging, QQ is unified for following form:User name, user good friend 1, use
Family good friend 2, user good friend 3 etc.;It is following form by the video information stipulations from networks such as bean cotyledon, youku.coms:Video name, regard
Generation, director, protagonist, region, type, scoring, label in consecutive years;And by the data integration from disparate databases after stipulations to phase
In same database.
Step S230:Customer data base is built, and multi-source data after pretreatment is stored in the database;
Preferably, by userspersonal information's storage into Sqlserver or Mysql relevant database and by user society
In the figure relational database for handing over network information storage MongoDB.It is appreciated that Sqlserver or Mysql or MongoDB data
Storehouse is a kind of preferred embodiment therein, and in practice can also be using other databases.
Step S240:The microblog data sent according to user obtains the personality and mood of user;
Referring to Fig. 6, the microblog data sent according to user obtains the personality and mood of user, comprise the steps:
Step S241:Mathematical abstractions are the vector space of a N-dimensional, and each vector correspondence personality/mood is wanted substantially
Element, wherein personality fundamental is:Introversion, flare, stably, it is unstable, mood fundamental is:It is happy, indignation, sad, probably
Fear, detest, in surprise, the vector space of the N-dimensional is Q=[x1、x2、……];
Step S242:Participle is carried out to the microblog data, semantic feature is obtained, is defined as C=[c1、c2、……];
Step S243:Set up Function Mapping relation, Q=f (C), wherein, Q is basic and key element the set of personality/mood, C
It is the semantic feature set obtained after microblog data participle, f is corresponding mapping function;
Step S244:Microblog data C is collected from the microblogging, invites the professional person with psychological consultation experience to read
Microblogging, judges the score of each fundamental, so as to obtain personality/mood fundamental Q, composing training data set;
Step S245:Carry out learning to obtain model of fit for training dataset using neural network algorithm, then basis
Obtained model prediction user personality/mood.
Step S250:To good friend's clustering of user, the friend circle of the user is obtained, the friend circle includes close friend
Circle, good friend's circle and maximum propagation influence power good friend;
Referring to Fig. 7, to good friend's clustering of user, obtaining the friend circle of the user, comprising the steps:
Step S251:The good friend of user is expressed as to the set of series of features vector, the set includes region, year
Age, sex, occupation, the video type liked and the list of videos watched;
Step S252:The distance between Companion Vector is calculated to characterize the similarity between good friend and good friend, and according to phase
If using clustering algorithm KMeans by good friend's automatic cluster of user for dry type like degree;
Step S253:Choose higher preceding 20 good friends of similarity and constitute its close friend's circle;
Step S254:The propagation effect of user good friend is calculated according to the microblogging quantity forwarded and forwarding quantity of user good friend
Power, the calculation formula is P=0.2S+0.8F, and wherein P is the propagating influence of good friend, and S is microblogging quantity forwarded, and F is microblogging
The number of times being forwarded.It is appreciated that determining its shared ratio in analysis recommending module according to the propagating influence of its good friend
Weight, the proportion that the bigger good friend of influence power accounts in analysis recommending module is bigger.
Step S260:According to the personality, mood and friend circle of the user, the video of the user preferences is chosen;
According to the personality, mood and friend circle of the user, the step flow chart of the video of the user preferences, bag are chosen
Include following step:
Step S261:The video film of user preferences is chosen according to the personality of user and mood;:
Referring to Fig. 8, choosing the video film of user preferences according to the personality of user and mood, comprise the steps:
Step S2611:Every film is described with region, age, protagonist, four attributes of type;
Step S2612:Preference of the user for the film with some attribute is obtained according to the personality and mood of user
Degree, and assign weights;
Step S2613:For any one film, the preference that the user corresponds to the film native is added
With obtain fancy grade of the user to the film;
Step S2614:All films are calculated, and choose 10 maximum films of fancy grade and recommend user.
It is appreciated that it is a kind of wherein preferred mode to choose 10 films to recommend user, it can also select in practice
Other film quantity are taken to user.
Step S262:The video film of user preferences is chosen according to the friend circle of user.
Referring to Fig. 9, choosing the video film of user preferences according to the friend circle of user, comprise the steps:
Step S2621:Calculate every good friend in user's friend circle and watch more video jointly, extract preceding 10 films and make
For recommendation results;
Step S2622:Watch more video in the close friend's circle for finding user jointly, extract preceding 10 films as pushing away
Recommend result;
Step S2623:According to the propagating influence of user good friend, preceding 10 films are extracted as recommendation results.
Step S270:For showing the video chosen to user.
It is appreciated that the video recommendations to user can be realized by above-mentioned steps S210~S270.Above-mentioned video recommendations
Method can also comprise the steps:
Weights are assigned to the video that the video recommendations module is chosen according to friend circle, and are ranked up according to the weights
Generate recommendation list.Specifically, the recommendation of the maximum good friend of the recommendation results that are drawn respectively for close friend's circle, propagating influence
As a result and commonly the recommendation results of circle assign weights;Calculate the weights sum of all films appeared in recommendation results;And
Generation recommendation list is ranked up according to the weights, consequently recommended result is used as.A) the recommendation knot drawn for close friend's circle
Fruit assigns weights 1, assigns weights 1.5 for the recommendation results that the larger good friend of influence power draws, is drawn for common circle
Recommendation results assign weights 0.8;The weights sum of all films appeared in recommendation results is calculated again;Selection is gone too far after sequence
Sub maximum preceding 20 videos, are used as consequently recommended result.
Video recommendation system and method that the present invention is provided, by data obtaining module obtain include userspersonal information,
The multi-source data of the user social contact network information and video classification information, is carried out pre- by data preprocessing module to the multi-source data
Processing, then the user data library storage multi-source data after pretreatment built through data memory module, then pass through user's feature
The microblog data that analysis module is sent according to user obtains the personality and mood and social network analysis module of user to user's
Good friend's clustering, obtains the friend circle of the user, and video recommendations module is according to the personality, mood and good friend of the user
Circle, chooses the video of the user preferences and shows the video chosen to the user by front end display module.
It is understood that for the person of ordinary skill of the art, can be done with technique according to the invention design
Go out other various corresponding changes and deformation, and all these changes and deformation should all belong to the protection model of the claims in the present invention
Enclose.
Claims (15)
1. a kind of video recommendation system, it is characterised in that including:
Data obtaining module, includes many of userspersonal information, the user social contact network information and video classification information for obtaining
Source data, the userspersonal information includes age, sex, occupation, nationality, location, the video type liked, watched
List of videos, content of microblog and transmission that the friend information of the user social contact network information including user, user push
List of videos that the when and where of microblogging, user watched, the video scoring information of user, the video classification packet are included
Title, director, age, type, video scoring and video tab;
Data preprocessing module, for being pre-processed to the multi-source data, the data preprocessing module includes:Data are clear
Submodule is washed, for rejecting incomplete data in the multi-source data, hough transformation submodule, for different platform will to be come from
The form of multi-source data unified, and data integration submodule, for by the data from disparate databases after stipulations
It is integrated into identical database;
Data memory module, builds customer data base, and multi-source data after pretreatment is stored in the database;
User's characteristic analysis module, the microblog data for being sent according to user obtains the personality and mood of user;
Social network analysis module, for good friend's clustering to user, obtains the friend circle of the user;
Video recommendations module, for the personality according to the user, mood and friend circle, chooses the video of the user preferences;
And
Front end display module, for showing the video chosen to the user;
The social network analysis module is analyzed for user social contact network data, including:
The good friend of user is expressed as to the set of series of features vector, the set includes region, age, sex, occupation, happiness
Joyous video type and the list of videos watched;The similarity between good friend and good friend is calculated, and will be used according to similarity
If the good friend at family is divided into dry type;
The intimate degree of user and its good friend are calculated, its close friend's circle is obtained;
The propagating influence of user is calculated according to the microblogging quantity forwarded and forwarding quantity of user good friend:P=0.2S+0.8F,
Wherein P is the propagating influence of good friend, and S is microblogging quantity forwarded, and F is the number of times that microblogging is forwarded.
2. video recommendation system according to claim 1, it is characterised in that described information acquisition module includes:
Userspersonal information's acquisition submodule, by the user registration module of webpage version, obtains the personal information of user;
Social network information acquisition submodule, the user social contact network information is obtained by intercommunion platform;And
Video classification acquisition of information submodule, video classification information is obtained by video website.
3. video recommendation system according to claim 1, it is characterised in that the data memory module is used to use described
Family personal information storage is stored into Sqlserver or Mysql relevant database and by the user social contact network information
In MongoDB figure relational database.
4. video recommendation system according to claim 1, it is characterised in that user's characteristic analysis module includes:
User's character analysis submodule, the microblog data for being sent according to user in the past analyzes the personality of user;And
User emotion analyzes submodule, and the microblog data for being sent according to user in the past analyzes the mood of user.
5. video recommendation system according to claim 1, it is characterised in that the video recommendations module includes:
Userspersonal information's recommending module, the video film of user preferences is chosen for the personality according to user and mood;And
Social network information recommending module, the video film for choosing user preferences according to the friend circle of user.
6. video recommendation system according to claim 1, it is characterised in that also including user profile real-time capture module,
For the social network information of user in real, and it is stored in the database.
7. video recommendation system according to claim 1, it is characterised in that also including result optimizing module, for institute
The video imparting weights that video recommendations module is chosen according to friend circle are stated, and generation is ranked up according to the weights and recommend row
Table.
8. video recommendation system according to claim 7, it is characterised in that the front end display module includes webpage shape
Formula, television terminal or mobile phone terminal.
9. a kind of video recommendation method, it is characterised in that comprise the steps:
Acquisition includes the multi-source data of userspersonal information, the user social contact network information and video classification information, the user
People's information includes age, sex, occupation, nationality, location, the video type liked, the list of videos watched, the use
Content of microblog that the friend information of family social network information including user, user push and send microblogging when and where,
List of videos that user watched, the video scoring information of user, the video classification packet include title, director, age, class
Type, video scoring and video tab;
The multi-source data is pre-processed, including:Incomplete data in the multi-source data are proposed, will be from different flat
The form of the multi-source data of platform is unified, by the data integration from disparate databases after stipulations to identical database
In;
Customer data base is built, and multi-source data after pretreatment is stored in the database;
The microblog data sent according to user obtains the personality and mood of user;
To good friend's clustering of user, the friend circle of the user is obtained, the friend circle includes close friend's circle, good friend's circle
And maximum propagation influence power good friend;
According to the personality, mood and friend circle of the user, the video of the user preferences is chosen;And
For showing the video chosen to the user;
To good friend's clustering of user, the friend circle of the user is obtained, is comprised the steps:
The good friend of user is expressed as to the set of series of features vector, the set includes region, age, sex, occupation, happiness
Joyous video type and the list of videos watched;
The distance between Companion Vector is calculated to characterize the similarity between good friend and good friend, and calculated using cluster according to similarity
If good friend's automatic cluster of user is dry type by method KMeans;
Choose higher preceding 20 good friends of similarity and constitute its close friend's circle;
The propagating influence of user good friend is calculated according to the microblogging quantity forwarded and forwarding quantity of user good friend, the calculating is public
Formula is P=0.2S+0.8F, and wherein P is the propagating influence of good friend, and S is microblogging quantity forwarded, and F is the number of times that microblogging is forwarded.
10. video recommendation method according to claim 9, it is characterised in that the microblog data sent according to user is obtained
The personality and mood of user, comprises the steps:
Mathematical abstractions are the vector space of a N-dimensional, and each vector correspondence personality/mood fundamental, wherein personality base
Essentiality is:Introversion, flare, stably, it is unstable, mood fundamental is:Happy, indignation, it is sad, frightened, detest, it is surprised,
The vector space of the N-dimensional is designated as Q=[x1、x2、……];
Participle is carried out to the microblog data, semantic feature is obtained, is defined as C=[c1、c2、……];
Set up Function Mapping relation, Q=f (C), wherein, Q is the set of personality/mood fundamental, and C is microblog data participle
The semantic feature set obtained afterwards, f is corresponding mapping function;
Microblog data C is collected from the microblogging, the score of each fundamental is judged, is wanted substantially so as to obtain personality/mood
Plain Q, composing training data set;
Carry out learning to obtain model of fit for training dataset using neural network algorithm, then according to obtained model prediction
User's personality/mood.
11. video recommendation method according to claim 9, it is characterised in that become reconciled according to the personality of the user, mood
Friend's circle, chooses the video of the user preferences, comprises the steps:
The video film of user preferences is chosen according to the personality of user and mood;And
The video film of user preferences is chosen according to the friend circle of user.
12. video recommendation method according to claim 11, it is characterised in that chosen and used according to the personality of user and mood
The video film of family hobby, comprises the steps:
Every film is described with region, age, protagonist, four attributes of type;
Preference of the user for the film with some attribute is obtained according to the personality and mood of user, and assigns power
Value;
For any one film, the preference that the user corresponds to the film native is summed up, and obtains the use
Fancy grade of the family to the film;
All films are calculated, and choose 10 maximum films of fancy grade and recommend user.
13. video recommendation method according to claim 11, it is characterised in that user's happiness is chosen according to the friend circle of user
Good video film, comprises the steps:
Calculate every good friend in user's friend circle and watch more video jointly, extract preceding 10 films and be used as recommendation results;
More video is watched in the close friend's circle for finding user jointly, preceding 10 films is extracted and is used as recommendation results;
According to the propagating influence of user good friend, preceding 10 films are extracted as recommendation results.
14. video recommendation method according to claim 9, it is characterised in that also comprise the steps:The video is pushed away
The video imparting weights that module is chosen according to friend circle are recommended, and generation recommendation list is ranked up according to the weights.
15. video recommendation method according to claim 14, it is characterised in that to the video recommendations module according to good friend
The video that circle is chosen assigns weights, and is ranked up generation recommendation list according to the weights, comprises the steps:
Recommendation results, the recommendation results of the good friend of propagating influence maximum and the common circle drawn respectively for close friend's circle
Recommendation results assign weights;
Calculate the weights sum of all films appeared in recommendation results;
And generation recommendation list is ranked up according to the weights, it is used as consequently recommended result.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310684807.7A CN103714130B (en) | 2013-12-12 | 2013-12-12 | Video recommendation system and method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310684807.7A CN103714130B (en) | 2013-12-12 | 2013-12-12 | Video recommendation system and method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN103714130A CN103714130A (en) | 2014-04-09 |
CN103714130B true CN103714130B (en) | 2017-08-22 |
Family
ID=50407105
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201310684807.7A Active CN103714130B (en) | 2013-12-12 | 2013-12-12 | Video recommendation system and method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN103714130B (en) |
Families Citing this family (52)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105224576A (en) * | 2014-07-01 | 2016-01-06 | 上海视畅信息科技有限公司 | A kind of video display intelligent recommendation method |
CN106156030A (en) | 2014-09-18 | 2016-11-23 | 华为技术有限公司 | The method and apparatus that in social networks, information of forecasting is propagated |
CN105824818A (en) * | 2015-01-04 | 2016-08-03 | 中国移动通信集团河北有限公司 | Informationized management method, platform and system |
CN106034256B (en) * | 2015-03-10 | 2019-11-05 | 腾讯科技(北京)有限公司 | Video social contact method and device |
CN104765778A (en) * | 2015-03-18 | 2015-07-08 | 百度在线网络技术(北京)有限公司 | Method and device for providing information to be transmitted based on user behaviors |
CN104834969A (en) * | 2015-05-05 | 2015-08-12 | 东南大学 | Film evaluation prediction method and system |
CN106202103A (en) * | 2015-05-06 | 2016-12-07 | 阿里巴巴集团控股有限公司 | Music recommends method and apparatus |
WO2016187768A1 (en) * | 2015-05-25 | 2016-12-01 | 武克易 | Video information pushing method and apparatus |
CN106339146A (en) * | 2015-07-09 | 2017-01-18 | 华为终端(东莞)有限公司 | Method and device for recommending applications |
CN105138551A (en) * | 2015-07-14 | 2015-12-09 | 青岛海信传媒网络技术有限公司 | Method and apparatus for obtaining user interest tag |
CN105022699B (en) * | 2015-07-14 | 2018-04-24 | 惠龙易通国际物流股份有限公司 | The preprocess method and system of buffer area data |
CN105095516B (en) * | 2015-09-16 | 2019-02-15 | 中国传媒大学 | The radio and television tenant group system and method integrated based on spectral clustering |
CN105915954A (en) * | 2015-10-30 | 2016-08-31 | 乐视移动智能信息技术(北京)有限公司 | Video recommendation method based on mobile phone screen and video recommendation system thereof |
CN105915957A (en) * | 2015-12-15 | 2016-08-31 | 乐视致新电子科技(天津)有限公司 | Intelligent television playing content display method, device and system |
US10824941B2 (en) * | 2015-12-23 | 2020-11-03 | The Toronto-Dominion Bank | End-to-end deep collaborative filtering |
CN106528584B (en) * | 2016-02-15 | 2019-10-29 | 中山大学 | A kind of group recommending method based on ensemble learning |
CN105740473B (en) * | 2016-03-14 | 2021-03-02 | 腾讯科技(深圳)有限公司 | User generated content display method and device |
CN105704231B (en) * | 2016-03-17 | 2019-04-30 | 珠海格力电器股份有限公司 | Information pushing method and device for network playing resources |
CN105930425A (en) * | 2016-04-18 | 2016-09-07 | 乐视控股(北京)有限公司 | Personalized video recommendation method and apparatus |
CN105930532B (en) * | 2016-06-16 | 2019-08-02 | 上海聚力传媒技术有限公司 | A kind of method and apparatus from multimedia resource to user that recommending |
CN106131684A (en) * | 2016-06-24 | 2016-11-16 | 依偎科技(南昌)有限公司 | A kind of content recommendation method and terminal |
CN106202252A (en) * | 2016-06-29 | 2016-12-07 | 厦门趣处网络科技有限公司 | Method, system are recommended in a kind of trip analyzed based on user emotion |
CN106201184A (en) * | 2016-06-29 | 2016-12-07 | 腾讯科技(深圳)有限公司 | Edit methods, device and the terminal of a kind of SNS message |
WO2018023329A1 (en) * | 2016-08-01 | 2018-02-08 | Linkedin Corporation | Quality industry content mixed with friend's posts in social network |
CN106202570A (en) * | 2016-08-11 | 2016-12-07 | 乐视控股(北京)有限公司 | A kind of user information acquiring method and device |
CN106649655A (en) * | 2016-12-13 | 2017-05-10 | 宁夏宁信信息科技有限公司 | Personalized recommending method and system in video app |
CN106909907A (en) * | 2017-03-07 | 2017-06-30 | 佛山市融信通企业咨询服务有限公司 | A kind of video communication sentiment analysis accessory system |
US10521482B2 (en) | 2017-04-24 | 2019-12-31 | Microsoft Technology Licensing, Llc | Finding members with similar data attributes of a user for recommending new social connections |
CN108959323B (en) * | 2017-05-25 | 2021-12-07 | 腾讯科技(深圳)有限公司 | Video classification method and device |
CN107918652B (en) * | 2017-11-15 | 2020-10-02 | 浙江大学 | Method for recommending movies based on social relations by utilizing multi-modal network learning |
CN107948732B (en) * | 2017-12-04 | 2020-12-01 | 京东方科技集团股份有限公司 | Video playing method, video playing device and video playing system |
CN108632671A (en) * | 2018-03-29 | 2018-10-09 | 北京恒信彩虹信息技术有限公司 | A kind of recommendation method and system |
CN108681601B (en) * | 2018-05-21 | 2020-12-11 | 北京奇艺世纪科技有限公司 | Video sharing method and device and electronic equipment |
CN108848152B (en) * | 2018-06-05 | 2021-09-21 | 腾讯科技(深圳)有限公司 | Object recommendation method and server |
CN109002490B (en) * | 2018-06-26 | 2020-09-04 | 腾讯科技(北京)有限公司 | User portrait generation method, device, server and storage medium |
CN109104620B (en) * | 2018-07-26 | 2020-05-19 | 腾讯科技(深圳)有限公司 | Short video recommendation method and device and readable medium |
CN109245989A (en) * | 2018-08-15 | 2019-01-18 | 咪咕动漫有限公司 | Processing method and device based on information sharing and computer readable storage medium |
CN109698820A (en) * | 2018-09-03 | 2019-04-30 | 长安通信科技有限责任公司 | A kind of domain name Similarity measures and classification method and system |
CN109783724A (en) * | 2018-12-14 | 2019-05-21 | 深圳壹账通智能科技有限公司 | Management method, terminal device and the medium of social network information |
CN109783686A (en) * | 2019-01-21 | 2019-05-21 | 广州虎牙信息科技有限公司 | Behavioral data processing method, device, terminal device and storage medium |
CN109902753B (en) * | 2019-03-06 | 2023-01-13 | 深圳市珍爱捷云信息技术有限公司 | User recommendation model training method and device, computer equipment and storage medium |
CN109933700A (en) * | 2019-03-07 | 2019-06-25 | 王芃翰 | Students ' reading based on big data suggests generation method and relevant device |
CN110096613B (en) * | 2019-04-12 | 2021-07-20 | 北京奇艺世纪科技有限公司 | Video recommendation method and device, electronic equipment and storage medium |
TWI716033B (en) * | 2019-07-15 | 2021-01-11 | 李姿慧 | Video Score Intelligent System |
CN110647678B (en) * | 2019-09-02 | 2022-11-15 | 杭州数理大数据技术有限公司 | Recommendation method based on user character label |
CN110837598B (en) * | 2019-11-11 | 2021-03-19 | 腾讯科技(深圳)有限公司 | Information recommendation method, device, equipment and storage medium |
CN111859102A (en) * | 2020-02-17 | 2020-10-30 | 北京嘀嘀无限科技发展有限公司 | Prompt information determination method, system, medium and storage medium |
CN111680224A (en) * | 2020-04-22 | 2020-09-18 | 威比网络科技(上海)有限公司 | Cross-platform course pushing method and device, electronic equipment and storage medium |
CN112365447B (en) * | 2020-10-20 | 2022-08-19 | 四川长虹电器股份有限公司 | Multidimensional movie and television scoring method |
CN113204577A (en) * | 2021-04-15 | 2021-08-03 | 北京沃东天骏信息技术有限公司 | Information pushing method and device, electronic equipment and computer readable medium |
CN113239041A (en) * | 2021-05-13 | 2021-08-10 | 大连交通大学 | Computer big data processing acquisition system and method |
CN116739814B (en) * | 2023-04-23 | 2024-05-14 | 广州市疾病预防控制中心(广州市卫生检验中心、广州市食品安全风险监测与评估中心、广州医科大学公共卫生研究院) | Method for preventing disease transmission and social platform |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102184235A (en) * | 2011-05-13 | 2011-09-14 | 广州星海传媒有限公司 | Set top box-based digital television program recommending method and system |
CN103327045A (en) * | 2012-03-21 | 2013-09-25 | 腾讯科技(深圳)有限公司 | User recommendation method and system in social network |
CN103327400A (en) * | 2012-03-22 | 2013-09-25 | 鸿富锦精密工业(深圳)有限公司 | Customer premise equipment and method for creating social video channel |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8433670B2 (en) * | 2011-03-03 | 2013-04-30 | Xerox Corporation | System and method for recommending items in multi-relational environments |
CN102571978A (en) * | 2012-02-10 | 2012-07-11 | 上海视畅信息科技有限公司 | Intelligent video viewing system |
-
2013
- 2013-12-12 CN CN201310684807.7A patent/CN103714130B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102184235A (en) * | 2011-05-13 | 2011-09-14 | 广州星海传媒有限公司 | Set top box-based digital television program recommending method and system |
CN103327045A (en) * | 2012-03-21 | 2013-09-25 | 腾讯科技(深圳)有限公司 | User recommendation method and system in social network |
CN103327400A (en) * | 2012-03-22 | 2013-09-25 | 鸿富锦精密工业(深圳)有限公司 | Customer premise equipment and method for creating social video channel |
Also Published As
Publication number | Publication date |
---|---|
CN103714130A (en) | 2014-04-09 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103714130B (en) | Video recommendation system and method | |
CN108959603B (en) | Personalized recommendation system and method based on deep neural network | |
Vargas-Govea et al. | Effects of relevant contextual features in the performance of a restaurant recommender system | |
CN104866474B (en) | Individuation data searching method and device | |
CN105426528B (en) | A kind of retrieval ordering method and system of commodity data | |
WO2018041168A1 (en) | Information pushing method, storage medium and server | |
Cufoglu | User profiling-a short review | |
KR101579376B1 (en) | Personalized place recommendation system and method by using subjectivity analysis for user classification | |
JP5386663B1 (en) | Information processing apparatus, information processing method, information processing program, and recording medium | |
KR102227552B1 (en) | System for providing context awareness algorithm based restaurant sorting personalized service using review category | |
US8997008B2 (en) | System and method for searching through a graphic user interface | |
US20140288999A1 (en) | Social character recognition (scr) system | |
CN110737834A (en) | Business object recommendation method and device, storage medium and computer equipment | |
Lee et al. | A hybrid collaborative filtering-based product recommender system using search keywords | |
CN112380451A (en) | Favorite content recommendation method based on big data | |
CN115408618B (en) | Point-of-interest recommendation method based on social relation fusion position dynamic popularity and geographic features | |
CN115878903B (en) | Information intelligent recommendation method based on big data | |
CN115712780A (en) | Information pushing method and device based on cloud computing and big data | |
CN106708871A (en) | Method and device for identifying social service characteristics user | |
CN110083766B (en) | Query recommendation method and device based on meta-path guiding embedding | |
CN115168700A (en) | Information flow recommendation method, system and medium based on pre-training algorithm | |
WO2020111329A1 (en) | Automatic answering method and system using similar user matching | |
Ashraf et al. | Personalized news recommendation based on multi-agent framework using social media preferences | |
CN117670439A (en) | Restaurant recommendation method and system based on user portrait | |
CN117436956A (en) | Intelligent marketing management system based on advertisement pushing |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |