CN110457517A - The implementation method of the similar suggested design of program request based on picture similitude - Google Patents
The implementation method of the similar suggested design of program request based on picture similitude Download PDFInfo
- Publication number
- CN110457517A CN110457517A CN201910764033.6A CN201910764033A CN110457517A CN 110457517 A CN110457517 A CN 110457517A CN 201910764033 A CN201910764033 A CN 201910764033A CN 110457517 A CN110457517 A CN 110457517A
- Authority
- CN
- China
- Prior art keywords
- poster
- program
- request
- feature vector
- database
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/55—Clustering; Classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
- G06F16/58—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/583—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
-
- 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
- G06F16/73—Querying
- G06F16/738—Presentation of query results
- G06F16/739—Presentation of query results in form of a video summary, e.g. the video summary being a video sequence, a composite still image or having synthesized frames
-
- 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
- G06F16/783—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using metadata automatically derived from the content
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
- H04N21/20—Servers specifically adapted for the distribution of content, e.g. VOD servers; Operations thereof
- H04N21/25—Management operations performed by the server for facilitating the content distribution or administrating data related to end-users or client devices, e.g. end-user or client device authentication, learning user preferences for recommending movies
- H04N21/251—Learning process for intelligent management, e.g. learning user preferences for recommending movies
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
- H04N21/60—Network structure or processes for video distribution between server and client or between remote clients; Control signalling between clients, server and network components; Transmission of management data between server and client, e.g. sending from server to client commands for recording incoming content stream; Communication details between server and client
- H04N21/61—Network physical structure; Signal processing
- H04N21/6106—Network physical structure; Signal processing specially adapted to the downstream path of the transmission network
- H04N21/6125—Network physical structure; Signal processing specially adapted to the downstream path of the transmission network involving transmission via Internet
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- Multimedia (AREA)
- Data Mining & Analysis (AREA)
- Library & Information Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Signal Processing (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The present invention discloses the implementation method of the similar suggested design of program request based on picture similitude, is related to intelligent dibbling technical field;Establish the poster database of all request program posters of OTT VOD system, feature extraction is carried out to poster using deep learning, establish poster feature database, and feature vector is converted by feature, most like poster is obtained by calculating the cosine similarity of different characteristic vector, carry out program request recommendation, compared with prior art, the matchmaker of program request record and program that the present invention does not need user provides label information and carries out collaborative filtering calculating, processing calculating only is carried out to request program poster, difference is distinguished from direction using cosine similarity, and it is insensitive to absolute numerical value, the initiation problem of numerically gap can be effectively avoided on the characteristic value similar calculating of picture the problem of, improve the accuracy of recommendation, enhance user's viscosity, improve user experience.
Description
Technical field
The present invention discloses the implementation method of the similar suggested design of program request based on picture similitude, is related to intelligent dibbling technology
Field.
Background technique
With the fast development of Internet technology, OTT set-top box technology is flourished.But it is simple in OTT set-top box
Single list search formula VOD system has been unable to meet the completely new demand of user.Most of user more thinks in on-demand process
The similar film of viewing film is got, is based on this, program request recommender system is come into being, but existing program request recommender system is usual
It is realized and is recommended using collaborative filtering method, frequently result in popular film by using data to find similar user and film
It is easier than minority film recommended;Simultaneously as the use data that the film newly shown is not too many, therefore utilize and cooperateed with
Filtering to user recommends any New cinema very unrealistic.
The present invention provides the implementation method of the similar suggested design of program request based on picture similitude, and using being based on, picture is similar
Property the similar suggested design of program request do not need user program request record and program matchmaker provide label information carry out collaborative filtering calculating,
Processing calculating only is carried out to request program poster and gets similar request program, the accuracy of recommendation is improved, enhances user
Viscosity improves user experience.
Summary of the invention
The present invention is directed to problem of the prior art, provides the realization side of the similar suggested design of program request based on picture similitude
Method carries out processing calculating to request program poster and gets similar request program, improves the accuracy of recommendation, enhance user
Viscosity improves user experience.
Concrete scheme proposed by the present invention is:
The implementation method of the similar suggested design of program request based on picture similitude: request program in OTT VOD system is obtained
Poster establish poster database,
Characteristics extraction is carried out to poster in poster database based on deep learning algorithm, establishes poster feature database, it will be special
Value indicative is converted into feature vector, and traversal calculates the cosine similarity of feature vector, most like sea is determined by cosine similarity
Report carries out request program recommendation.
The formula of the cosine similarity of feature vector is calculated in the implementation method are as follows:
A, B respectively indicate the feature vector of two different posters, and Ai indicates that each component of A feature vector, Bi indicate B feature
Each component of vector.
The calculating threshold value that cosine similarity is controlled in the implementation method is determined by the calculating threshold value of cosine similarity
Most like poster obtains the request program list of the poster.
By timed task in the implementation method, in time by the poster of the new request program that OTT VOD system is added
It is sophisticated to poster database.
The poster and request program id that request program in OTT VOD system is obtained in the implementation method, according to program request
Program id and poster corresponding relationship establish poster database.
The realization system of the similar suggested design of program request based on picture similitude include recommendation unit, OTT VOD system and
Client,
The poster that recommendation unit obtains request program in OTT VOD system establishes poster database,
Characteristics extraction is carried out to poster in poster database based on deep learning algorithm, establishes poster feature database, it will be special
Value indicative is converted into feature vector, and traversal calculates the cosine similarity of feature vector, sees by the way that cosine similarity is determining with client
The most like poster seen, recommends client for request program.
Recommendation unit calculates the formula of the cosine similarity of feature vector in the realization system are as follows:
A, B respectively indicate the feature vector of two different posters, and Ai indicates that each component of A feature vector, Bi indicate B feature
Each component of vector.
The calculating threshold value of recommendation unit control cosine similarity, passes through the calculating of cosine similarity in the realization system
Threshold value determines most like poster, obtains the request program list of the poster.
Recommendation unit is by timed task in the realization system, in time by the new program request section that OTT VOD system is added
Purpose poster is sophisticated to poster database.
Recommendation unit obtains the poster and request program id of request program in OTT VOD system in the realization system,
Poster database is established according to request program id and poster corresponding relationship.
Usefulness of the present invention is:
The present invention provides the implementation method of the similar suggested design of program request based on picture similitude, establishes OTT program request system
It unites the poster databases of all request program posters, feature extraction is carried out to poster using deep learning, establishes poster feature database,
And feature vector is converted by feature, most like poster is obtained by calculating the cosine similarity of different characteristic vector, into
Row program request is recommended, and compared with prior art, the matchmaker of program request record and program that the present invention does not need user provides label information and carries out
Collaborative filtering calculates, and only carries out processing calculating to request program poster, distinguishes difference from direction using cosine similarity, and right
Absolute numerical value is insensitive, the characteristic value similar calculating of picture the problem of on can effectively avoid the initiation of numerically gap
Problem, improves the accuracy of recommendation, and enhancing user's viscosity improves user experience.
Detailed description of the invention
Fig. 1 is the method for the present invention flow diagram;
Fig. 2 is present system topology schematic diagram;
Fig. 3 is film poster schematic diagram present in OTT VOD system.
Specific embodiment
The implementation method of program request similar suggested design of the present invention offer based on picture similitude: OTT VOD system is obtained
The poster of middle request program establishes poster database,
Characteristics extraction is carried out to poster in poster database based on deep learning algorithm, establishes poster feature database, it will be special
Value indicative is converted into feature vector, and traversal calculates the cosine similarity of feature vector, most like sea is determined by cosine similarity
Report carries out request program recommendation.
The realization system packet of the similar suggested design of the program request based on picture similitude corresponding with the above method is provided simultaneously
Recommendation unit, OTT VOD system and client are included,
The poster that recommendation unit obtains request program in OTT VOD system establishes poster database,
Characteristics extraction is carried out to poster in poster database based on deep learning algorithm, establishes poster feature database, it will be special
Value indicative is converted into feature vector, and traversal calculates the cosine similarity of feature vector, sees by the way that cosine similarity is determining with client
The most like poster seen, recommends client for request program.
The present invention will be further explained below with reference to the attached drawings and specific examples, so that those skilled in the art can be with
It more fully understands the present invention and can be practiced, but illustrated embodiment is not as a limitation of the invention.
When carrying out program request recommendation using the method for the present invention, the request program poster and program request section of OTT VOD system are first obtained
Mesh id establishes poster database according to request program id and poster corresponding relationship;
Characteristics extraction is carried out to poster in poster database based on deep learning algorithm, establishes poster feature database, it will be special
Value indicative is converted into feature vector, and traversal calculates the cosine similarity of feature vector, wherein the formula of cosine similarity are as follows:
A, B respectively indicate the feature vector of two different posters, and Ai indicates that each component of A feature vector, Bi indicate B feature
Each component of vector converts the picture feature value of extraction to the spy of one 25800 dimension for each poster in poster library
Vector is levied, poster in poster database is successively then calculated into similarity, such as the poster picture pair of program X and program Y two-by-two
The feature vector answered is A respectively1,A2... ..., A25800And B1,B2... ..., B25800, then their similarity formula are as follows:
Value range by the cosine similarity of feature vector is between [0,1], and value more becomes close to 1, represent two to
Amount similarity is closer, i.e., two pictures are more similar;More become close to 0, it is smaller, more independent to represent two vector similarities, i.e.,
Two pictures are more dissimilar,
It determines most like poster by cosine similarity, carries out request program recommendation, with reference in Fig. 3, when user watches
When animated film, recommending a part in the request program of user using the method for the present invention accordingly is to correspond to poster in Fig. 3
Film, the film that poster is corresponded in Fig. 3 is present in request program list.
It, can also be by timed task, in time by the new point that OTT VOD system is added using the method for the present invention in the above process
The poster for broadcasting program is sophisticated to poster database, periodically carries out feature using machine learning algorithm to the poster picture in poster library and mentions
It takes, the pretreatment operations such as vector conversion obtain the feature vector of poster picture, carry out cosine similarity meter to poster feature vector
It calculates, obtains most like poster, carry out request program recommendation.
When carrying out program request recommendation using present system, recommendation unit first obtains the request program poster of OTT VOD system
With request program id, poster database is established according to request program id and poster corresponding relationship;
Characteristics extraction is carried out to poster in poster database based on deep learning algorithm, establishes poster feature database, it will be special
Value indicative is converted into feature vector, and traversal calculates the cosine similarity of feature vector, wherein the formula of cosine similarity are as follows:
A, B respectively indicate the feature vector of two different posters, and Ai indicates that each component of A feature vector, Bi indicate B feature
Each component of vector converts the picture feature value of extraction to the spy of one 25800 dimension for each poster in poster library
Vector is levied, poster in poster database is successively then calculated into similarity, such as the poster picture pair of program X and program Y two-by-two
The feature vector answered is A respectively1,A2... ..., A25800And B1,B2... ..., B25800, then their similarity formula are as follows:
The value range of the cosine similarity of feature vector is between [0,1] by recommendation unit, and value more becomes close to 1, generation
Two vector similarities of table are closer, i.e., two pictures are more similar;More become close to 0, it is smaller, more to represent two vector similarities
Independent, i.e., two pictures are more dissimilar,
Recommendation unit determines most like poster by cosine similarity, carries out request program recommendation, reference to client
In Fig. 3, when user watches animated film, a part is recommended in the request program of user i.e. using present system accordingly
For the film for corresponding to poster in Fig. 3, the film that poster is corresponded in Fig. 3 is present in request program list.
It, can also be by timed task, in time by the new point that OTT VOD system is added using present system in the above process
The poster for broadcasting program is sophisticated to poster database, periodically carries out feature using machine learning algorithm to the poster picture in poster library and mentions
It takes, the pretreatment operations such as vector conversion obtain the feature vector of poster picture, carry out cosine similarity meter to poster feature vector
It calculates, obtains most like poster, request program recommendation is carried out to client.
Embodiment described above is only to absolutely prove preferred embodiment that is of the invention and being lifted, protection model of the invention
It encloses without being limited thereto.Those skilled in the art's made equivalent substitute or transformation on the basis of the present invention, in the present invention
Protection scope within.Protection scope of the present invention is subject to claims.
Claims (10)
1. the implementation method of the similar suggested design of program request based on picture similitude, it is characterized in that obtaining OTT VOD system midpoint
The poster for broadcasting program establishes poster database,
Characteristics extraction is carried out to poster in poster database based on deep learning algorithm, poster feature database is established, by characteristic value
It is converted into feature vector, traversal calculates the cosine similarity of feature vector, most like poster is determined by cosine similarity, into
Row request program is recommended.
2. implementation method according to claim 1, it is characterized in that calculating the formula of the cosine similarity of feature vector are as follows:
A, B respectively indicate the feature vector of two different posters, AiIndicate each component of A feature vector, BiIndicate B feature vector
Each component.
3. implementation method according to claim 1 or 2, it is characterized in that the calculating threshold value of control cosine similarity, by remaining
The calculating threshold value of string similarity determines most like poster, obtains the request program list of the poster.
4. implementation method according to claim 3, it is characterized in that OTT program request system will newly be added in time by timed task
The poster of the request program of system is sophisticated to poster database.
5. implementation method according to claim 4, it is characterized in that obtaining the poster and point of request program in OTT VOD system
Program id is broadcast, poster database is established according to request program id and poster corresponding relationship.
6. the realization system of the similar suggested design of program request based on picture similitude, it is characterized in that including recommendation unit, OTT program request
System and client,
The poster that recommendation unit obtains request program in OTT VOD system establishes poster database,
Characteristics extraction is carried out to poster in poster database based on deep learning algorithm, poster feature database is established, by characteristic value
It is converted into feature vector, traversal calculates the cosine similarity of feature vector, passes through cosine similarity determination and client viewing
Request program is recommended client by most like poster.
7. realization system according to claim 6, it is characterized in that recommendation unit calculates the cosine similarity of feature vector
Formula are as follows:
A, B respectively indicate the feature vector of two different posters, and Ai indicates that each component of A feature vector, Bi indicate B feature vector
Each component.
8. realization system according to claim 6 or 7, it is characterized in that the calculating threshold of recommendation unit control cosine similarity
Value, determines most like poster by the calculating threshold value of cosine similarity, obtains the request program list of the poster.
9. realization system according to claim 8, it is characterized in that recommendation unit will be newly added in time by timed task
The poster of the request program of OTT VOD system is sophisticated to poster database.
10. realization system according to claim 9, it is characterized in that recommendation unit obtains request program in OTT VOD system
Poster and request program id, poster database is established according to request program id and poster corresponding relationship.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910764033.6A CN110457517A (en) | 2019-08-19 | 2019-08-19 | The implementation method of the similar suggested design of program request based on picture similitude |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910764033.6A CN110457517A (en) | 2019-08-19 | 2019-08-19 | The implementation method of the similar suggested design of program request based on picture similitude |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110457517A true CN110457517A (en) | 2019-11-15 |
Family
ID=68487559
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910764033.6A Pending CN110457517A (en) | 2019-08-19 | 2019-08-19 | The implementation method of the similar suggested design of program request based on picture similitude |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110457517A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112905894A (en) * | 2021-03-24 | 2021-06-04 | 合肥工业大学 | Collaborative filtering recommendation method based on enhanced graph learning |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102421016A (en) * | 2011-12-20 | 2012-04-18 | 深圳市同洲视讯传媒有限公司 | Method, system and terminal for acquiring request program information |
CN106169083A (en) * | 2016-07-05 | 2016-11-30 | 广州市香港科大霍英东研究院 | The film of view-based access control model feature recommends method and system |
CN107291845A (en) * | 2017-06-02 | 2017-10-24 | 北京邮电大学 | A kind of film based on trailer recommends method and system |
CN109034953A (en) * | 2018-07-02 | 2018-12-18 | 西南交通大学 | A kind of film recommended method |
-
2019
- 2019-08-19 CN CN201910764033.6A patent/CN110457517A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102421016A (en) * | 2011-12-20 | 2012-04-18 | 深圳市同洲视讯传媒有限公司 | Method, system and terminal for acquiring request program information |
CN106169083A (en) * | 2016-07-05 | 2016-11-30 | 广州市香港科大霍英东研究院 | The film of view-based access control model feature recommends method and system |
CN107291845A (en) * | 2017-06-02 | 2017-10-24 | 北京邮电大学 | A kind of film based on trailer recommends method and system |
CN109034953A (en) * | 2018-07-02 | 2018-12-18 | 西南交通大学 | A kind of film recommended method |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112905894A (en) * | 2021-03-24 | 2021-06-04 | 合肥工业大学 | Collaborative filtering recommendation method based on enhanced graph learning |
CN112905894B (en) * | 2021-03-24 | 2022-08-19 | 合肥工业大学 | Collaborative filtering recommendation method based on enhanced graph learning |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104199896B (en) | The video similarity of feature based classification is determined and video recommendation method | |
CN107852252B (en) | Fingerprint matching is recognized by automated content to change to detect channel | |
KR102405115B1 (en) | Using machine learning to recommend live-stream content | |
CN107944629B (en) | Recommendation method and device based on heterogeneous information network representation | |
CN104965889B (en) | Content recommendation method and device | |
CN104935963B (en) | A kind of video recommendation method based on timing driving | |
CN111310045A (en) | Network-embedded movie recommendation method based on meta-path | |
CN105913273A (en) | Intelligent recommendation all-terminal display method and device | |
CN103517087B (en) | Stream editing method, system and device, storage method, system and providing system | |
CN103955535A (en) | Individualized recommending method and system based on element path | |
CN102929928A (en) | Multidimensional-similarity-based personalized news recommendation method | |
CN101419606B (en) | Semi-automatic image labeling method based on semantic and content | |
CN106407418A (en) | A face identification-based personalized video recommendation method and recommendation system | |
CN107071578A (en) | IPTV program commending methods | |
CN103365936A (en) | Video recommendation system and method thereof | |
CN103426003A (en) | Implementation method and system for enhancing real interaction | |
CN104008139B (en) | The creation method and device of video index table, the recommendation method and apparatus of video | |
CN103886001A (en) | Personalized commodity recommendation system | |
CN105045818A (en) | Picture recommending method, apparatus and system | |
CN103686344A (en) | Enhanced video system and method | |
CN102523511A (en) | Network program aggregation and recommendation system and network program aggregation and recommendation method | |
CN104394471A (en) | Method for intelligently recommending favorite program to user | |
Li et al. | Content-based filtering recommendation algorithm using HMM | |
CN105282616A (en) | Combined collaborative filtering IPTV program recommendation method | |
CN103500228A (en) | Similarity measuring method improved through collaborative filtering recommendation algorithm |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20191115 |
|
RJ01 | Rejection of invention patent application after publication |