CN101763351A - Data fusion based video program recommendation method - Google Patents
Data fusion based video program recommendation method Download PDFInfo
- Publication number
- CN101763351A CN101763351A CN200810207608A CN200810207608A CN101763351A CN 101763351 A CN101763351 A CN 101763351A CN 200810207608 A CN200810207608 A CN 200810207608A CN 200810207608 A CN200810207608 A CN 200810207608A CN 101763351 A CN101763351 A CN 101763351A
- Authority
- CN
- China
- Prior art keywords
- user
- program
- behavioral data
- data
- video
- 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
Images
Abstract
A data fusion based video program recommendation method comprises the following steps: 1) collection of behavioral data of users: collecting the behavioral data of the users in browsing video websites and storing the behavioral data in a database; 2) data fusion: converting the behavioral data of a user in browsing a video program into the interest value of the user in the video program; and 3) program recommendation: according to the interest value of the user in a few programs, recommending the programs in which the user has high interest value to other users having the same favorites, wherein the behavioral data in browsing comprise behavioral data in playing. The duration ratio r is obtained by dividing the duration of the programs played by the users by the total duration of the programs. The higher the duration ratio r is, the higher the interest value is. As long as the users play the programs and other behavioral data of the users related to the interest value are fused, the program lists to be recommended can be generated more accurately and the degree of satisfaction of the users with the recommended programs is greatly improved.
Description
Technical field
The present invention relates to video program recommendation method.
Background technology
Along with the development of video network, watch video frequency program to become one of online purpose of numerous network users by network.How to attract the user to greatest extent, become the major issue that to consider of video website.A kind of effective method is to recommend video frequency program according to user's the hobby of watching to the user.
By analysis, the user has the hobby of watching certain class program usually.For example, the user who has likes watching sports cast, and the user who has then likes watching entertainment, and the user who has only likes watching time program or the like.
(Proceedings of the 10 in the tenth Internet meeting
ThInternational conference onWorld Wide Web), people such as Badru and George has proposed also to mention a kind of collaborative filtering recommending system based on user (User) simultaneously based on the collaborative filtering recommending system of project (Item) (Item-based Collaborative Filtering Recommendation Algorithms) in this piece article.The ultimate principle of these two kinds of filtering systems is as follows: according to known user the fancy grade of some program is calculated, recommend other hobby programs of known users then to the user with identical hobby.This fancy grade is by dividing value representation, and this score value is selected by the user, and for example certain program is disliked in expression in 1 fen, and expression in 5 fens enjoys a lot certain program.
Yet in video website, adopt this method carry out program commending the time, but run into many difficulties.For example: one, the user is unwilling painstakingly to go to give a mark to program.Usually the user is to watch video to the purpose of video website, and they are unwilling painstakingly to go to a program marking.Even if two have a spot of user to give a mark to video frequency program, these number of programs of being given a mark are very little of the ratio of all video frequency programs, usually less than 1%.Above-mentioned these difficulties can cause above-mentioned collaborative filtering recommending system " cold start-up problem " to occur, (seeing also " based on the recommendation of classification---a kind of method that solves cold start-up problem in collaborative the recommendation " that be published in " computer research and development "), this cold start-up problem can cause the commending system can't operate as normal, produces wrong recommendation results.
Summary of the invention
At the deficiency in the existing commending system, the invention provides a kind of video program recommendation method based on data fusion, this method is at first collected with interest-degree relevant the browse behavior of user on video website, after browsing behavior and marking by analysis then, carry out data fusion, score value after the fusion, just certain user carries out the recommendation of program to the interest level of certain video frequency program according to this interest level.The present invention specifically is achieved through the following technical solutions:
A kind of video program recommendation method based on data fusion may further comprise the steps:
1) user behavior data is collected step, collects the browse behavioral data of user in video website, and deposits in the database;
2) data fusion step is converted into the interest level of this user to certain video frequency program with certain user to the behavioral data of browsing of certain video frequency program;
3) program commending step according to the interest level of user to some program, is recommended this user's the high program of other interest level to other users with identical hobby;
Wherein, browse behavioral data and be included as the broadcast behavioral data, the duration of user's broadcast program is obtained duration ratio r divided by the length of this program, duration ratio r is big more, and interest level is high more.
Further, the time span of collection data is a time period before in a certain moment in the described step 1).
Further, the interest level described step 2) is that the interior behavioral data conversion of browsing of a time period before a certain moment obtains.
Further, adopt regular mode to recommend this user's the high program of other interest-degrees to other users in the described step 3) with identical hobby.
Further, described step 2) be capable with Customs Assigned Number in, video frequency program is numbered row, and the user is a content to the interest level of video frequency program, forms the interest level form of the user of a two dimension to video frequency program; In the described step 3), on the basis of this interest level table, adopt recommend method that known users and unknown subscriber are produced programs recommended tabulation, wherein, have the user of corresponding interest level to be known users, otherwise be the unknown subscriber based on collaborative filtering.
Further, also comprise the program that other users had watched in the removal recommend programs in the described step 3).
Further, the described behavioral data of browsing also comprises in the following behavior one, a plurality of or whole:
To program marking behavioral data, favorites behavioral data, dig behavioral data, bury behavioral data or upload the program behavioral data.
The present invention has avoided the user under the situation of being unwilling painstakingly to go to give a mark to program, causes above-mentioned collaborative filtering recommending system the problem of cold start-up to occur, as long as the user has the broadcast behavior, just can produce correct recommendation results.And merged the user other with interest-degree relevant browse behavioral data after, can produce the recommend programs tabulation more exactly, the technical scheme of painstakingly marking by this user of need not, improved the user greatly to programs recommended satisfaction, reduced video server and blindly clicked the burden that causes because of the user.
Description of drawings
Fig. 1 is the video program recommendation method embodiment process flow diagram that the present invention is based on data fusion.
Embodiment
As shown in Figure 1, a kind of video program recommendation method based on data fusion may further comprise the steps:
1) user behavior data is collected step, collects the browse behavioral data of user in video website, and deposits in the database;
2) data fusion step is converted into the interest level of this user to certain video frequency program with certain user to the behavioral data of browsing of certain video frequency program;
3) program commending step according to the interest level of user to some program, is recommended this user's the high program of other interest level to other users with identical hobby;
Wherein, browse behavioral data and be included as the broadcast behavioral data, the duration of user's broadcast program is obtained duration ratio r divided by the length of this program, duration ratio r is big more, and interest level is high more.
Wherein, the time span of collection data is a time period before in a certain moment in the described step 1).The user is at the behavioral data of browsing of video website in for example nearest one month to three months.
Wherein, the interest level described step 2) is that the interior behavioral data conversion of browsing of a time period before a certain moment obtains.
Wherein, adopt regular mode to recommend this user's the high program of other interest-degrees to other users in the described step 3) with identical hobby.
Wherein, described step 2) be capable with Customs Assigned Number in, video frequency program is numbered row, and the user is a content to the interest level of video frequency program, forms the interest level form of the user of a two dimension to video frequency program; In the described step 3), on the basis of this interest level table, adopt recommend method that known users and unknown subscriber are produced programs recommended tabulation, wherein, have the user of corresponding interest level to be known users, otherwise be the unknown subscriber based on collaborative filtering.
Wherein, also comprise the program that other users had watched in the removal recommend programs in the described step 3).
Wherein, the described behavioral data of browsing also comprises in the following behavior one, a plurality of or whole:
To program marking behavioral data, favorites behavioral data, dig behavioral data, bury behavioral data or upload the program behavioral data.
Wherein, described step 2) in the data fusion step, for the behavior of browsing of same type, the score value of the behavior that the back occurs covers the score value of the behavior that occurs earlier.
Wherein, the order of data fusion is followed successively by: play behavioral data, to program marking behavior, favorites behavior, dig (or top step on) behavior bury behavior (or top step on) or upload the program behavior.
Adopt the numeric representation interest level of 1,2,3,4,5 five discretize in the present embodiment.Wherein 1 expression is very disagreeable, and 2 expressions are general disagreeable, and 3 expressions are general, and 4 expressions are liked, and 5 expressions are delithted with.The user is in nearest one month in the intercepting scope of browsing behavioral data of video website.The fusion rule of user browsing behavior is as shown in table 1.
The fusion rule of the various user browsing behavior data of table 1
Browse behavior | Original value r | Explanation | Corresponding value | The intercepting scope |
Play | 0%~100% | Be as the criterion the duration ratio of r for playing: 1:r≤0.1 with last broadcast; 2:0.1<r≤0.3; 3:0.3<r≤0.7; 4:0.7<r≤0.9; 5:0.9<r. | 1,2,3,4,5 | Nearly one month |
Marking | 1~5 | By initial value | 1,2,3,4,5 | Nearly one month |
Collection | 1 or 0 | R=1 represents that the user collects some programs, makes a call to 5 fens; During r=0, then do not register students' marks. | 5 | Nearly one month |
Dig behavior | 1 or 0 | R=1 represents that the user digs the some programs of behavior, makes a call to 5 fens; During r=0, then do not register students' marks. | 5 | Nearly one month |
Browse behavior | Original value r | Explanation | Corresponding value | The intercepting scope |
Bury behavior | 1 or 0 | R=1 represents that the user buries the some programs of behavior, makes a call to 1 fen; During r=0, then do not register students' marks. | 1 | Nearly one month |
Upload | 1 or 0 | R=1 represents that the user uploads some programs, makes a call to 5 fens; During r=0, then do not register students' marks. | 5 | Nearly one month |
Same user is merged the above-mentioned behavioral data value of browsing of certain video frequency program, be about to program marking behavioral data, favorites behavioral data, dig behavior bury behavior (or top step on) data or upload after the corresponding value of r adds up in the program behavioral data, obtain the interest level of this user to certain video frequency program.And after obtaining the interest level table, adopt project-based collaborative filtering method to carry out the recommendation of video frequency program.And by the more programs recommended tabulation of twice frequency every day.
Above-described embodiment only is used to illustrate technological thought of the present invention and characteristics, its purpose makes those skilled in the art can understand content of the present invention and is implementing according to this, when can not only limiting claim of the present invention with present embodiment, be all equal variation or modifications of doing according to disclosed spirit, still drop in the claim of the present invention.
Claims (10)
1. video program recommendation method based on data fusion is characterized in that may further comprise the steps:
1) user behavior data is collected step, collects the browse behavioral data of user in video website, and deposits in the database;
2) data fusion step is converted into the interest level of this user to certain video frequency program with certain user to the behavioral data of browsing of certain video frequency program;
3) program commending step according to the interest level of user to some program, is recommended this user's the high program of other interest level to other users with identical hobby;
Wherein, browse behavioral data and be included as the broadcast behavioral data, the duration of user's broadcast program is obtained duration ratio r divided by the length of this program, duration ratio r is big more, and interest level is high more.
2. the video program recommendation method based on data fusion according to claim 1 is characterized in that:
The time span of collecting data in the described step 1) is a time period before a certain moment.
3. the video program recommendation method based on data fusion according to claim 2 is characterized in that:
Described step 2) interest level in is that the interior behavioral data conversion of browsing of a time period before a certain moment obtains.
4. the video program recommendation method based on data fusion according to claim 1 is characterized in that:
Adopt regular mode to recommend this user's the high program of other interest-degrees to other users in the described step 3) with identical hobby.
5. according to the described video program recommendation method of arbitrary claim in the claim 1 to 4, it is characterized in that based on data fusion:
Described step 2) be capable with Customs Assigned Number in, video frequency program is numbered row, and the user is a content to the interest level of video frequency program, forms the interest level form of the user of a two dimension to video frequency program; In the described step 3), on the basis of this interest level table, adopt recommend method that known users and unknown subscriber are produced programs recommended tabulation, wherein, have the user of corresponding interest level to be known users, otherwise be the unknown subscriber based on collaborative filtering.
6. the video program recommendation method based on data fusion according to claim 5 is characterized in that:
Also comprise in the described step 3) and remove the program that other users had watched in the recommend programs.
7. the video program recommendation method based on data fusion according to claim 6 is characterized in that:
The described behavioral data of browsing also comprises in the following behavior one, a plurality of or whole:
To program marking behavioral data, favorites behavioral data, dig behavioral data, bury behavioral data or upload the program behavioral data.
8. the video program recommendation method based on data fusion according to claim 7 is characterized in that:
Described step 2) in the data fusion step, for the behavior of browsing of same type, the score value of the behavior that the back occurs covers the score value of the behavior that occurs earlier.
9. the video program recommendation method based on data fusion according to claim 8 is characterized in that:
Described step 2) in, adopt the numeric representation interest level of 1,2,3,4,5 five discretize, wherein 1 expression is very disagreeable, 2 expressions are general disagreeable, 3 expressions are general, and 4 expressions are liked, and 5 expressions are delithted with, the user is in nearest one month in the intercepting scope of browsing behavioral data of video website, the fusion rule of various user browsing behavior data such as following table:
With same user to the marking behavioral data of certain video frequency program, collect behavioral data, dig behavioral data, bury behavioral data and upload after the corresponding value of r adds up in the program behavioral data, obtain the interest level of this user to certain video frequency program.
10. the video program recommendation method based on data fusion according to claim 9 is characterized in that: after obtaining the interest level table, press the more programs recommended tabulation of twice frequency every day in the described step 3).
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN200810207608A CN101763351A (en) | 2008-12-23 | 2008-12-23 | Data fusion based video program recommendation method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN200810207608A CN101763351A (en) | 2008-12-23 | 2008-12-23 | Data fusion based video program recommendation method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN101763351A true CN101763351A (en) | 2010-06-30 |
Family
ID=42494515
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN200810207608A Pending CN101763351A (en) | 2008-12-23 | 2008-12-23 | Data fusion based video program recommendation method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN101763351A (en) |
Cited By (19)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102456201A (en) * | 2010-10-26 | 2012-05-16 | 阿里巴巴集团控股有限公司 | Customer operation data analysis method and customer operation data analysis method device |
CN102571978A (en) * | 2012-02-10 | 2012-07-11 | 上海视畅信息科技有限公司 | Intelligent video viewing system |
CN102609474A (en) * | 2012-01-18 | 2012-07-25 | 北京搜狗信息服务有限公司 | Access information providing method and system |
CN102630049A (en) * | 2011-12-31 | 2012-08-08 | 上海聚力传媒技术有限公司 | Method for determining interest degree of user about playing video and equipment thereof |
CN103116589A (en) * | 2011-11-17 | 2013-05-22 | 腾讯科技(深圳)有限公司 | Method and device of sending recommendation information |
CN103297853A (en) * | 2013-06-07 | 2013-09-11 | 华东师范大学 | IPTV (internet protocol television) program recommendation method based on context recognition for multiple users |
CN103491441A (en) * | 2013-09-09 | 2014-01-01 | 东软集团股份有限公司 | Recommendation method and system of live television programs |
CN103544256A (en) * | 2013-10-15 | 2014-01-29 | 北京奇虎科技有限公司 | Method and device for providing network multimedia resources |
CN104104999A (en) * | 2013-04-10 | 2014-10-15 | 华为技术有限公司 | Audio and video information recommending method and device |
CN104994408A (en) * | 2015-06-25 | 2015-10-21 | 青岛海信电器股份有限公司 | Intelligent television program recommendation method, device and intelligent television |
WO2015180385A1 (en) * | 2014-05-28 | 2015-12-03 | 小米科技有限责任公司 | Multimedia resource recommendation method and device |
CN105430505A (en) * | 2015-11-13 | 2016-03-23 | 云南大学 | IPTV program recommending method based on combined strategy |
CN105792000A (en) * | 2014-12-23 | 2016-07-20 | 北京数码视讯科技股份有限公司 | Video recommendation method and device |
CN106095819A (en) * | 2016-05-31 | 2016-11-09 | 北京奇艺世纪科技有限公司 | A kind of video recommendation method and device |
CN107391713A (en) * | 2017-07-29 | 2017-11-24 | 内蒙古工业大学 | A kind of method and system for solving the problems, such as cold start-up in collaborative filtering recommending technology |
CN108322779A (en) * | 2017-01-17 | 2018-07-24 | 南宁富桂精密工业有限公司 | The collection and processing method of video information, device and server |
CN111147940A (en) * | 2019-12-30 | 2020-05-12 | 百度在线网络技术(北京)有限公司 | Video playing method and device, computer equipment and medium |
CN111161555A (en) * | 2018-11-07 | 2020-05-15 | 北京嘀嘀无限科技发展有限公司 | Information collection method and system |
CN114691858A (en) * | 2022-03-15 | 2022-07-01 | 电子科技大学 | Improved UNILM abstract generation method |
-
2008
- 2008-12-23 CN CN200810207608A patent/CN101763351A/en active Pending
Cited By (28)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102456201A (en) * | 2010-10-26 | 2012-05-16 | 阿里巴巴集团控股有限公司 | Customer operation data analysis method and customer operation data analysis method device |
CN102456201B (en) * | 2010-10-26 | 2016-09-14 | 阿里巴巴集团控股有限公司 | A kind of analysis method and apparatus of guest operation data |
CN103116589A (en) * | 2011-11-17 | 2013-05-22 | 腾讯科技(深圳)有限公司 | Method and device of sending recommendation information |
CN102630049B (en) * | 2011-12-31 | 2014-12-10 | 上海聚力传媒技术有限公司 | Method for determining interest degree of user about playing video and equipment thereof |
CN102630049A (en) * | 2011-12-31 | 2012-08-08 | 上海聚力传媒技术有限公司 | Method for determining interest degree of user about playing video and equipment thereof |
CN102609474A (en) * | 2012-01-18 | 2012-07-25 | 北京搜狗信息服务有限公司 | Access information providing method and system |
CN102609474B (en) * | 2012-01-18 | 2016-04-06 | 北京搜狗信息服务有限公司 | A kind of visit information supplying method and system |
CN102571978A (en) * | 2012-02-10 | 2012-07-11 | 上海视畅信息科技有限公司 | Intelligent video viewing system |
CN104104999B (en) * | 2013-04-10 | 2018-10-09 | 华为技术有限公司 | Audio/video information recommends method and apparatus |
CN104104999A (en) * | 2013-04-10 | 2014-10-15 | 华为技术有限公司 | Audio and video information recommending method and device |
CN103297853A (en) * | 2013-06-07 | 2013-09-11 | 华东师范大学 | IPTV (internet protocol television) program recommendation method based on context recognition for multiple users |
CN103297853B (en) * | 2013-06-07 | 2016-04-06 | 华东师范大学 | A kind of IPTV program commending method based on multi-user's context identification |
CN103491441A (en) * | 2013-09-09 | 2014-01-01 | 东软集团股份有限公司 | Recommendation method and system of live television programs |
CN103544256A (en) * | 2013-10-15 | 2014-01-29 | 北京奇虎科技有限公司 | Method and device for providing network multimedia resources |
WO2015180385A1 (en) * | 2014-05-28 | 2015-12-03 | 小米科技有限责任公司 | Multimedia resource recommendation method and device |
US10289621B2 (en) | 2014-05-28 | 2019-05-14 | Xiaomi Inc. | Method and apparatus for recommending multimedia resource |
CN105792000A (en) * | 2014-12-23 | 2016-07-20 | 北京数码视讯科技股份有限公司 | Video recommendation method and device |
CN104994408A (en) * | 2015-06-25 | 2015-10-21 | 青岛海信电器股份有限公司 | Intelligent television program recommendation method, device and intelligent television |
CN105430505A (en) * | 2015-11-13 | 2016-03-23 | 云南大学 | IPTV program recommending method based on combined strategy |
CN105430505B (en) * | 2015-11-13 | 2018-07-03 | 云南大学 | A kind of IPTV program commending methods based on combined strategy |
CN106095819A (en) * | 2016-05-31 | 2016-11-09 | 北京奇艺世纪科技有限公司 | A kind of video recommendation method and device |
CN108322779A (en) * | 2017-01-17 | 2018-07-24 | 南宁富桂精密工业有限公司 | The collection and processing method of video information, device and server |
CN107391713B (en) * | 2017-07-29 | 2020-04-28 | 内蒙古工业大学 | Method and system for solving cold start problem in collaborative filtering recommendation technology |
CN107391713A (en) * | 2017-07-29 | 2017-11-24 | 内蒙古工业大学 | A kind of method and system for solving the problems, such as cold start-up in collaborative filtering recommending technology |
CN111161555A (en) * | 2018-11-07 | 2020-05-15 | 北京嘀嘀无限科技发展有限公司 | Information collection method and system |
CN111147940A (en) * | 2019-12-30 | 2020-05-12 | 百度在线网络技术(北京)有限公司 | Video playing method and device, computer equipment and medium |
CN114691858B (en) * | 2022-03-15 | 2023-10-03 | 电子科技大学 | Improved UNILM digest generation method |
CN114691858A (en) * | 2022-03-15 | 2022-07-01 | 电子科技大学 | Improved UNILM abstract generation method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN101763351A (en) | Data fusion based video program recommendation method | |
CN101764661A (en) | Data fusion based video program recommendation system | |
Dellarocas et al. | A statistical measure of a population’s propensity to engage in post-purchase online word-of-mouth | |
CN101489107B (en) | Collaborative filtering recommendation method based on population attribute keyword vector | |
CN103686237B (en) | Recommend the method and system of video resource | |
CN108875022B (en) | Video recommendation method and device | |
CN101047826B (en) | Electronic apparatus, information browsing method thereof | |
CN103686382B (en) | Program recommendation method | |
US9955231B2 (en) | Relevant video content pushed to a mobile phone | |
CN107071578A (en) | IPTV program commending methods | |
CN105915949A (en) | Video content recommending method, device and system | |
CN102207972B (en) | Television program recommending method and device for digital television | |
CN103649981B (en) | Method and apparatus for delivering targeted content | |
CN107818166A (en) | A kind of information recommends method, apparatus, server and system | |
CN110430471A (en) | It is a kind of based on the television recommendations method and system instantaneously calculated | |
CN103136275A (en) | System and method for recommending personalized video | |
JP4370850B2 (en) | Information processing apparatus and method, program, and recording medium | |
EP1891588A1 (en) | Method and apparatus for estimating total interest of a group of users directing to a content | |
Oh et al. | Social media analytics framework: the case of Twitter and Super Bowl ads | |
CN104038788A (en) | Community social network system and content recommendation method | |
KR101054088B1 (en) | How to automatically recommend customized IP programs using collaborative filtering | |
Xu et al. | Catch-up TV recommendations: show old favourites and find new ones | |
CN105183925A (en) | Content association recommending method and content association recommending device | |
CN106612461A (en) | A recommendation method, apparatus and system for a product package | |
Cremonesi et al. | Time-evolution of IPTV recommender systems |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C12 | Rejection of a patent application after its publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20100630 |