CN109241348B - Video recommendation method and system - Google Patents
Video recommendation method and system Download PDFInfo
- Publication number
- CN109241348B CN109241348B CN201811256921.9A CN201811256921A CN109241348B CN 109241348 B CN109241348 B CN 109241348B CN 201811256921 A CN201811256921 A CN 201811256921A CN 109241348 B CN109241348 B CN 109241348B
- Authority
- CN
- China
- Prior art keywords
- video
- videos
- list
- content
- user
- 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
Images
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Two-Way Televisions, Distribution Of Moving Picture Or The Like (AREA)
Abstract
The invention discloses a video recommendation method and a video recommendation system, which comprise the following steps: screening the existing videos according to the historical clicking behaviors of the user, and placing the screened videos into a pre-established video database; distributing the videos in the video database to more than one pre-established content bucket according to the video attributes; each content bucket has a preset weight, and the weight is adjustable; extracting videos from the content bucket according to the weight, and forming more than one video list by the extracted videos; and randomly selecting one video list to recommend to the user. The technical scheme provided by the invention can effectively carry out personalized information recommendation on the user, and has the advantages of simple system structure and low development cost.
Description
Technical Field
The invention relates to the technical field of communication, in particular to a video recommendation method and system.
Background
With the development of the mobile internet, the network information is greatly increased, and especially the wide application of short videos greatly meets the requirement of users for obtaining information. However, with the increase of the number of short videos, users face unprecedented difficulty in information selection, and cannot quickly and intuitively acquire interesting contents from a huge video database, so that the use efficiency of information is reduced to a certain extent.
At present, an effective way to solve the above problems is to adopt system recommendation, and recommend video information suitable for a user to the user according to characteristics of the user, such as information demand and personal interest, so as to avoid tedious search work performed by the user.
However, the existing video recommendation system is complex in structure and adopts technologies such as data mining, machine learning and artificial intelligence, and a large amount of labor cost and time cost are required for building the system, which cannot be borne by many small and medium-sized enterprises.
Disclosure of Invention
The invention aims to provide a video recommendation method and system, which can effectively perform personalized information recommendation on a user, and has the advantages of simple system structure and low development cost.
In order to achieve the purpose, the technical scheme adopted by the invention is as follows:
a video recommendation method, comprising: screening the existing videos according to the historical clicking behaviors of the user, and placing the screened videos into a pre-established video database; distributing the videos in the video database to more than one pre-established content bucket according to the video attributes; each content bucket has a preset weight, and the weight is adjustable; extracting videos from the content bucket according to the weight, and forming more than one video list by the extracted videos; and randomly selecting one video list to recommend to the user.
Further, before said randomly selecting one of said video lists and recommending it to the user, the method further comprises: and performing video rearrangement processing on the more than one video list.
Further, still include: the recommended video list is deleted periodically.
Further, still include: and removing the videos in the video database according to a preset removing rule at regular intervals.
Preferably, the existing video comprises: historical videos and newly-built videos; screening the historical video comprises: sequencing the historical videos according to content quality, and selecting the top 5% of videos in the sequencing, wherein the number of the top 5% of videos is not more than 150; or sequencing the historical videos according to content quality, and selecting the top 10% of videos in the sequencing, wherein the number of the top 10% of videos is not more than 300; the screening of the new video comprises the following steps: and performing quality audit on the newly-built video.
Preferably, the video attributes include: exposure, heat value, release time, activity and country; the calculation method of the heat value comprises the following steps:
wherein R is a calorific value, c1A number of views having a viewing duration of more than 5 seconds, c2To order praise, c3For delivering the number of flowers, c4As the number of comments, c5For sharing success number, d is video playing number, a1，a2，a3，a4And b is constant.
A video recommendation system comprising: the screening unit is used for screening the existing videos according to the historical clicking behaviors of the users and placing the screened videos into a pre-established video database; the distribution unit is used for distributing the videos in the video database to more than one pre-established content bucket according to the video attributes; each content bucket has a preset weight, and the weight is adjustable; the extracting unit is used for extracting videos from the content bucket according to the weight and forming more than one video list by the extracted videos; and the recommending unit is used for randomly selecting one video list and recommending the video list to the user.
Further, still include: and the duplicate removal unit is used for carrying out video duplicate removal processing on more than one video list before randomly selecting one video list and recommending the video list to a user.
Further, still include: and the list deleting unit is used for deleting the recommended video list periodically.
Further, still include: and the video removing unit is used for removing the videos in the video database according to a preset removing rule at regular intervals.
According to the video recommendation method and system provided by the embodiment of the invention, videos are screened according to historical clicking behaviors of users, then the videos are classified according to video attributes, and then the videos with different video attributes are extracted from different content buckets according to weights and recommended to the users. Because the video attribute can be defined and the extraction weight can be adjusted, each enterprise can establish different content buckets according to the actual situation of the video of the enterprise and set different weights for the content buckets, thereby recommending different personalized video contents to different users. Moreover, the system has simple structure and low development cost.
Drawings
FIG. 1 is a flow chart of a method of an embodiment of the present invention;
FIG. 2 is a schematic diagram of a system configuration according to an embodiment of the present invention;
FIG. 3 is a system layout of an embodiment of the present invention;
fig. 4 is a system architecture diagram of an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail below with reference to the accompanying drawings.
FIG. 1 is a flow chart of a method of an embodiment of the present invention, including the steps of:
in this step, the existing video includes: historical video and new video. Screening the historical video comprises: sequencing the historical videos according to content quality, selecting the top 5% of videos in the sequencing, rounding by decimal point carry, wherein the number of the top 5% of videos is not more than 150; or sequencing the historical videos according to the content quality, selecting the top 10% of videos in the sequencing, rounding by decimal point carry, wherein the number of the top 10% of videos is not more than 300. The screening process can be manual screening or automatic screening of the system.
The screening of the new video comprises the following steps: and performing quality audit on the newly-built video. For example, a set of video quality criteria is established, which is divided into S, A, B, C four quality levels, and each new video (i.e. newly released video) is scored for quality. For video of quality C, it is directly discarded and for video of quality S, A, B it is placed in the video database described above.
102, distributing the videos in the video database to more than one pre-established content bucket according to video attributes; each content bucket has a preset weight, and the weight is adjustable;
in this step, the video attributes include: exposure, heat value, release time, activity, the country, and the like, and each enterprise can specify other different video attributes according to the characteristics of the video. Each video attribute corresponds to a bucket of content, the number of buckets of content increasing or decreasing with increasing or decreasing video attributes, respectively.
The calculation method of the heat value comprises the following steps:
wherein R is a calorific value, c1A number of views having a viewing duration of more than 5 seconds, c2To order praise, c3For delivering the number of flowers, c4As the number of comments, c5For sharing success number, d is video playing number, a1，a2，a3，a4B is a constant, in particular a is 0.5, a1＝1，a2＝3，a3＝5，a4＝10，a520, b is 1. The above parameters support on-line updating at any time.
According to the video attributes of exposure, heat value, release time, activity and the country, a corresponding content bucket can be established: low exposure bucket, high heat bucket, new content bucket, active PGC bucket, national bucket. Wherein, the low exposure bucket stores the video with the video watching amount less than 100. The high-heat bucket stores videos with heat values larger than a certain threshold, and the threshold can be set according to actual conditions. The new content bucket stores the video with the release time of the latest 7 days. And storing videos meeting the following conditions in the active PGC bucket: the number of times that the video publisher starts App in the last 7 days is more than 1, the video publishing time is within 1 month, and the video quality is more than A level. Wherein the video quality is judged according to the video quality judgment standard in step 101. The national bucket is classified by the country to which the video belongs, and specifically, may be classified into china (including australia, port), japanese korean, thailand, vietnam, malaysia, philippines, indonesia, southeast asia (other non-divided countries), europe and america (usa, canada, singapore, australia, new zealand, europe), south american africa, and the like.
103, extracting videos from the content bucket according to the weight, and forming more than one video list by the extracted videos;
the weight is the weight of the content bucket, and a high weight of a content bucket indicates that the number of videos extracted from the content bucket is large. Assuming that each video list is composed of 30 videos, in the present embodiment, the method for extracting the 30 videos from the content bucket is as follows: the first 3 videos were extracted from the active pgc bucket, randomly ordered; extracting 2 videos from the new content bucket, and randomly sequencing; the remaining 25 videos are extracted according to the above weight rule. For the old user, the weighting rule is: 10% of the low exposure bucket, 10% of the new content bucket, 30% of the high heat bucket, 30% of the active PGC bucket, and 20% of the home bucket, that is, 10%, 30%, and 20% of the video amount is extracted from the low exposure bucket, the high heat bucket, the new content bucket, the active PGC bucket, and the home bucket, respectively. For the new user, the weighting rule is: low exposure bucket 0%, new content bucket 5%, high heat bucket 35%, active pgc bucket 35%, home bucket 25%, i.e. 0, 5%, 35% and 25% of the video amount is extracted from low exposure bucket, new content bucket, high heat bucket, active pgc bucket, home bucket, respectively. Different enterprises can set different video attributes and different extraction weights according to the video characteristics of the enterprises.
And 104, randomly selecting one video list to recommend to the user.
In this embodiment, since different video lists may include the same video, before randomly selecting one of the video lists and recommending the selected video list to the user, the video rearrangement processing needs to be performed on the more than one video lists. Specifically, a cache record is made for the video that the client has watched, and if the user pulls the video that has been watched again after a period of time, the video will not be displayed at the user side.
In this embodiment, the method further includes: the recommended video list is periodically deleted, and the video data in the content bucket is periodically deleted. And, the videos in the video database need to be removed periodically according to a preset removing rule, and the video database needs to be supplemented with videos according to the method in step 101. The specific elimination rule is as follows: and eliminating videos which are exposed for a period of time and have the exposure value of more than or equal to 50 and the heat value of less than a certain threshold value from the video database. The threshold value can be adjusted according to the characteristics of the video and the viewing feedback of the user. It should be noted that the rejected video is never recommended to the user unless someone intervenes.
The invention also discloses a video recommendation system, as shown in fig. 2, comprising: the screening unit is used for screening the existing videos according to the historical clicking behaviors of the users and placing the screened videos into a pre-established video database; the distribution unit is used for distributing the videos in the video database to more than one pre-established content bucket according to the video attributes; each content bucket has a preset weight, and the weight is adjustable; the extracting unit is used for extracting videos from the content bucket according to the weight and forming more than one video list by the extracted videos; and the recommending unit is used for randomly selecting one video list and recommending the video list to the user.
Further, still include: and the duplicate removal unit is used for carrying out video duplicate removal processing on more than one video list before randomly selecting one video list and recommending the video list to a user. And the list deleting unit is used for deleting the recommended video list periodically. And the video removing unit is used for removing the videos in the video database according to a preset removing rule at regular intervals.
The system adopts the following technology in the development process: PHP, Redis, MongoDB. The PHP has a wide application foundation in the WEB service development of a service end, and has the advantages of flexible language, convenient deployment and high development efficiency; redis is a popular memory storage service program, and is used for providing a Cache function; MongoDB is used as an excellent open-source non-relational database, has low cost, convenient use and easy expansion, and is used for persistent storage of data. The architecture of the system is shown in figure 4.
The development process of the system is as follows:
1. a video database is established in Redis, and videos are manually or automatically screened and placed into the video database.
2. And regularly extracting videos in the video database by adopting the Linux crontab, and distributing the videos to different content buckets according to the video attributes.
3. And extracting videos from the content bucket at regular time by adopting Linux crontab according to the weight, and forming more than one video list.
4. And randomly selecting a video list to recommend to the user.
5. And the Linux crontab is adopted to reject the videos in the video database regularly according to rejection rules, and meanwhile, the videos in the video database can be manually rejected.
6. And collecting click behavior data of the user, and continuously optimizing historical click behaviors of the user so as to optimize video recommendation quality.
Therefore, an effective, low-cost and quickly-realized video recommendation system is built. On the basis, each enterprise can take the video attribute needing to participate in recommendation as the classification standard of the content bucket according to the actual situation of the video of the enterprise, can newly add different content buckets, and then adjust the weight of each content bucket so as to achieve the recommendation system suitable for the video content of the enterprise.
According to the video recommendation method and system provided by the embodiment of the invention, videos are screened according to historical clicking behaviors of users, then the videos are classified according to video attributes, and then the videos with different video attributes are extracted from different content buckets according to weights and recommended to the users. Because the video attribute can be defined and the extraction weight can be adjusted, each enterprise can establish different content buckets according to the actual situation of the video of the enterprise and set different weights for the content buckets, thereby recommending different personalized video contents to different users. Moreover, the system has simple structure and low development cost.
The above description is only for the specific embodiments of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present invention, and all the changes or substitutions should be covered within the scope of the present invention.
Claims (9)
1. A method for video recommendation, comprising:
screening the existing videos according to the historical clicking behaviors of the user, and placing the screened videos into a pre-established video database, wherein the existing videos comprise: the method comprises the steps that historical videos and newly-built videos are screened according to the quality of video contents;
distributing the videos in the video database to more than one pre-established content bucket according to the video attributes; each content bucket has a preset weight, and the weight is adjustable; the video attributes include: exposure, heat value, release time, activity and country; the calculation method of the heat value comprises the following steps:
wherein the content of the first and second substances,Rthe value of the heat is the value of the heat,c 1 for browsing numbers with a browsing duration longer than 5 seconds,c 2 in order to count the number of praise,c 3in order to feed the flower in a number of flowers,c 4in order to be a number of the comments,c 5in order to share the success number of the session,din order to play the number of videos,a 1 ,a 2 ,a 3 ,a 4 ,a 5 ,bare all constant;
extracting videos from the content bucket according to the weight, and forming more than one video list by the extracted videos;
and randomly selecting one video list to recommend to the user.
2. The video recommendation method of claim 1, further comprising, before said randomly selecting one of said video lists for recommendation to a user: and performing video rearrangement processing on the more than one video list.
3. The video recommendation method of claim 2, further comprising: the recommended video list is deleted periodically.
4. The video recommendation method of claim 3, further comprising: and removing the videos in the video database according to a preset removing rule at regular intervals.
5. The video recommendation method of claim 1,
screening the historical video comprises: sequencing the historical videos according to content quality, and selecting the top 5% of videos in the sequencing, wherein the number of the top 5% of videos is not more than 150; or sequencing the historical videos according to content quality, and selecting the top 10% of videos in the sequencing, wherein the number of the top 10% of videos is not more than 300;
the screening of the new video comprises the following steps: and performing quality audit on the newly-built video.
6. A video recommendation system, comprising:
the screening unit is used for screening the existing videos according to the historical clicking behaviors of the users and placing the screened videos into a pre-established video database, wherein the existing videos comprise: the method comprises the steps that historical videos and newly-built videos are screened according to the quality of video contents;
the distribution unit is used for distributing the videos in the video database to more than one pre-established content bucket according to the video attributes; each content bucket has a preset weight, and the weight is adjustable; the video attributes include: exposure, heat value, release time, activity and country; the calculation method of the heat value comprises the following steps:
wherein the content of the first and second substances,Rthe value of the heat is the value of the heat,c 1 for browsing numbers with a browsing duration longer than 5 seconds,c 2 in order to count the number of praise,c 3in order to feed the flower in a number of flowers,c 4in order to be a number of the comments,c 5in order to share the success number of the session,din order to play the number of videos,a 1 ,a 2 ,a 3 ,a 4 ,a 5 ,bare all constant;
the extracting unit is used for extracting videos from the content bucket according to the weight and forming more than one video list by the extracted videos;
and the recommending unit is used for randomly selecting one video list and recommending the video list to the user.
7. The video recommendation system of claim 6, further comprising:
and the duplicate removal unit is used for carrying out video duplicate removal processing on more than one video list before randomly selecting one video list and recommending the video list to a user.
8. The video recommendation system of claim 7, further comprising:
and the list deleting unit is used for deleting the recommended video list periodically.
9. The video recommendation system of claim 8, further comprising:
and the video removing unit is used for removing the videos in the video database according to a preset removing rule at regular intervals.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811256921.9A CN109241348B (en) | 2018-10-25 | 2018-10-25 | Video recommendation method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811256921.9A CN109241348B (en) | 2018-10-25 | 2018-10-25 | Video recommendation method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109241348A CN109241348A (en) | 2019-01-18 |
| CN109241348B true CN109241348B (en) | 2022-03-22 |
Family
ID=65082343
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811256921.9A Active CN109241348B (en) | 2018-10-25 | 2018-10-25 | Video recommendation method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109241348B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113781086A (en) * | 2021-01-21 | 2021-12-10 | 北京沃东天骏信息技术有限公司 | Article recommendation method, device, medium and electronic equipment |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106326413A (en) * | 2016-08-23 | 2017-01-11 | 达而观信息科技（上海）有限公司 | Personalized video recommending system and method |
| CN106339447A (en) * | 2016-08-23 | 2017-01-18 | 达而观信息科技（上海）有限公司 | System and method for automatically predicting hot videos |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7590616B2 (en) * | 2006-11-17 | 2009-09-15 | Yahoo! Inc. | Collaborative-filtering contextual model based on explicit and implicit ratings for recommending items |
-
2018
- 2018-10-25 CN CN201811256921.9A patent/CN109241348B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106326413A (en) * | 2016-08-23 | 2017-01-11 | 达而观信息科技（上海）有限公司 | Personalized video recommending system and method |
| CN106339447A (en) * | 2016-08-23 | 2017-01-18 | 达而观信息科技（上海）有限公司 | System and method for automatically predicting hot videos |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109241348A (en) | 2019-01-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108108821B (en) | Model training method and device | |
| CN104717124B (en) | A kind of friend recommendation method, apparatus and server | |
| CN103295147B (en) | method, device and system for advertising | |
| CN104091287B (en) | A kind of task implementation method and device based on data-driven | |
| US9224105B2 (en) | Method and system of automatically downloading media content in a preferred network | |
| CN100384134C (en) | Client-end resource search under broadcast-storage network environment and automatic downloading method | |
| CN106529714A (en) | Method and system predicting user loss | |
| CN109635199B (en) | Application list dynamic recommendation method and system based on user behaviors | |
| CN104461742B (en) | Calculate equipment optimization method and device | |
| CN103631954A (en) | Individualized recommendation method and device | |
| CN102004994A (en) | Online product recommendation and selection method, device and system | |
| CN106815254A (en) | A kind of data processing method and device | |
| CN110929141B (en) | Group mining method, device, equipment and storage medium | |
| CN106815260A (en) | A kind of index establishing method and equipment | |
| CN105701135A (en) | Application recommendation method and device | |
| CN108595526B (en) | Resource recommendation method and device | |
| CN104992482A (en) | Athletic competition data processing system and method thereof | |
| CN109241348B (en) | Video recommendation method and system | |
| CN108320168A (en) | A kind of data analysing method and device | |
| CN106487642A (en) | A kind of method and apparatus of pushed information | |
| CN103593543B (en) | A kind of method and system updating P2P network user's subjectivity Analysis model of network behaviors parameter | |
| CN106910082A (en) | It is a kind of to transfer the method and apparatus that advertisement is pushed | |
| CN113010790A (en) | Content recommendation method, device, server and storage medium | |
| CN108763506A (en) | A kind of message push processing method and device | |
| CN110874609B (en) | User clustering method, storage medium, device and system based on user behaviors |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |