CN-115129929-B - Video pushing method, device, server and storage medium

Abstract

The method comprises the steps of obtaining historical interaction videos corresponding to an account, determining video categories and video information corresponding to the historical interaction videos, obtaining videos matched with the video categories and videos matched with the video information to obtain candidate video sets, obtaining first features, second features and third features of all videos in the candidate video sets, wherein the first features are used for representing minimum values of feature distances between the videos and the video categories, the second features are used for representing minimum values of feature distances between the videos and the video information, the third features are used for representing correlation degrees between the videos and video search information currently input by the account, determining recommendation degrees of all videos in the candidate video sets according to the first features, the second features and the third features, screening target videos from the candidate video sets according to the recommendation degrees, and pushing the target videos to the account. By adopting the method, the accuracy of the pushed video is improved.

Inventors

• ZHANG SHUIFA

Assignees

• 北京达佳互联信息技术有限公司
• 北京达佳互联信息技术有限公司

Dates

Publication Date

20260421

Application Date

20210326

Priority Date

20210326

Claims (19)

1. 1. A video pushing method, comprising: Acquiring a history interaction video corresponding to an account, and determining a video category and video information corresponding to the history interaction video, wherein the history interaction video is a video with interaction of the account on a video recommendation page, and the video category refers to a category to which video characteristics of the history interaction video belong; acquiring videos matched with the video categories and videos matched with the video information to obtain a candidate video set; The method comprises the steps of obtaining first characteristics, second characteristics and third characteristics of each video in a candidate video set, wherein the first characteristics are used for representing the minimum value of characteristic distances between video characteristics of each video and category characteristics of each video category, the second characteristics are used for representing the minimum value of characteristic distances between information characteristics of video information of each video and information characteristics of each video information, and the third characteristics are used for representing the correlation degree between the video characteristics of each video and video search information currently input by an account; Determining the recommendation degree of each video in the candidate video set according to the first feature, the second feature and the third feature, wherein the recommendation degree of each video is determined according to the target feature of each video; and screening at least one target video from the candidate video set according to the recommendation degree, and pushing the target video to the account.
2. 2. The video pushing method according to claim 1, wherein the determining the video category corresponding to the historical interactive video includes: Acquiring a plurality of video sets which are classified in advance, wherein the video sets are obtained based on historical recommended video classification corresponding to the account; determining target video sets corresponding to each historical interactive video from the plurality of video sets; and respectively taking the video categories of the target video sets corresponding to the historical interactive videos as the video categories corresponding to the historical interactive videos.
3. 3. The video pushing method according to claim 2, wherein determining a target video set corresponding to each historical interactive video from the plurality of video sets includes: acquiring a first feature vector of each historical interactive video and a second feature vector of a collection center of each video collection; according to the first feature vector and the second feature vector, obtaining feature distances between each historical interactive video and the collection center of each video collection; And aiming at each historical interaction video, taking the video set with the minimum feature distance as a target video set corresponding to the historical interaction video.
4. 4. The method for pushing video according to claim 1, wherein the obtaining the video matching the video category and the video matching the video information to obtain the candidate video set includes: Acquiring a video to be recommended corresponding to the account; Respectively screening videos matched with the video categories and videos matched with the video information from the videos to be recommended to obtain an initial video set; And if the initial video set contains the historical interactive video, deleting the historical interactive video in the initial video set to obtain the candidate video set.
5. 5. The video pushing method according to claim 4, wherein the video information includes video author information, and the second feature is used to characterize a minimum value of feature distances between the respective videos and the respective video author information; And screening at least one target video from the candidate video set according to the recommendation degree and pushing the target video to the account, wherein the method comprises the following steps: screening at least one video with the recommendation degree larger than a preset recommendation degree from all videos included in the candidate video set, and correspondingly serving as at least one target video; sorting the target videos according to the second characteristics of the target videos; pushing the ordered target video to the account.
6. 6. The video pushing method according to claim 1, wherein the determining the recommendation degree of each video in the candidate video set according to the first feature, the second feature, and the third feature includes: and inputting the target characteristics of each video into a pre-trained video prediction model to obtain the recommendation degree of each video.
7. 7. The video pushing method according to claim 6, wherein the pre-trained video prediction model is trained by: Acquiring a first feature, a second feature and a third feature of a sample video; Inputting the first feature, the second feature and the third feature of the sample video into a video prediction model to be trained to obtain the prediction recommendation degree of the sample video; obtaining a loss value according to the difference value between the predicted recommendation degree and the actual recommendation degree of the sample video; and adjusting model parameters of the video prediction model to be trained according to the loss value to obtain an adjusted video prediction model, and repeatedly training the adjusted video prediction model until the loss value obtained according to the trained video prediction model is smaller than a preset threshold value, and taking the trained video prediction model as the pre-trained video prediction model.
8. 8. The video pushing method according to claim 1, further comprising, before determining the recommendation level of each video in the candidate video set according to the first feature, the second feature, and the third feature: Acquiring fourth characteristics of each video in the candidate video set, wherein the fourth characteristics are used for representing video quality information of each video; The determining the recommendation degree of each video in the candidate video set according to the first feature, the second feature and the third feature comprises the following steps: Splicing the first feature, the second feature, the third feature and the fourth feature of each video in the candidate video set to obtain target features of each video; and obtaining the recommendation degree of each video according to the target characteristics of each video.
9. 9. A video pushing device, comprising: The information determining unit is configured to acquire a history interaction video corresponding to an account and determine a video category and video information corresponding to the history interaction video, wherein the history interaction video is a video with interaction on a video recommendation page of the account, and the video category refers to a category to which video characteristics of the history interaction video belong; a video acquisition unit configured to perform acquisition of a video matching the video category and a video matching the video information, resulting in a candidate video set; a feature acquisition unit configured to perform acquisition of a first feature, a second feature, and a third feature of each video in the candidate video set, the first feature being used for characterizing a minimum value of feature distances between video features of each video and category features of each video category, the second feature being used for characterizing a minimum value of feature distances between information features of video information of each video and information features of each video information, the third feature being used for characterizing a correlation between video features of each video and video search information currently input by the account; The recommendation degree determining unit is configured to determine recommendation degrees of all videos in the candidate video set according to the first feature, the second feature and the third feature, wherein the recommendation degrees of all videos are determined according to target features of all videos; and the video pushing unit is configured to screen at least one target video from the candidate video set according to the recommendation degree and push the target video to the account.
10. 10. The video pushing device according to claim 9, wherein the information determining unit is further configured to perform obtaining a plurality of video sets classified in advance, wherein the plurality of video sets are obtained based on historical recommended video classification corresponding to the account, determine target video sets corresponding to each historical interactive video from the plurality of video sets, and respectively correspond video categories of the target video sets corresponding to each historical interactive video to the video categories corresponding to each historical interactive video.
11. 11. The video pushing device according to claim 10, wherein the information determining unit is further configured to perform obtaining a first feature vector of each historical interaction video and a second feature vector of a collection center of each video collection, obtain a feature distance between each historical interaction video and the collection center of each video collection according to the first feature vector and the second feature vector, and regarding each historical interaction video, use a video collection with the minimum feature distance as a target video collection corresponding to the historical interaction video.
12. 12. The video pushing device according to claim 9, wherein the video obtaining unit is further configured to obtain videos to be recommended corresponding to the account, screen videos matched with the video category and videos matched with the video information from the videos to be recommended respectively to obtain an initial video set, and delete the historical interaction videos in the initial video set if the initial video set contains the historical interaction videos to obtain the candidate video set.
13. 13. The video pushing device of claim 12, wherein the video information comprises video author information, the second feature being used to characterize a minimum value of feature distances between the respective videos and the respective video author information; The video pushing unit is further configured to perform screening of at least one video with the recommendation degree larger than a preset recommendation degree from all videos included in the candidate video set, and correspond to the at least one target video, order the target videos according to second characteristics of the target videos, and push the ordered target videos to the account.
14. 14. The video pushing device according to claim 9, wherein the recommendation determining unit is further configured to perform inputting the target feature of each video into a pre-trained video prediction model to obtain the recommendation of each video.
15. 15. The video pushing device according to claim 14, further comprising a model training unit configured to perform obtaining a first feature, a second feature and a third feature of a sample video, input the first feature, the second feature and the third feature of the sample video into a video prediction model to be trained to obtain a prediction recommendation of the sample video, obtain a loss value according to a difference between the prediction recommendation and an actual recommendation of the sample video, adjust model parameters of the video prediction model to be trained according to the loss value to obtain an adjusted video prediction model, and repeatedly train the adjusted video prediction model until the loss value obtained according to the trained video prediction model is smaller than a preset threshold, and use the trained video prediction model as the pre-trained video prediction model.
16. 16. The video pushing device of claim 9, further comprising an acquisition unit configured to perform acquiring a fourth feature of each video in the candidate video set, the fourth feature being used to characterize video quality information of the each video; the recommendation degree determining unit is further configured to splice the first feature, the second feature, the third feature and the fourth feature of each video in the candidate video set to obtain target features of each video, and obtain recommendation degrees of each video according to the target features of each video.
17. 17. A server for a server, which comprises a server and a server, characterized by comprising the following steps: A processor; a memory for storing the processor-executable instructions; Wherein the processor is configured to execute the instructions to implement the video push method of any one of claims 1 to 8.
18. 18. A computer readable storage medium, characterized in that instructions in the computer readable storage medium, when executed by a processor of a server, enable the server to perform the video push method of any one of claims 1 to 8.
19. 19. A computer program product comprising a computer program, characterized in that the computer program, when executed by a processor, implements the video push method of any of claims 1 to 8.

Description

Video pushing method, device, server and storage medium Technical Field The present disclosure relates to the field of video pushing technologies, and in particular, to a video pushing method, a device, a server, a storage medium, and a computer program product. Background With the development of video technology, various videos are layered endlessly, and in the process that a user browses videos through an application program, some videos can be pushed to corresponding users based on user requests. In the related art, the current video pushing method generally finds out videos related to the search word from a massive number of videos according to the search word input by a user, and then pushes the videos to the user, but the video wanted by the user is influenced by various factors, and only the search word input by the user is considered, so that the accuracy of the pushed video is lower. Disclosure of Invention The present disclosure provides a video pushing method, apparatus, server, storage medium and computer program product, to at least solve the problem of low accuracy of the video pushed in the related art. The technical scheme of the present disclosure is as follows: According to a first aspect of an embodiment of the present disclosure, there is provided a video pushing method, including: Acquiring a history interaction video corresponding to an account, and determining video category and video information corresponding to the history interaction video, wherein the history interaction video is a video with interaction of the account on a video recommendation page; acquiring videos matched with the video categories and videos matched with the video information to obtain a candidate video set; The method comprises the steps of obtaining first characteristics, second characteristics and third characteristics of each video in a candidate video set, wherein the first characteristics are used for representing the minimum value of characteristic distances between each video and each video category, the second characteristics are used for representing the minimum value of characteristic distances between each video and each video information, and the third characteristics are used for representing the correlation degree between each video and video search information currently input by an account; determining the recommendation degree of each video in the candidate video set according to the first feature, the second feature and the third feature; and screening at least one target video from the candidate video set according to the recommendation degree, and pushing the target video to the account. In an exemplary embodiment, the determining the video category corresponding to the historical interactive video includes: Acquiring a plurality of video sets which are classified in advance, wherein the video sets are obtained based on historical recommended video classification corresponding to the account; determining target video sets corresponding to each historical interactive video from the plurality of video sets; and respectively taking the video categories of the target video sets corresponding to the historical interactive videos as the video categories corresponding to the historical interactive videos. In an exemplary embodiment, the determining, from the plurality of video sets, a target video set corresponding to each of the historical interactive videos includes: acquiring a first feature vector of each historical interactive video and a second feature vector of a collection center of each video collection; according to the first feature vector and the second feature vector, obtaining feature distances between each historical interactive video and the collection center of each video collection; And aiming at each historical interaction video, taking the video set with the minimum feature distance as a target video set corresponding to the historical interaction video. In an exemplary embodiment, the obtaining the video matching the video category and the video matching the video information to obtain a candidate video set includes: Acquiring a video to be recommended corresponding to the account; Respectively screening videos matched with the video categories and videos matched with the video information from the videos to be recommended to obtain an initial video set; And if the initial video set contains the historical interactive video, deleting the historical interactive video in the initial video set to obtain the candidate video set. In an exemplary embodiment, the video information includes video author information, the second feature is used to characterize a minimum value of feature distances between the respective videos and the respective video author information; And screening at least one target video from the candidate video set according to the recommendation degree and pushing the target video to the account, wherein the method comprises the following steps: screening at least