Video type acquisition method, device and electronic device based on multi-model

Abstract

The invention relates to the field of classification models, and discloses a multi-model-based video type acquisition method, comprising: acquiring a behavior feature set, classifying the behavior feature set, and obtaining a first behavior feature set and a second behavior feature set; The behavioral feature set and the second behavioral feature set are input to the improved classification model, and a first feature subset is selected from the first behavioral feature set and the second behavioral feature set by using the feature distribution layer to send to the first classification sub-model, and Use the first classification sub-model to classify the first video playback set corresponding to the first feature subset to click class, and obtain the first type quantity set and the second type quantity set; Weight calculation, select the video type that the target user is interested in from the weight calculation result. The present invention can realize rapid judgment of the video type that the user is interested in, thereby improving the accuracy of pushing the video type of the user's interest.

Description

technical field [0001] The present invention relates to the field of classification models, and in particular, to a method, apparatus, electronic device and computer-readable storage medium for acquiring video types based on multiple models. Background technique [0002] With the continuous development of communication technology, the threshold for video playback is gradually lowered, and short videos have gradually become the mainstream of entertainment in the era. Pushing interesting videos to target users has become the main way for major mainstream media to attract users. [0003] At present, users' likes are mainly used to judge the video types that users like, but sometimes there are problems such as "users forget to like" or "users accidentally like", which makes it impossible to obtain more and more accurate data volume, resulting in The prediction results obtained by the prediction model of the push system are not accurate enough, which also makes it impossible to a...

AI Technical Summary

Problems solved by technology

[0003] At present, the method of users' likes is mainly used to judge the type of videos that users like, but sometimes there are problems such as "the user forgot to like" or "the user accidentally likes", which leads to the inability to obtain more and more accurate data volume, resulting in The prediction results obtained by the prediction model of the push system are not accurate enough, which also makes it impossible to accurately obtain the user's interested videos, which in turn leads to inaccurate video push

Images