Live video topic classification method and device and electronic device

Abstract

The embodiment of the invention provides a live video topic classification method and device and an electronic device. The method comprises: through collecting bullet screen information of a live video and splicing bullet screen information of the live video, obtaining bullet screen long text, matching the bullet screen long text with bullet screen information in a corpus, and taking single characters or words matched in the bullet screen long text as keywords; and determining the keywords of which the first occurrence frequency is greater than a first preset threshold and the second occurrence frequency is less than a second preset threshold as topic keywords, and determining a topic classification result of the live video based on the topic keywords. By adopting the method and the device, the topic classification of the live video can be automatically realized based on the bullet screen information generated in the live video, and the topic classification of the live video does not need to be manually performed, so that the cost of the topic classification of the live video is reduced and the classification efficiency is improved.

Description

technical field [0001] The present invention relates to the field of multimedia technology, in particular to a live video topic classification method, device and electronic equipment based on barrage data. Background technique [0002] As a new type of entertainment, live streaming has a broad market and a certain audience base, and it is more and more closely related to people's lives. The content of the live broadcast mainly comes from bloggers with strong autonomy, so the theme of the live broadcast highlights the characteristics of fragmentation, so it is very urgent to obtain the broadcast content of the live broadcast in a timely manner. [0003] Usually, the classification of live video topics is mainly based on manual work. One method is defined by the anchor. Before the broadcast on the live broadcast platform, the anchor must select the theme category to start broadcasting, such as games, beauty makeup, two-dimensional, etc., which can be defined by the anchor. Th...

AI Technical Summary

Problems solved by technology

[0003] Usually, the classification of live video topics is mainly based on manual work. One method is defined by the anchor. Before the broadcast on the live broadcast platform, the anchor must select the theme category to start broadcasting, such as games, beauty makeup, two-dimensional, etc., which can be defined by the anchor. The category determines the topic category of the live content. Using this method to distinguish the topic category of the live video has the problem of inaccurate topic classification

Another method is that the company can manually monitor the live video content, and distinguish the subject category of the live video according to the monitored live video content. If there is sensitive information in the live video content, the live account can be banned or other processing, and the live video can be processed in this way. Classification of topics is costly and inefficient

Images