An extraction method of a social video topic extraction system based on a topic model

Abstract

The invention discloses a social video theme extraction system and method based on a topic model, and relates to the social video theme extraction technology. The extraction system comprises a user comment collection module, an original comment database, a user comment preprocessing module, a theme detection module, a theme database and a result display module. The extraction method comprises thefollowing steps: firstly, crawling user comment data of a to-be-processed video within a certain period of time by the user comment collection module; preprocessing each user comment data by the usercommend preprocessing module, calculating a preprocessing result by the theme detection module by adopting a theme extraction algorithm to obtain a fine-grained theme keyword set; writing the fine-grained theme keyword set into the theme database for storing; and calling a fine-grained theme keyword in the theme database for display by the result display module. According to the social video themeextraction system and the method based on the topic model provided by the invention, gaps are filled based on bullet screen data mining of characters, so that the problem of bullet screen semantic sparsity is relieved to a certain extent, and the display result is better.

Description

technical field [0001] The invention relates to a social video topic extraction technology, in particular to an extraction method of a topic model-based social video topic extraction system. Background technique [0002] With the rapid popularity of social video applications such as barrage videos, more and more users are accustomed to posting comments on video content in real time during the process of watching videos. It floats across the middle screen, so it is called the barrage. [0003] Social video applications represented by barrage not only enhance the user's interactive experience, but also expand a large number of user-contributed text information related to the video plot, which facilitates the extraction of fine-grained video topics. [0004] The current video topic extraction methods mostly extract low-level features from video frames, and then map the low-level features to high-level semantic concepts. Due to the natural gap between low-level features and hi...

AI Technical Summary

Problems solved by technology

Due to the natural gap between low-level features and high-level semantic concepts, this method is difficult to accurately describe video content; and the processing of video frames also requires a lot of computing resources

Images