Video content retrieval method based on semantic search

Abstract

The invention discloses a video content retrieval method based on semantic search. According to the video content retrieval method based on semantic search, the search ambiguity is improved, and the problem that a traditional accurate search method cannot well process analyzed audio texts full of interference characters is solved; According to the video content retrieval method based on semantic search, semantic association is supported, and fuzzy matching at the levels of synonyms, synonyms and associated words is supported. Compared with a traditional semantic retrieval method, the invention provides a fuzzy retrieval method which is more suitable for life use, and corresponding synonyms, synonyms and associated words are intelligently matched with specific meanings of words in the target text in the text; according to the video content retrieval method based on semantic search, question and answer type search is supported, answers are accurately positioned in the materials according to the questions of the user, and convenience and possibility of search are improved.

Description

technical field [0001] The invention relates to the related field of video content retrieval, in particular to a video content retrieval method based on semantic search. Background technique [0002] With the continuous development of Internet technology and the continuous increase of network bandwidth today, online video has attracted a large number of users with its convenient access experience, diverse video sources and real-time update speed, making online video a part of users' online life. Indispensable and important component. [0003] With the massive growth of various videos in the network, video users often obtain interested video content through video retrieval. Existing video retrieval services mainly search for videos that meet the search criteria in the video library and provide them to users through information such as the video name, video format, video duration, or video resolution input by video users. However, with the continuous development of terminal ...

AI Technical Summary

Problems solved by technology

[0006] 1. The audio text parsed from the video is generally fuzzy and filled with a large number of interfering characters, which cannot be blurred by traditional precise search methods;

[0007] 2. Character-level search methods cannot handle sentence-level semantic information, because a word difference in the language may lead to semantic subversion, and character-level search methods cannot capture sentence-level semantic information;

[0008] 3. The question-and-answer mode cannot be realized, and the user's input text can only be matched. If the user's input text is a "question sentence", the answer to the question sentence cannot be found from massive video information

Images