False news interpretability detection system and method based on induction consciousness

Abstract

The invention discloses a false news interpretability detection system and a false news interpretability detection method based on induction consciousness, valuable comment sub-chains of questionable news are discovered by exploring evolution rules and hierarchical structure characteristics of a site in a comment thread, and the defects of deficiency of relevance mining and mutual influence among comments in the prior art are overcome. According to the method, interdisciplinary knowledge and a neural network model are combined, and semantic association mining and mutual influence mining between comment nodes in a false news comment tree are explored; an evolution tree network from bottom to top considers two social psychology theories to research the evolution law of the site in the comment clue so as to strengthen the valuable comments of the questioned news; according to the top-down coordination tree network provided by the invention, information absorption between the father node and the brother node is coordinated, the hierarchical structure of comments is enhanced, valuable comments are obtained by fusing the hierarchical structure of comments with evolutionary tree network features, a comment subchain is highlighted, and the interpretability of a verification result is enhanced.

Description

technical field [0001] The invention relates to a false news interpretability detection system and method based on induction evolution and coordination tree network of inductive consciousness. Background technique [0002] The development of social media has opened the door for the production and dissemination of fake news. Compared with authentic and reliable news content, planned and packaged fake news is easier to attract the audience's attention and consumption. The WHO chief has defined the problem as an information epidemic, saying fake news spreads faster and easier than the virus itself. Therefore, how to detect and analyze fake news and take intervention measures to reduce the spread of fake news has become one of the urgent problems in the field of social media. [0003] Typical fake news detection methods usually extract text features such as semantics, emotion, and stylistic features around news content, and use supervised learning (such as CNN, RNN) for binary...

AI Technical Summary

Problems solved by technology

However, although the multi-task learning method of detecting fake news based on comment position is effective, it often regards the comments in the comment tree as independent individuals, and lacks effective exploration of the correlation between comments, because subsequent comment semantics and gestures may Will be influenced or disturbed by current or previous comments

Images