Social network rumor detection method based on content and user multi-factor analysis

Abstract

The invention provides a social network rumor detection method based on content and user multi-factor analysis. The method comprises the steps that firstly, a text information case is obtained, and text information and user information of the text information case are obtained; secondly, text content feature models of the text information case are built according to the text information, wherein the text content feature models comprise a keyword matching model, an emotion tendency model, an emotion fluctuation model, a theme clustering matching model and a content influence evaluation model; thirdly, user feature models of the text information case are established according to the user information, wherein the user feature models comprise a content consistency judgment model and a user influence evaluation model; fourthly, feature vectors are constructed according to the text content feature models and the user feature models, a classifier trained, the feature vectors are input into the classifier, a result is output, and a social network rumor is recognized. The method is used for conducting detection without depending on a single feature, misidentification is avoided, and the detection precision is improved.

Description

technical field [0001] The invention belongs to the fields of sentiment analysis, text mining and deep learning in natural language processing, and in particular relates to a method for detecting social network text rumors. Background technique [0002] At present, the detection and discrimination of online rumors mainly rely on keyword matching, manual review, comment sentiment and classifier models. [0003] The keyword matching method uses a large-scale keyword corpus for text search and comparison, and once relevant content is found, block or delete it. Such an approach has many disadvantages. For example, the error rate is high, and many messages with little influence or irrelevant content are directly deleted by the system, and even some unintentional and accidental spellings will be judged as rumors. [0004] Another disadvantage of the above method is that sometimes the appearance of keywords does not mean that things have been distorted or misunderstood, but only ...

AI Technical Summary

Problems solved by technology

[0008] The main purpose of the present invention is to provide an analysis method that combines text content and user features to comprehensively detect rumors, so as to eliminate the current problems of large detection errors and low efficiency that rely on a single feature

Images