Cold start fraud comment detection method based on social attention mechanism representation learning

Abstract

The invention discloses a cold start fraud comment detection method based on social attention mechanism representation learning, and the method comprises the steps: constructing an initial target function representing the entity relation of a user, a project, comments and scores based on a given online comment data set; constructing an explicit user characteristic matrix of the display relationship between the users and an implicit user characteristic matrix of the implicit relationship between the users according to the scores, and then constructing a social coupling matrix of the users; integrating the social coupling matrix of the user into the user representation matrix by adopting an attention mechanism, and adjusting the initial target function to obtain a new target function; and determining an attention mechanism of the new user, and identifying whether the comment is a fraudulent comment or not according to the determined classifier. The entity relationship, the user social coupling relationship and the fraud related information are embedded into the user representation space of the social attention mechanism, so that the defect of lack of user historical information in the cold start problem is effectively overcome, and the fraud comments under the cold start condition can be effectively detected.

Description

technical field [0001] The present application relates to the field of computer technology, in particular to a method for detecting cold-start fraudulent comments based on social attention mechanism representation learning. Background technique [0002] With the wide application of the Internet, its rich information resources have brought great convenience to people, and the comments published by Internet users have seriously affected people's decision-making. The existence of fraudulent reviews on the Internet greatly damages users' decision-making, and fraudsters who write fraudulent reviews to confuse honest users can obtain excellent commercial value and reputation. Most of the existing methods are based on the user’s comment content to detect fraudulent comments, but when a new user just publishes a new comment, the detection method based on the comment content fails due to the lack of sufficient new user history information, this kind of problem is called Cold start p...

AI Technical Summary

Problems solved by technology

Although the performance is obviously improved, this method only captures the user common comment relationship, ignoring other complex social relationships, such as users with the same attitude and similar preferences

In addition, this method does not consider the relevant fraudulent information in the entity relationship learning process, which may reduce the fraudulent review detection accuracy especially in the cold start situation.

Images