False comment detection method and system based on knowledge integration

Abstract

The invention discloses a false comment detection method and system based on knowledge integration. The method comprises the following steps: obtaining comment data to be detected; performing false detection on the comment data to be detected by adopting a false comment detection model, wherein the false comment detection model is used for extracting text embedding characteristics based on a knowledge embedding unit, extracting context embedding characteristics based on a depth embedding unit, fusing the text embedding characteristics and the context embedding characteristics, and carrying out false detection by adopting the fused embedding characteristics. According to the method, the text embedding features and the context features of the comment data are integrated, so that the accuracy of feature semantic expression is enhanced, and the detection precision of false comments is improved.

Description

technical field [0001] The invention belongs to the technical field of big data analysis, and in particular relates to a method and system for detecting false comments based on knowledge integration. Background technique [0002] The statements in this section merely provide background information related to the present disclosure and do not necessarily constitute prior art. [0003] Online consumption, such as online shopping, booking hotels, booking train tickets, etc., has become the current mainstream consumption mode due to its rich content, convenience, and good user experience. Potential consumers usually visit product reviews before deciding whether to consume or not. However, due to factors such as merchants improving their own reputation or competition among merchants, there are often some false reviews, which bring inconvenience to consumers' decision-making, and It is also not conducive to effective regulation of online markets. [0004] At present, there are r...

AI Technical Summary

Problems solved by technology

Potential consumers usually visit product reviews before deciding whether to consume or not. However, due to factors such as merchants improving their own reputation or competition among merchants, there are often some false reviews, which bring inconvenience to consumers' decision-making, and It is also not conducive to effective online market regulation.

[0004] At present, there are related methods for evaluation screening based on machine learning. However, the existing methods usually only use bag-of-words or psycholinguistic tags to represent features when performing feature extraction, regardless of the context of the text, and it is difficult to capture the content of comments. Semantic information, unable to properly describe comments

Moreover, since reviews are sequences of varying lengths and have strong time dependence, deep models usually cannot balance the effects of long-term and short-term historical information, which will lead to time-consuming models and low accuracy

Finally, the deep learning model for fake review detection lacks interpretability, people can only see its input and output, but it is difficult to understand its work, which affects people's trust in the model and the improvement of the model effect

Images