Evidence and declaration combined extraction method for fact detection

Abstract

The invention relates to an evidence and declaration combined extraction method for fact detection. The method comprises the following steps: S1, specifying a pre-material library for retrieval and asection of to-be-verified declaration, cleaning a corpus, and performing entity extraction on the declaration to obtain an entity set; S2: carrying out document retrieval: for a given declaration, retrieving and constructing a corresponding candidate document set from the cleaned corpus by using an entity linking method according to the entity set, and taking all sentences in the set as a candidate sentence set; and S3, constructing evidences based on a greedy strategy evidence search method, using a pre-trained language model BERT as an evaluation model of the evidences, and training and testing the evaluation model to obtain final target evidences and categories. According to the invention, the evidence searching accuracy can be effectively improved.

Description

technical field [0001] The invention relates to the field of automatic fact detection, in particular to a method for joint extraction of evidence and statement oriented to fact detection. Background technique [0002] The purpose of automatic fact detection work is to enable computers to automatically identify and filter false information on the Internet to ensure the authenticity and reliability of information. With the successful application of deep learning in natural language processing in recent years, more and more research efforts have tried to combine deep learning technology into automatic fact checking and achieved good results. The fact detection task is a kind of automatic fact detection task. This task is used to determine the authenticity of a given statement. There are two goals involved: (1) Evidence mining, that is, for a given statement, retrieve it from Wikipedia The sentence set with the highest degree of correlation with the statement is used as evidenc...

AI Technical Summary

Problems solved by technology

[0007] In order to overcome the defect that the evidence cannot be accurately searched in the fact detection process described in the prior art, the present invention provides a method for joint extraction of evidence and statements oriented to fact detection

Images