Auditing knowledge graph entity extraction method based on deep learning algorithm

Abstract

The invention relates to an auditing knowledge graph entity extraction method based on a deep learning algorithm. The auditing knowledge graph entity extraction method comprises the following steps: step 1, inputting an auditing process description text sequence in an auditing record; 2, generating an encoder fused with audit dictionary knowledge based on an auditing process description text sequence in the audit record input in the step 1, and calculating an output feature vector of the encoder; and step 3, decoding the text feature vector output by the encoder by using the single-layer LSTM network to obtain a fine-grained problem clue triple sequence containing the entity potential relationship. According to detailed plots in the auditing process, the deep learning-based problem clue triple is adopted to extract fine-grained entity relationships for joint extraction, interaction and constraint of an entity model and a relationship model are better utilized, structured extraction and storage of auditing items are realized, and the auditing efficiency is improved. And the performance of the unstructured long text entity relationship extraction model is improved.

Description

technical field [0001] The invention belongs to the technical field of intelligent auditing, and relates to a method for extracting audit knowledge map entities, in particular to a method for extracting audit knowledge map entities based on deep learning algorithms. Background technique [0002] Audit records are first-hand information detailing review logic and problem examples, and the entity relationships contained in them, such as business rules, rules and regulations, and business profiles, are a form of domain knowledge. Through entity relationship extraction, the abstract semantic data in audit records is converted into visualized knowledge map information, which not only facilitates the learning and analysis of auditing professionals to study and judge the type, frequency, and form of problems, but also realizes the reasonable fusion and cross-border audit information. Domain analysis provides valuable reference for digital auditing. [0003] The wide application of...

AI Technical Summary

Problems solved by technology

[0007] The purpose of the present invention is to overcome the deficiencies of the prior art, and propose a method for extracting audit knowledge map entities based on deep learning algorithms. For the detailed plot of the audit process, the fine-grained extraction of triplets of question clues based on deep learning is used to fuse domain dictionary knowledge The entity-relationship joint extraction makes better use of the interaction and constraints between the entity model and the relationship model, realizes structured extraction and storage of audit items, and improves the performance of the unstructured long-text entity-relationship extraction model

Images