Problem equivalence discrimination method combining semi-supervised learning and ensemble learning

Abstract

The invention discloses a problem equivalence discrimination method combining semi-supervised learning and ensemble learning. The method comprises the following steps: S1, synonym normalization: 1) embedding word vector; 2) judging word similarity; 3) manual judgment; S2, performing semantic equivalence identification: calculating a dual network based on LSTM; calculating a dual network model based on the CNN; calculating on the basis of a Match Pyramid model; extracting features artificially. According to the method, synonyms are generated and introduced in a semi-supervised mode, so that the whole system can be flexibly adjusted according to specific fields, semantic equivalence of problems is judged through four different models of ensemble learning, and the advantages of the differentmodels are brought into play.

Description

technical field [0001] The invention relates to the technical field of data mining, and in particular to a method for judging equivalence of problems by combining semi-supervised learning with integrated learning. Background technique [0002] In the form of one question and one answer, the intelligent question answering system accurately locates the questioning knowledge needed by users and provides users with personalized information services. With the development of artificial intelligence technology, intelligent question-answering systems have also gained more and more applications in banking, insurance, service, government and other industries. [0003] Intelligent question answering systems can generally be divided into two categories: domain specific or general domain. For the former, since accurate answers must be based on professional knowledge in this field, intelligent question answering systems generally rely on the "question-answer" knowledge base. Therefore, ...

AI Technical Summary

Problems solved by technology

On the one hand, it is difficult for this method to describe the equivalence / inequivalence in the deep semantics of the problem

On the other hand, it takes a lot of time and effort to manually construct a dictionary of synonyms, and it is difficult to exhaust the possibilities of synonyms

In addition to the traditional problem equivalence judgment method, although the method based on deep learning avoids the above problems in a sense, it is difficult to incorporate business prior knowledge, so it is difficult to adjust according to specific fields

Images