Chinese semantic parsing method and device

Abstract

The invention relates to a Chinese semantic parsing method and device. The Chinese semantic parsing method comprises the steps: obtaining a Chinese target text, and obtaining named entity words contained in the Chinese target text and relation words between the named entity words through a syntactic analysis tool; vectoring all words and named entity words forming the Chinese target text and relational words between the named entity words to obtain word vectors, and performing one-hot encoding on the word vectors to obtain a discrete feature vector of the word vector; conducting semantic feature extraction on the Chinese target text through a word embedding tool, and obtaining continuous feature vectors of word vectors in the Chinese target text; performing semantic analysis on the continuous feature vector and the discrete feature vector to obtain a first semantic analysis result of the Chinese target text; and correcting the first semantic analysis result in combination with a knowledge base to obtain a second semantic analysis result of the Chinese target text. According to the Chinese semantic parsing method and device, the capability of analyzing abstract semantics of the Chinese target text is improved.

Description

technical field [0001] The invention relates to the field of natural language processing, in particular to a Chinese semantic analysis method and device. Background technique [0002] In recent years, the field of semantic analysis has become a research hotspot in natural language processing. At the same time, neural network technology is also used in various aspects of natural language processing. Therefore, the research on semantic analysis combined with the efficient expression ability of neural network technology is very important. [0003] Parsing natural language into various designed semantic representations has always been an important research content of natural language processing. Common semantic representation structures include: logical semantic representation based on lambda expression, grammar based on combined category grammar (CCG) And semantic representation and semantic knowledge representation based on knowledge base, etc. Traditional semantic analysis m...

AI Technical Summary

Problems solved by technology

Traditional semantic analysis methods usually use traditional machine learning methods such as SVM, linear regression, and random forest to analyze natural language sentences; these methods can obtain logical semantic representations to a certain extent, but need to define very complex heuristics in advance rules and external knowledge and the traditional semantic representation contains less semantics

Images