Grammar compiling method, semantic parsing method and corresponding devices

Abstract

The invention provides a grammar compiling method, a semantic parsing method and corresponding devices. The grammar compiling method includes preliminarily defining corresponding grammar description files and corresponding word category description files on the basis of logical grammar by manifest language (LGML) according to commonly used semantic sentence expression; respectively generating grammar trees of the grammar description files and word category trees of the word category description files by reduction processes according to preset sequences, and hooking the word category trees at positions of corresponding word categories in the grammar trees to form semantic grammar trees so as to implement grammar compiling. Commonly used sentence description in the grammar description files comprises the word categories, operators and functions. The word category description files are used for describing particular values of the word categories. The grammar compiling method, the semantic parsing method and the corresponding devices have the advantages that the semantic grammar trees created on the basis of the mode can be used for semantic parsing in a mode of whole sentence matching or semantic mapping matching or a combination of the whole sentence matching and the semantic mapping matching.

Description

【Technical field】 [0001] The present invention relates to the field of computer application technology, in particular to a syntax compiling method, a semantic analysis method and a corresponding device. 【Background technique】 [0002] Speech recognition and spoken semantic analysis are two important technologies of voice interactive products. Speech recognition is to convert the content of speech into text, and spoken semantic analysis is to understand the information carried by spoken voice signals. The speech recognition rate and the accuracy of spoken language semantic analysis directly affect the user experience. In order to improve the speech recognition rate and the accuracy of spoken language semantic analysis, two context-free grammars specified by the W3C (World Wide Web Consortium) are currently used to construct the recognition space, that is, using the grammar in the form of BNF (Backus-Naur Form) or based on XML ( Extensible Markup Language) format syntax. [...

AI Technical Summary

Problems solved by technology

[0003] The way people speak is natural spoken language, and there are big differences in behavior, wording and text input. For example, there are loose grammar and word order reversal in spoken language. The BNF grammar recommended by W3C and the grammar in XML format are mainstream large-vocabulary continuous speech However, due to the deep interpretation levels of BNF and XML, the definition of semantic mapping and syntax understanding using these two grammars is very complicated, with the defects of poor readability and maintainability, and it is difficult to perform related grammar compilation and semantic analysis

Images