Semantic-based Chinese network text emotion extracting method

Abstract

The invention provides a method capable of accurately extracting network text emotion information. The method comprises the following steps of: (1) detecting special symbols such as emotion symbols, abbreviations, acronyms, interjections and the like in a text; (2) pre-processing the plain text by using a professional syntax analyzer to obtain a basic syntactic relationship of sentences; (3) analyzing words separated from the sentences, and endowing the words with emotion vectors; (4) obtaining phrase emotion vectors according to the types of different phrases and the corresponding emotion rules; and (5) obtaining final emotion vectors of the sentences according to different types of the sentences by combining the phrase emotion vectors. The network text emotion extracting method can comprehensively and detailedly identify the emotion types, and has high accuracy.

Description

(1) Technical field [0001] The invention relates to a method for extracting network text emotion information, and mainly relates to the fields of natural language processing and emotion computing. (2) Background technology [0002] Natural language is one of the unique means of communication for human beings, and with the development of the Internet, natural language has also become an important means of network communication, and gradually derived a new language method: Internet language. Network text information contains rich emotional information, corresponding to the corresponding psychological state of the user, so the research on text emotion extraction is of great significance in the field of emotional computing and intelligent interaction. It enables the computer to experience human emotions and make appropriate responses from text information, and can be used in man-machine dialogue systems and bionic agent interaction systems. How to accurately and effectively rec...

AI Technical Summary

Problems solved by technology

[0005] In terms of research on the emotional information of Chinese language texts, most of the current research is limited to the two emotional polarities of praise and criticism, and there is a lack of corresponding models for network language processing, which greatly restricts the development of Chinese human-computer emotional interaction.

The text processing model for Internet language combined with the emotion extraction method of six basic emotion models of Ekman can effectively solve the problems of lack of emotional information and low recognition rate in Chinese texts.

Images