中文短文本多级情感分析

Multi-level Sentiment Analysis of Short Text in Chinese

多年来,文本情感分析精度不高和适用面窄等问题一直是国内外专家研究的热点和难点。首先设计了一种新的句子分割和关键信息抽取的算法,使得一般评论能够得到较为理想的断句;其次,设计了一套适用于常规语言环境下中文句子情感细粒度分析的框架模型,能够对微博等短文本语句通过简单训练进行快速的初步分析,并提供有实质意义的相关信息。

For low precision and limited applicability, text multi-level sentiment analysis has been a heated as well as hard issue round the globe. Our proposition makes two contributions to it. Firsdy, we designed a new algorithm to segment Chinese sentences and ex- tract key information from them. If the passage is not too long, the result seems satisfactory. Secondly, we propose a model to analyze the polarity and intensity of the sentiment in context, which can work fast by sim- ple training and submit valuable information.