基于信息熵的文本语句整体褒贬倾向的识别方法

An Identification Method Text of Overall Commendatory and Derogatory Tendency of Sentences Based on Information Entropy

在人工智能和大数据时代,通过分析数据来发现规律已经成为一种趋势。作为当前较为普遍且流行的数据信息,文本数据已经广泛地被用于各种分析。在文本数据分析领域,文本语句的褒贬含义研究是目前的一个重要研究方向。通过对一些主观性言论情感色彩的倾向性分析,这种研究有助于了解公众对某些热点事件的看法和意见。本文通过基于信息熵的文本挖掘理论,结合褒贬词的比例,提出一种文本语句整体褒贬判断的方法—比例信息熵。实验结果表明,相比于目前所流行的几种典型的信息熵方法,比例信息熵对文本整体信息的褒贬倾向性分析是有效的。

In the era of artificial intelligence and big data,to find implicit laws from data analysis has been a new trend.Currently,as a kind of common and popular information,text data have been widely applied in various analyses.Studying the commendatory and derogatory meaning of text sentences is an important way in the field of text data analysis.By analyzing the emotion color tendency of some subjective speech,the research contributes to illustrate how the public to feel about some hot issues.In this paper,we combine the proportion of commendatory and derogatory words to develop a novel methodology to analyze the positive or negative trends of the whole sentences.Experiments reveal that this method performs better than some selected typical entropy methodologies.Therefore,the proportion information entropy can be regarded to be helpful for positive or negative judgments of sentences.