基于语义的文本话题倾向性分析

Semantic-Based Text Topic Sentiment Orientation Analysis

摘要: 文本情感倾向性研究是人工智能的分支学科，涉及了计算语言学，数据挖掘，自然语言处理等多个学科。基于语义的情感倾向研究和基于机器学习的情感倾向研究是情感倾向性分析的两个方向。本文采用了基于语义的方法，利用HowNet提供的情感词词典来进行文本的语义分析，对文本短语或词逐一赋予一个情感值，然后用语义和义元相似度计算的方法，计算文本中词语的语义相似度，得到词语的情感极性和强度，从而对文本的情感倾向给出一个量化的倾向程度值。通过实验表明这种方法文本研判的正确率较高。

Abstract: The text emotional tendency research is a subdiscipline of artificial intelligence, involving computational linguistics, data mining, natural language processing, and other disciplines. Semantic-based study and research based on machine learning are two divided directions of emotional tendency analysis. In this paper, we applied the semantic-based method, using emotion word dictionary provided by HowNet on text semantic analysis, give each text phrase or word an emotional value, then use the semantic similarity and semantic unit similarity calculation method to calculate the semantic similarity of words in the text, thereby get their emotional polarity and intensity, and finally obtain a quantify value of text emotional tendency. Experiments show that the accuracy of this method is relatively high.