Twitter推文与情感词典SentiWordNet匹配算法研究

A Matching Algorithm Between the Tweets in Twitter and SentiWordNet

在Twitter情感分类研究中,经常会采用将推文中的单词匹配情感词典中的同义词条查找相应情感值的方法 .但推文书写比较随意,包含许多俚语、缩写和特殊符号,导致许多词汇与情感词典中的词条无法匹配,匹配率不高直接影响推文的情感分类性能.针对Twitter的语言特征,提出了一套Twitter推文与情感词典SentiWordNet的匹配算法.该算法首先通过对推文内容进行数据清洗、替代处理、词性标注和词形还原等预处理,增加了命名实体识别、对hashtags内容的断词处理、基于Word Clusters的否定句处理和词组匹配等方法 .实验结果表明,采用此方法的匹配率可达90%以上.

In the research of the Twitter sentiment classification, a method is widely used to obtain sentiment values by mapping tweets ＊ words with the synonym terms in the sentiment lexicon. However, tweets are usually written informally, which contain slangs, abbreviations and special symbols, many words in the tweets cannot be found in the terms of senti-ment lexicon. Lower matching rate directly impacts the performance of sentiment classification. Based on the features of Twitter, a set of matching algorithm between tweets and sentiment lexicon SentiWordNet is proposed in the article. In this method, tweets are processed by data cleaning, alternative processing, POS tagging and word lemmatizing, along with some algorithms such as named entity recognition, hash tags word segmentation, negated context recognition with Word Clusters and phrase matching. Experimental results show that the matching rate reaches over 90%.