【摘要】 评价文本的比较句识别是比较句情感分析的一项基础任务,具有重要的研究价值。提出中文比较句自动识别的方法,首先对包含至少一个关键词的候选比较句进行抽取,从而形成候选比较句集合,而后通过对候选比较句集合采用多特征融合的分类方法进行分类。实验结果表明比较句识别的性能达到87.26%的F1值。更多还原

【Abstract】 Comparative sentence identification in evaluative text is an essential task in comparative sentiment analysis,which has important research value. This paper proposes a comparative sentence identification method in Chinese. First,extract comparative candidates which contain at least a keyword to form a set of comparative candidates. Then identify comparative sentences from the set of candidates based on feature fusion. The experiment result shows the F1- score is 87.26%.更多还原