摘要
旨在对中文微博文本的句子中评价对象进行识别。评价对象识别是指识别出评论中情感表达所针对的对象,进行评价对象的识别有助于对事件发展状况进行监控管理。目前,针对中文微博领域评价对象识别的研究较少。由于微博文本的句子简短、语言表达不够规范且表达的观点缺少带情感倾向性的词语(评价词),因而传统的通过评价词来找到评价对象的方法不适用于微博文本。利用词性分析提取和过滤评价对象候选词,并结合语义分析对句子中的候选词进行分类,基于相似的句子有着相似的评价对象的假设,采用候选词的相似性迭代算法识别中文微博文本句子中的评价对象。实验结果表明,通过深入分析微博文本的语言特征提出的方法,提高了对评价对象识别的精度。
It focuses on extracting opinion targets in Chinese microblogs. Opinion target extraction aims to find the object to which the o- pinion is expressed, helping to monitor and control the development of events. At present,there are few researches on opinion target extraction in Chinese microblogs. Due to short text span,colloquial writing style and the lack of words with emotional tendency （ opinion words） in Microblogs,the traditional approaches rely on opinion words are not suitable for microblogs. In this paper, we use the part-of -speech analysis to extract and filter candidate words,and combine semantic analysis to classify candidate words. Based on the assumption that similar messages may have similar opinion targets, a similarity iterative algorithm of candidate words is proposed to extract opinion targets. Experimental results show that by deeply analyzing language features of Microblogs, the proposed method has improved high ac- curacy.
出处
《计算机技术与发展》
2017年第1期6-10,共5页
Computer Technology and Development
基金
国家自然科学基金资助项目(61202132)
关键词
评价对象
候选词提取
语义分析
相似性计算
opinion target
candidate extraction
semantic analysis
similarity calculation
