摘要
目前的主观性文本识别方法多是基于词典和统计的方法,忽略了包含丰富情感信息的主观线索,准确率不高。针对这一问题,提出以主观线索为特征的主观性文本识别方法。选取关联词、情感词以及指示性动词、感叹词、程度副词、带有情感色彩的标点符号等6种主观线索特征作为主观性文本识别依据,建立主观线索特征词表,用朴素贝叶斯分类器对主观性文本进行识别。实验结果表明,相比朴素贝叶斯+主观线索(不含关联词)方法和传统朴素贝叶斯方法,该方法准确率分别平均提高8%和14%。
Current subjective text recognition method is dictionary-based and statistical-based, which ignores the subjective cluescontaining rich emotional information, and the accuracy is not high. In view of this problem, a method of subjective text recogni- tion characterized by subjective clues was proposed. Six kinds of subjective clues ingredients including associated words, emotional words, indicative verbs, interjections, degree adverbs, punctuations with sentiment were selected as the subjective text recogni-tion basis. Subiective clues features dictionary was established, and the subiective text with the Naive Bayes classifier was identi-well as the method of traditional Naive Bayes, the proposed scheme can increase the accuracy of subjective text recognition by about 8% and 14% respectively.
出处
《计算机工程与设计》
北大核心
2015年第9期2572-2577,F0003,共7页
Computer Engineering and Design
基金
武汉大学软件工程国家重点实验室开放课题基金项目(SKLSE2012-09-30)
山西省自然科学基金项目(2013011015-2)
关键词
主观性文本识别
情感计算
主观线索
特征词表
朴素贝叶斯
subjective text recognition
emotion computation
subjective clues
features dictionary
Naive Bayes
