祁瑞华 杨明昕 徐琳宏 关菁华


摘 要:[目的/意义]图书评论是出版社、图书馆和用户研究读者观点的重要线索,评论特征抽取研究是提高图书评论观点精准挖掘效率和准确率的基础性工作。[方法/过程]分别从评论特征抽取研究和图书评论特征聚类、语义表示、隐性特征抽取的典型方法等方面对国内外研究现状进行客观分析,梳理相关领域研究发展脉络和趋势。[结果/结论]指出图书评论特征抽取效率和准确率的提高需要考虑特征聚类、语义表示和隐性特征抽取等关键问题。
关键词:图书评论;特征抽取;综述
DOI:10.3969/j.issn.1008-0821.2019.09.018
〔中图分类号〕N99 〔文献标识码〕A 〔文章编号〕1008-0821(2019)09-0160-08
Abstract:[Purpose/Significance]Book review is an important clue for publishers,libraries and users to evaluate works.Feature extraction is the basic work to improve the efficiency and accuracy of the fine grained opinion mining of book reviews.[Method/Process]This paper made an objective analysis of the current research situation at home and abroad from the perspectives of the research on the reviewfeatureextraction,the clustering of book review feature,semantic representation,and the typical methods of the extraction of implicit feature,and sorted out the development context and trends of the research in related fields.[Result/Conclusion]It is pointed out that to improve the efficiency and accuracy of featureextractionin book reviews,some key issues should be considered,such as feature clustering,semantic representation and implicit featureextraction.
Key words:book review;feature extraction;survey
图书评论是出版社、图书馆和用户研究读者对作品评价信息的重要线索。图书评论特征抽取的目标是从评论文本中抽取关键要素,在此基础上研究用户对图书具体方面所持的评价和态度。评论特征抽取既能帮助用户详细了解图书质量做出购买决策,也能帮助出版社和用户实现精准观点挖掘,从而提高图书评论观点挖掘的效率和准确率。
评论特征抽取属于细粒度的观点挖掘,在情报学和自然语言处理领域受到广泛关注。在中国知网全部数据库高级检索下,检索条件为(主题=“评论”或者题名=“评论”)并且(主题=“特征”或者题名=“特征”或者主题=“挖掘”或者题名=“挖掘”或者主题=“抽取”或者题名=“抽取”)时,检索到“评论特征抽取”相关文献4 166篇。但是图书评论领域得到的关注明显不足,当检索条件加上:并且(主题=“图书”或者题名=“图书”)时文献稀缺,仅检索到63篇文献,发展年度趋势见图1。……