Ontology-based Opinion Tag Extraction of Product Reviews
The massive increase of product reviews makes the opinion mining of product reviews become a hot topic.The products and their opinion words in the product reviews are named opinion tag, and the tag extraction is the core task of opinion mining.In order to quickly extract opinion tags from the reviews for opinion mining, this paper firstly builds the model of opinion tag extraction by combining ontology, CRFs model, and dependency syntactic parsing together.Specifically, it constructs product review ontology,combined ontology and CRFs model to recognize and extract evaluation objects and evaluation words, applies syntactic parsing method to generate the opinion.Then, it selects reviews of automobile products as the raw data and conducts the comparative experiments between single syntactic parsing method and combined ontology-based method.The results show that the ontology-based method is more accurate than single syntactic parsing method, showing better applicability in practice.Finally, the discussions, and further research are summarized.
opinion tag extraction product reviews ontology syntactic parsing
WANG Ying DING Shengchun
Department of Information Management, Nanjing University of Science and Technology, Nanjing 210094, China
国际会议
武汉
英文
301-305
2016-10-21（万方平台首次上网日期，不代表论文的发表时间）