摘要
在线评论已成为互联网环境下用户需求获取的重要数据资源.然而,评论质量的良莠不齐严重干扰了需求挖掘的准确性和可信性.如何发现能够准确描述用户需求的有用评论是提高需求获取技术有效性的前提保障.针对这一问题,文中提出一种基于复杂网络的评论有用性分析方法,利用评论间的语义关联,从宏观的角度分析评论对于用户需求识别的有用程度,进而发现能够准确描述用户需求的评论.作者将评论看作一种内容互连的网络拓扑的形态,利用评论网络节点的重要性来度量评论的有用性,并通过拓扑势理论将用户的主观评价与网络拓扑结构的客观影响有机融合对评论网络节点重要性进行分析.实验结果表明,该方法所确定的高有用性评论能够保证用户需求获取具有较高的准确率和覆盖率.
Online reviews have become a novel data resource for requirements elicitation. How- ever, the quality differences of reviews cause great difficulty in eliciting highly accurate and relia- ble user requirements by automatic mining technology. The prerequisite of user requirements elicitation is how to discover more helpful reviews that can describe user requirements accurately. For this problem, this paper proposes an approach to analyzing the helpfulness of a review for requirements elicitation based on Complex Network Theory. In our approach, we employ the semantic association between the reviews to analyze the degree to which a review is helpful for identifying user requirements, and then discover the reviews that can describe user requirements accurately. Our approach regards the reviews as a network topology with associated content, and uses the node importance of a review in the network to measure the helpfulness of the review. The node importance of a review is computed based on topological potential by integrating the users＇ subjective evaluation and the objective influence of the network topology. Experimental results show that our approach can identify more helpful reviews to support user requirements elicitation with sufficient accuracy and coverage.
出处
《计算机学报》
EI
CSCD
北大核心
2013年第1期119-131,共13页
Chinese Journal of Computers
基金
国家自然科学基金(61170087)
软件开发环境国家重点实验室自主研究课题(SKLSDE-2012ZX-13)资助~~
关键词
需求获取
在线评论有用性
语义关联
复杂网络
拓扑势
requirements elicitation
helpfulness of online reviews
semantic association
complexnetwork
topological potential
