Combining opinion mining with social network analysis in opinion leader identification

Daniel Zeng, Hengmin Zhou, Yilu Zhou

Research output: Contribution to conferencePaper

Abstract

The increasing popularity of social networking websites facilitates expressing people's personal opinions in these Web 2.0 online channels. While social network analysis has been an important approach in opinion leader identification, actual opinions expressed in online communities have not been taken into consideration during this process. We propose Opinion Rank to enrich traditional social network analysis by incorporating sentiment of opinions mined from online user-generated content. In addition, we incorporate Opinion Rank with traditional node-based ranking to identify opinion leaders. Our experiments on real-world datasets of online reviews show that a hybrid approach combining both members' activity scores and Opinion Rank scores achieved the best performance in finding opinion leaders in large networks. Our study bridges two research areas, opinion mining and social network analysis, and demonstrates that sentiments among members can provide important clues for identifying experts or authorities in online communities.

Original languageEnglish (US)
Pages157-162
Number of pages6
StatePublished - Jan 1 2011
Event21st Workshop on Information Technologies and Systems, WITS 2011 - Shanghai, China
Duration: Dec 3 2011Dec 4 2011

Other

Other21st Workshop on Information Technologies and Systems, WITS 2011
CountryChina
CityShanghai
Period12/3/1112/4/11

ASJC Scopus subject areas

  • Information Systems

Fingerprint Dive into the research topics of 'Combining opinion mining with social network analysis in opinion leader identification'. Together they form a unique fingerprint.

  • Cite this

    Zeng, D., Zhou, H., & Zhou, Y. (2011). Combining opinion mining with social network analysis in opinion leader identification. 157-162. Paper presented at 21st Workshop on Information Technologies and Systems, WITS 2011, Shanghai, China.