Message Feature Identification of On-line Discussions toward Consumer Based Marketing Strategies

Noriko Imafuji Yasui, Shunsuke Saruwatari, Xavier Llorà, David E. Goldberg (2007)

Illinois Technical Report No. 2007021 Link to the PDF.

This paper focuses on the online discussions done by exchanging text based messages on online message boards. The goal of this paper is to delineate discussion topics and enhance understanding discussants message features. Toward this goal, first, we propose a method for extracting important messages and terms from discussions. This method is based on a well-know link analysis algorithm, HITS. Next, we propose a message feature map, which is visualization of messages plots on a plane with the axes; centrality and novelty. Then, we classify message characteristics into four types on the message feature map, and describe a message feature transition graph. We also examine how our approaches can be used on real world setting by using data obtained by online focus group discussions. The experimental results indicate that our approach gives us intuitive understanding of the discussion topics and the message features, and helps us to identify the messages that should be focused on from the marketing view point.

About this entry

You’re currently reading “ Message Feature Identification of On-line Discussions toward Consumer Based Marketing Strategies ,” an entry on the DISCUS project

Published:
12.21.07 / 12pm
Analyze:
KeyGraph it!
Summarize it!
Category:
Selected Publications