What’s with the Attitude? Identifying Sentences with Attitude in Online Discussions

Proceedings of the 2010 Conference on Empirical Methods in Natural Language Processing, pages 1245–1255, MIT, Massachusetts, USA, 9-11 October 2010. c⃝2010 Association for Computational Linguistics What’s with the Attitude? Identifying Sentences with Attitude in Online Discussions Ahmed Hassan Vahed Qazvinian University of Michigan Ann Arbor Ann Arbor, Michigan, USA hassanam,vahed,radev@umich.edu Dragomir Radev Abstract Mining sentiment from user generated content is a very important task in Natural Language Processing. An example of such content is threaded discussions which act as a very im- portant tool for communication and collabo- ration in the Web. Threaded discussions in- clude e-mails, e-mail lists, bulletin boards, newsgroups, and Internet forums. Most of the work on sentiment analysis has been centered around finding the sentiment toward products or topics. In this work, we present a method to identify the attitude of participants in an online discussion toward one another. This would enable us to build a signed network representation of participant interaction where every edge has a sign that indicates whether the interaction is positive or negative. This is different from most of the research on so- cial networks that has focused almost exclu- sively on positive links. The method is exper- imentally tested using a manually labeled set of discussion posts. The results show that the proposed method is capable of identifying at- titudinal sentences, and their signs, with high accuracy and that it outperforms several other baselines. 1 Introduction Mining sentiment from text has a wide range of applications from mining product reviews on the Web (Morinaga et al., 2002; Turney and Littman, 2003) to analyzing political speeches (Thomas et al., 2006). Automatic methods for sentiment mining are very important because manual extraction of them is very costly, and inefficient. A new application of sentiment mining is to automatically identify atti- tudes between participants in an online discussion. An automatic tool to identify attitudes will enable us to build a signed network representation of par- ticipant interaction in which the interaction between two participants is represented using a positive or a negative edge. Even though using signed edges in social network studies is clearly important, most of the social networks research has focused only on positive links between entities. Some work has re- cently investigated signed networks (Leskovec et al., 2010; Kunegis et al., 2009), however this work was limited to a few number of datasets in which users were allowed to explicitly add negative, as well as positive, relations. This work will pave the way for research efforts to examine signed social networks in more detail. It will also allow us to study the re- lation between explicit relations and the text under- lying those relation. Although similar, identifying sentences that dis- play an attitude in discussions is different from iden- tifying opinionated sentences. A sentence in a dis- cussion may bear opinions about a definite target (e.g., price of a camera) and yet have no attitude to- ward the other participants in the discussion. For in- stance, in the following discussion Alice’s …

Content truncated to 500 words