Digesting Virtual “Geek” Culture: The Summarization of Technical Internet Relay Chats

Proceedings of the 43rd Annual Meeting of the ACL, pages 298–305, Ann Arbor, June 2005. c⃝2005 Association for Computational Linguistics Digesting Virtual “Geek” Culture: The Summarization of Technical Internet Relay Chats Liang Zhou and Eduard Hovy University of Southern California Information Sciences Institute 4676 Admiralty Way Marina del Rey, CA 90292-6695 {liangz, hovy} @isi.edu Abstract This paper describes a summarization system for technical chats and emails on the Linux kernel. To reflect the complex- ity and sophistication of the discussions, they are clustered according to subtopic structure on the sub-message level, and immediate responding pairs are identified through machine learning methods. A re- sulting summary consists of one or more mini-summaries, each on a subtopic from the discussion. 1 Introduction The availability of many chat forums reflects the formation of globally dispersed virtual communi- ties. From them we select the very active and growing movement of Open Source Software (OSS) development. Working together in a virtual community in non-collocated environments, OSS developers communicate and collaborate using a wide range of web-based tools including Internet Relay Chat (IRC), electronic mailing lists, and more (Elliott and Scacchi, 2004). In contrast to conventional instant message chats, IRCs convey engaging and focused discussions on collaborative software development. Even though all OSS par- ticipants are technically savvy individually, sum- maries of IRC content are necessary within a virtual organization both as a resource and an or- ganizational memory of activities (Ackerman and Halverson, 2000). They are regularly produced manually by volunteers. These summaries can be used for analyzing the impact of virtual social in- teractions and virtual organizational culture on software/product development. The emergence of email thread discussions and chat logs as a major information source has prompted increased interest in thread summariza- tion within the Natural Language Processing (NLP) community. One might assume a smooth transition from text-based summarization to email and chat-based summarizations. However, chat falls in the genre of correspondence, which re- quires dialogue and conversation analysis. This property makes summarization in this area even more difficult than traditional summarization. In particular, topic “drift” occurs more radically than in written genres, and interpersonal and pragmatic content appears more frequently. Questions about the content and overall organization of the sum- mary must be addressed in a more thorough way for chat and other dialogue summarization sys- tems. In this paper we present a new system that clus- ters sub-message segments from correspondences according to topic, identifies the sub-message segment containing the leading issue within the topic, finds immediate responses from other par- ticipants, and consequently produces a summary for the entire IRC. Other constructions are possi- ble. One of the two baseline systems described in this paper uses the timeline and dialogue structure to select summary content, and is quite effective. We use the term chat loosely in this paper. Input IRCs for our system is a mixture of chats and 298 emails that are indistinguishable in format ob- served from the downloaded corpus (Section 3). In the following sections, we summarize previ- …

Content truncated to 500 words