Online conversation mining for author characterization and topic identification
Authors:
Open Access
- 28 October 2011
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 19-26
Abstract
The increasing popularity of online-based services (Twitter, Facebook, IRC, Myspace, blogs, just to mention few of them) results in a production of a huge amount of novel documents. These documents present properties that can not be found in standard edited documents. In particular the messages generated during the use of Instant Messages Services (IM) like chat-rooms or Twitter are short, user-generated and "noisy". We are investigating two different but related aspect of the content of these colloquial messages: the topic identification and the author identification tasks. In the first case we would like to answer the question what is the conversation about? while in the second case the question is who are the people involved in the conversation?. The combination of these two tasks into a unique model is a novel and interesting research problem that is the main topic of our research.Keywords
This publication has 19 references indexed in Scilit:
- Statistics of Online User-Generated Short DocumentsLecture Notes in Computer Science, 2010
- A Comparative Study of Utilizing Topic Models for Information Retrieval2009
- PLDA: Parallel Latent Dirichlet Allocation for Large-Scale Applications2009
- A few chirps about twitter2008
- Chat mining: Predicting user and message attributes in computer-mediated communicationInformation Processing & Management, 2008
- Authorship AttributionFoundations and Trends® in Information Retrieval, 2008
- Semi-supervised Self-training for Sentence Subjectivity Classification2008
- Structural analysis of chat messages for topic detectionOnline Information Review, 2006
- Finding scientific topicsProceedings of the National Academy of Sciences, 2004
- Multi-document summarization using off the shelf compression software2003