Affiliation:
1. Humboldt-Universität zu Berlin, Germany
Abstract
E-mail has become one of the most important communication media for business and private purposes. Large amounts of past e-mail records reside on corporate servers and desktop clients. There is a huge potential for mining this data. E-mail filing and spam filtering are well-established e-mail mining tasks. E-mail filing addresses the assignment of incoming e-mails to predefined categories to support selective reading and organize large e-mail collections. First research on e-mail filing was conducted by Green and Edwards (1996) and Cohen (1996). Pantel and Lin (1998) and Sahami, Dumais, Heckerman, and Horvitz (1998) first published work on spam filtering. Here, the goal is to filter unsolicited messages. Recent research on e-mail mining addresses automatic e-mail answering (Bickel & Scheffer, 2004) and mining social networks from email logs (Tyler, Wilkinson, & Huberman, 2004). In Section Background we will categorize common e-mail mining tasks according to their objective, and give an overview of the research literature. Our Main Thrust Section addresses e-mail mining with the objective of supporting the message creation process. Finally, we discuss Future Trends and conclude.
Reference21 articles.
1. Bickel, S., & Scheffer, T. (2004). Learning from message pairs for automatic email answering. Proceedings of the European Conference on Machine Learning.
2. Boykin, P., & Roychowdhury, V. (2004). Personal e-mail networks: An effective anti-spam tool. Preprint, arXiv id 0402143.
3. Cohen, W. (1996). Learning rules that classify e-mail. Proceedings of the IEEE Spring Symposium on Machine learning for Information Access, Palo Alto, California, USA.
4. Support vector machines for spam categorization
5. Scale-free topology of e-mail networks.;H.Ebel;Physical Review E: Statistical, Nonlinear, and Soft Matter Physics,2002