Developing Dynamic and Adaptive Methods for Turkish Spam Filtering
BAP funded project on spam filtering methods for Turkish emails.
Status
Completed 2003 - 2004
Developing Dynamic and Adaptive Methods for Turkish Spam Filtering
Funding Agency: Bogazici University Research Fund, BAP (Project 04A101)
Project Manager: Tunga Güngör
Dates: 2003-2004
In this project, spam-prevention filtering methods were developed for Turkish email messages. At the time, spam messages constituted 10% of all email messages and caused significant time loss for users. While filtering algorithms existed for common languages like English, no such work had been done for Turkish messages.
The methods developed in this project were dynamic and based on neural networks and Bayesian networks techniques. The algorithms included two main components: a morphology module that performs morphological analysis of message contents and a learning module that classifies messages as normal and spam. Turkish's complex morphological structure was taken into account throughout the development.