AUTOMATIC CATEGORIZATION OF CALL-FOR-PAPERS MESSAGES

Abstract

Participants of discussion lists receive a great number of messages in their mail boxes. Most of the times, only a small fraction of those messages are useful to the user. An example of such lists is the one used to spread call-for-papers for conferences and scientific journals, which are extremely useful to research groups, professors and students who develop scientific-related activities. The diversity of call-for-papers for different research fields, however, makes the separation of the most relevant ones somewhat difficult. This paper describes an intelligent web service that organizes call-for-papers stored in electronic mail accounts. The service uses a supervised learning technique kNN in order to classify call-for-papers in six major computing areas. Experiments utilizing a test base have shown accuracy of about 89%. An extension of this web service for recommendations of call-for-papers based on automatic information extraction of researchers’ Lattes curricula (CNPq) is also presented.