Finding and auto-labeling of task groups on e-mails and documents

Abstract

We propose a new method which extracts task groups on user's numerous messages such as e-mails and documents by finding the family of subsets that share common topics, sender/receivers, date or author. In addition, the method automatically gives labels to the groups by the common items. The biggest difference between the proposed method and conventional methods is that the conventional methods find completely divided clusters of documents, while the proposed method finds overlapping clusters of documents on the basis of subclass method. Therefore, the proposed method provides multiple views on the documents. We carried out experiments on the authors' e-mails and document files in order to confirm the basic property and effectiveness of the proposed method. © Springer-Verlag Berlin Heidelberg 2005.