Abstract: In this paper, we study the use of support vector machine in text categorizatio n. Unlike other machine learning techniques, it allows easy incorporation of new d ocuments into an existing trained system. Moreover, dimension reduction, which is usually imperative, now becomes optiona l. Thus, SVM adapts efficiently in dynamic environments that require frequent additions to the document collection. Empirical results on the Reuters-22173 collection are also discussed.
Proceedings of the International Conference on Neural Information Processing (ICONIP), pp.347-351, Kitakyushu, Japan, October 1998.
Postscript: http://www.cs.ust.hk/~jamesk/papers/iconip99.ps.gz