Abstract

In this talk, we report on our recent experiments and results on automatic classification of free text documents into one or more of a given number of categories using an approach we developed recently. We use different kinds of feature extractors to derive information from documents and incorporate neural network learning into the approach. Here, neural networks also act as a fusion mechanism for the different features extracted from the documents to be classified. We then compare the effectiveness of individual feature extractors in classifying the text and that of different interesting combinations of feature extractors using different metrics. The text used in the classification experiments is taken from the Reuters-22173 collection. The results indicate that a larger "reference library" is not necessarily more effective. However, information fusion almost always performs better than information from the individual feature extractors, and certain combinations seem to do better than the others.

About the Speaker