Adaptive Information Filtering : Learning Drifting Concepts

The task of information filtering is to classify texts from a s tream of documents into relevant and nonrelevant, respectively, with respect to a particular categ ory or user interest, which may change over time. A filtering system should be able to adapt to such concept changes. This p aper explores methods to recognize concept changes and to maintain windows on the training data, whose size is ei ther fixed or automatically adapted to the current extent of concept change. Experiments with two simulated co ncept drift scenarios based on real-world text data and eight learning methods are performed to evaluate three indi cators for concept changes and to compare approaches with fixed and adjustable window sizes, respectively, to eac h other and to learning on all previously seen examples. Even using only a simple window on the data already improves t h performance of the classifiers significantly as compared to learning on all examples. For most of the classifi er , the window adjustments lead to a further increase in performance compared to windows of fixed size. The chosen ind icators allow to reliably recognize concept changes.