Discriminative Category Matching: Efficient Text Classification for Huge Discriminative Category Matching: Efficient Text Classification for Huge
Gabriel Pui Cheong Fung, Jeffrey Xu Yu, Hongjun Lü · International Conference on Data Mining · 2002
With the rapid growth of textual information availableon the Internet, having a good model for classifying andmanaging documents automatically is undoubtly important.When more documents are archived, new terms, new conceptsand concept-drift will frequently appear. Without adoubt, updating the classification model frequently ratherthan using the old model for a very long period is absolutelyessential. Here, the challenges are: a) obtain a highaccuracy classification model; b) consume low computationaltime for both model training and operation; and c)occupy low storage space. However, none of the existingclassification approaches could achieve all of these requirements.In this paper, we propose a novel text classificationapproach, called Discriminative Category Matching, whichcould achieve all of the stated characteristics. Extensive experimentsusing two benchmarks and a large real-life collectionare conducted. The encouraging results indicatedthat our approach is hignhly feasible.