TRANSDUCTIVE LEARNING FOR SHORT-TEXT CLASSIFICATION PROBLEMS USING LATENT SEMANTIC INDEXING

Abstract

This paper presents work that uses Transductive Latent Semantic Indexing (LSI) for text classification. In addition to relying on labeled training data, we improve classification accuracy by incorporating the set of test examples in the classification process. Rather than performing LSI's singular value decomposition (SVD) process solely on the training data, we instead use an expanded term-by-document matrix that includes both the labeled data as well as any available test examples. We report the performance of LSI on data sets both with and without the inclusion of the test examples, and we show that tailoring the SVD process to the test examples can be even more useful than adding additional training data. This method can be especially useful to combat possible inclusion of unrelated data in the original corpus, and to compensate for limited amounts of data. Additionally, we evaluate the vocabulary of the training and test sets and present the results of a series of experiments to illustrate how the test set is used in an advantageous way.