A Comparative Study using Vector Space Model with K-Nearest Neighbor on Text Categorization Data

Text categorization is one of the well studied problems in data mining and information retrieval. Given a large quantity of documents in a data set where each document is associated with its corresponding category. Categorization involves building a model from classified documents, in order to classify previously unseen documents as accurately as possible. In this paper, we investigate variations of vector space model using inverse document frequency (IDF) and weighted inverse document frequency (WIDF). Experimental results against eight different data sets provide evidence that the Cosine Coefficient outperformed Jaccard and Dice Coefficient approaches with regards to F1 measure results, and the Cosine-based IDF achieved the highest average scores.