An improved KNN algorithm for text classification

Abstract

This paper analyzes the advantages and disadvantages of KNN alogrithm and introduces an improved KNN alogrithm (WPSOKN) for text classification. It is based on particle swarm optimization which has the ability of random and directed global search within training document set. During the procedure for searching k nearest neighbors of the test sample, those document vectors that are impossible to be the k closest vectors are kicked out quickly. Besides it reduces the impact of individual particles from the overall. Moreover, the interference factor is introduced to avoid premature to find the k nearest neighbors of test samples quickly. We conducted an extensive experimental study using real datasets, and the results show that the WPSOKNN algorithm is more efficient than other KNN algorithm.