An improved KNN algorithm for text classification

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