An Improved KNN Text Classification Algorithm Based on K-Medoids and Rough Set
Yuxuan Tan · 2018
As an important algorithm used in text classifications, the K-Nearest Neighbors (KNN) has the advantage of simple and effective. However, its computational overhead is very high. Addressing this problem, an improved KNN algorithm named KM-RS-KNN which introduces the K-medoids and rough set to the KNN is proposed in this paper. The K-medoids is used to reduce the training data by similarity between the new incoming text and the cluster center, while the upper and lower approximations in the rough set theory are used to reduce the sample search space. The simulation shows that KM-RS-KNN has better classification efficiency and classification effect than the traditional KNN algorithm.