An improved <i>k</i>NN text classification method

• DOI: 10.1504/ijcse.2019.103944
• Publication Date: 2019-12-04T07:35:44Z
• Authors: Fengfei Wang; Zhen Liu; Chundong Wang
• Keywords: Feature vector; Feature (linguistics)

This paper proposes an improved kNN text classification method. The algorithm in vector space models (VSM) has several limitations, such as occupying excessive storage and all dimensions the share same weight, making inaccurate. To solve these problems, this a SOM neural network with principal component weighting. In model, analysis process is embedded into network. Specifically, used to extract main feature components of assessed target. Then, it inputted for computation. Meanwhile, variance contribution rates are introduced Euclidean distance function forms weights. Using weighting compute weights VSM together could effectively reduce space, increase precision speed kkNN