Research on Spectral Clustering Based on Latent Semantic Index

There is a problem that the text vector dimension is too high and the algorithm is easy to fall into local optimum problem in traditional text clustering. About this problem, this paper presents a spectral clustering method based on Latent Semantic Index (LSI), which uses the advantages of both. Not only analyzed the words and semantic relations between words, but also applies to any shape of the distribution of sample data clustering. The clustering experiment of Aviation Safety Report shows that this method has a good clustering result.