Application of LSA space´s dimension character in document multi-hierarchy clustering

In LSA space, dimensions corresponding to bigger singular values reflect the general concept of language elements, while dimensions corresponding to smaller singular values reflect particular concept of language elements. On this basis, different dimensions of LSA space are adopted for document clustering under various concept granularities. In addition, in the LSA-based algorithm of document clustering, better clustering results are obtained by taking the row vectors of document self-indexing matrix as the objects to be clustered, instead of the document vectors with low dimensionality.