CLUSTERING OF MEDLINE DOCUMENTS USING SEMI-SUPERVISED SPECTRAL CLUSTERING

AbinCherian, D.Saravanan, A.Jesudoss

Abstract: We are considering: local-content (LC) information, global-content (GC) information from PubMed and MESH (medical subject heading-MS) for the clustering of bio-medical documents. The performances of MEDLINE document clustering are enhanced from previous methods by combining both the LC and GC. We propose a semi-supervised spectral clustering method to overcome the limitations of representation space of earlier methods.

Keywords: document clustering, semi-supervised clustering, spectral clustering

DOI: https://doi.org/10.15623/ijret.2014.0303026