Open Access Open Access  Restricted Access Subscription Access
Cover Image

An Graph based Sentence Level Semantic Linkage Weighing Model for Efficient Text Clustering

Sharmila V, Tholkappia Arasu G, Balamurugan P


Text clustering being performed in different approaches from term level to document level similarity and there are many clustering approaches discussed earlier. The earlier methods performs clustering using similarity measures which is performed based on term level co-occurrence, which does not make sense where multi concept linkage present in documents. This paper, proposes a graph based sentence level semantic linkage weighting (GSSWM) approach to cluster the text documents. Each statement of the document is considered as a node of graph and for each node the method computes semantic similarity for each sentence of document towards the available classes. The method maintains number of semantic concepts and their linkage with the terms under the class as well as the terms of other semantic classes. The semantic linkage represents the interior and exterior relations the node has with the semantic graph. Based on computed sentence level semantic linkage weight, the text document is clustered and the method produces efficient clustering than other methods in all the factors of text clustering.

Full Text:



  • There are currently no refbacks.