CSSM: A Context-Based Semantic Similarity Measure

Semantic similarity is very critical in applications that manipulate data across heterogeneous, autonomous and distributed data sources. It helps connect these data sources. Nevertheless, current semantic similarity approaches were proven to achieve a very moderate accuracy. Furthermore, they are merely applicable when we wish to determine the semantic similarity between words. In this paper, we present CSSM, a Context-based Semantic Similarity Measure. Experimental results on Miller & Charles benchmark dataset show that CSSM performs semantic similarity comparisons properly, and obtains high accuracy.