A semantic similarity approach combining location and intrinsic information content

To measure semantic similarity between terms is an important issue in many research fields. In this paper, a new semantic similarity approach, which combines the intrinsic information content of the term and the location of the term in the directed acyclic graph, is presented. The approach first calculates the sub graphs of two terms based the directed acyclic graph, and then calculates the intersection and union of the sub graphs. The semantic similarity of two terms is the ratio of the total intrinsic information content of terms in the intersection to the total intrinsic information content of terms in the union. Experimental evaluations using MeSH biomedical ontology indicate that the proposed approach yields results that correlate more closely with human assessments than other.