To search, Click below search items.


All Published Papers Search Service


New Modified Semantic Similarity Measure based on Information Content Approach


Nababteh Mohammed, Deri Mohammed


Vol. 17  No. 3  pp. 73-76


In recent years, the semantic similarity measure has got a great concern especially in NLP, information retrieval and cognitive science. Several approaches have been introduced for computing the semantic similarity score among concepts. This paper presents a modified semantic similarity measure of LIN measure. The proposed method focuses on solving the low similarity score between synonyms, as well as avoiding zero similarity score when the concept has no occurrence in corpus. The proposed measure computes the similarity score using the parents of the compared concepts and least common subsumer. The experiments show that the proposed measure has achieved high correlation against LIN measure on the miller and Charles benchmark dataset, also the MSE value of the proposed measure was 0.263758, on the other hand the MSE of LIN measure on the same dataset was 0.344196.


Ontology, WordNet, semantic similarity, similarity measures.