Abstract

Sentence and document similarity assessment iskey to most NLP applications. This paper presents a novelmeasure of calculating the similarity between sentences orbetween documents using ontology. The similarity isassessed using sentence or document concept vector formingfrom finding the linkage between ontology terms andsentence or document content, the linage can be used togenerate semantic indexes of sentences or document andapply them to implement highly efficient searchingalgorithms to compute sentence or document similarity, andthe difference between the sentence and document similaritymeasurement is articulated. Results were verified throughexperiments. Experiments show that this technique isefficient and compares favorably to other similaritymeasures, and it is flexible enough to allow the user to makecomparisons without any additional dictionary or corpusinformation. We believe that this method can be applied in avariety of text knowledge representation and discoveryapplications.