Assessing Text Semantic Similarity Using Ontology

Sentence and document similarity assessment is key to most NLP applications. This paper presents a novel measure of calculating the similarity between sentences or between documents using ontology. The similarity is assessed using sentence or document concept vector forming from finding the linkage between ontology terms and sentence or document content, the linage can be used to generate semantic indexes of sentences or document and apply them to implement highly efficient searching algorithms to compute sentence or document similarity, and the difference between the sentence and document similarity measurement is articulated. Results were verified through experiments. Experiments show that this technique is efficient and compares favorably to other similarity measures, and it is flexible enough to allow the user to make comparisons without any additional dictionary or corpus information. We believe that this method can be applied in a variety of text knowledge representation and discovery applications.