A New Model of Information Content for Measuring the Semantic Similarity between Concepts

作者：Qingbo Yuan, Zhongqing Yu, Kaixi Wang · 年份：2013 · DOI：10.1109/cloudcom-asia.2013.25 · 被引用次数：18 · 研究领域：Advanced Text Analysis Techniques、Topic Modeling、Biomedical Text Mining and Ontologies

The information content (IC) of concepts is a fundamental dimension in semantic similarity calculation. The IC of a concept is able to provide an evaluation of its degree of semantic generality and concreteness, which is great important semantic evidence modeled in the ontology. A proper quantification of IC requires an accurate evaluation of the structural differences among different concepts. This paper analyses several existing IC models and some structural factors in the ontological structure. After that, this paper proposes a criterion for evaluating IC models and a novel model employing three important factors to compute the concept's IC. This model is evaluated on two different test datasets, and the experiments show that our IC model distinguishes concepts with different topology in the taxonomy more effectively than other IC models, and the corresponding similarities are better correlated with human judgments than most other works.