Semantic similarity measures are concerned with quantification of the likeness of the meanings of features that are compared. When features are obtained from classes in a taxonomy, semantic similarity is a function that assigns a numerical value to the similarity between two classes of objects. For example, two classes that have a direct common superclass are semantically more similar than two classes that are far apart in the taxonomy. CBR systems can use semantic approaches to retrieve cases. Typically, such CBR systems utilize a knowledge representation method combined with an approach for measuring the similarity of the meaning of things represented in the cases. The latter is known as semantic similarity. This chapter describes traditional similarity measures as well as recent approaches and applications.
