Measuring similarity between ontologies is essential for knowledge-intensive applications where ontologies provide precisely defined terms. However, with the widespread use of ontologies, problems arise due to their proliferation. Ontology engineers or users often have a core ontology for browsing or querying data but need to extend, adapt, or compare it with other ontologies. To detect and retrieve relevant ontologies, measures for similarity between ontologies are needed. This paper presents a set of ontology similarity measures and a multiple-phase empirical evaluation. The purpose of ontologies is to enable data exchange at both syntactic and semantic levels. As more ontologies are constructed and used on the WWW, new challenges emerge. Ontology engineers or users often need to extend, adapt, or compare their core ontology with other ontologies. To detect and retrieve relevant ontologies, measures for similarity between ontologies on a canonical scale (e.g., the reals in [0, 1]) are required. This paper introduces a comprehensive inventory of methods for measuring similarity between real-world ontologies and provides practical experiences with these measures. Five subjects, four novices, and one ontology engineering expert, modeled ontologies in three phases about a well-known domain using domain texts. The generated ontologies were used in an empirical evaluation study of the similarity measuring framework. The paper first formally defines the ontology structure and its semantics. Then, it proposes measures for describing the similarity of different ontology parts at the lexical and conceptual levels. In Section 5, it describes the empirical evaluation study and results. Finally, it relates to other research and concludes with an outlook on future challenges. The two-layer view of ontologies considers lexical and conceptual levels. At the lexical level, terms are used to convey meanings. At the conceptual level, conceptual relations between terms are investigated.Measuring similarity between ontologies is essential for knowledge-intensive applications where ontologies provide precisely defined terms. However, with the widespread use of ontologies, problems arise due to their proliferation. Ontology engineers or users often have a core ontology for browsing or querying data but need to extend, adapt, or compare it with other ontologies. To detect and retrieve relevant ontologies, measures for similarity between ontologies are needed. This paper presents a set of ontology similarity measures and a multiple-phase empirical evaluation. The purpose of ontologies is to enable data exchange at both syntactic and semantic levels. As more ontologies are constructed and used on the WWW, new challenges emerge. Ontology engineers or users often need to extend, adapt, or compare their core ontology with other ontologies. To detect and retrieve relevant ontologies, measures for similarity between ontologies on a canonical scale (e.g., the reals in [0, 1]) are required. This paper introduces a comprehensive inventory of methods for measuring similarity between real-world ontologies and provides practical experiences with these measures. Five subjects, four novices, and one ontology engineering expert, modeled ontologies in three phases about a well-known domain using domain texts. The generated ontologies were used in an empirical evaluation study of the similarity measuring framework. The paper first formally defines the ontology structure and its semantics. Then, it proposes measures for describing the similarity of different ontology parts at the lexical and conceptual levels. In Section 5, it describes the empirical evaluation study and results. Finally, it relates to other research and concludes with an outlook on future challenges. The two-layer view of ontologies considers lexical and conceptual levels. At the lexical level, terms are used to convey meanings. At the conceptual level, conceptual relations between terms are investigated.