Editing Semantic similarity (section)

=== In ontology matching ===
Semantic similarity plays a crucial role in [[ontology alignment]], which aims to establish correspondences between [[Ontology components|entities]] from different ontologies. It involves quantifying the degree of similarity between concepts or terms using the information present in the ontology for each entity, such as labels, descriptions, and hierarchical relations to other entities. Traditional metrics used in ontology matching are based on a lexical similarity between features of the entities, such as using the Levenshtein distance to measure the edit distance between entity labels.<ref>{{Cite conference|last1=Cheatham |first1=Michelle |last2=Hitzler |first2=Pascal |title=Advanced Information Systems Engineering |chapter=String Similarity Metrics for Ontology Alignment |date=2013 |editor-last=Alani |editor-first=Harith |editor2-last=Kagal |editor2-first=Lalana |editor3-last=Fokoue |editor3-first=Achille |editor4-last=Groth |editor4-first=Paul |editor5-last=Biemann |editor5-first=Chris |editor6-last=Parreira |editor6-first=Josiane Xavier |editor7-last=Aroyo |editor7-first=Lora |editor8-last=Noy |editor8-first=Natasha |editor9-last=Welty |editor9-first=Chris |conference =The Semantic Web – ISWC 2013 |series=Lecture Notes in Computer Science |volume=7908 |language=en |location=Berlin, Heidelberg |publisher=Springer |pages=294–309 |doi=10.1007/978-3-642-41338-4_19 |isbn=978-3-642-41338-4|s2cid=18372966 |doi-access=free }}</ref> However, it is difficult to capture the semantic similarity between entities using these metrics. For example, when comparing two ontologies describing conferences, the entities "Contribution" and "Paper" may have high semantic similarity since they share the same meaning. Nonetheless, due to their lexical differences, lexicographical similarity alone cannot establish this alignment. To capture these semantic similarities, [[Latent space|embeddings]] are being adopted in ontology matching.<ref name=":0">Sousa, G., Lima, R., & Trojahn, C. (2022). An eye on representation learning in ontology matching. ''OM@ISWC''.</ref> By encoding semantic relationships and contextual information, embeddings enable the calculation of similarity scores between entities based on the proximity of their vector representations in the embedding space. This approach allows for efficient and accurate matching of ontologies since embeddings can model semantic differences in entity naming, such as homonymy, by assigning different embeddings to the same word based on different contexts.<ref name=":0" />