Semantic similarity

Semantic proximity is a concept that the similarity of documents or terms by way of its contents (meaning, semantics ) describes. Can illustrate one semantic proximity using " maps" in which similar documents or terms are moved together close but dissimilar further apart. This happens, for example - often involuntarily - in the creation of Mind Maps.

There are also several visualization tools for the web, thanks to which illustrates the semantic proximity of websites ( content) you ( KartOO, WebBrain ).

The definition of the semantic proximity is an important problem area in the use of ontologies for semantic annotation and semantic search, in particular in the Semantic Web.

Formalization of the semantic proximity

Lying terms in a tree structure ordered before, such as in taxonomies, this is how the semantic proximity of two terms as define the length of the shortest path between the concept nodes.

If you have information about the frequency of occurrence ( p) of the hierarchized terms t, for example, from the analysis of a corpus of texts, these can be quantified by means of their information content I:

A metric for the concept tree can then be defined as follows:

If t1 superclass of t2 and

If t1 common superclass of t2 and t3.

Extensions of this simple rule relating the density of the tree node and the absolute depth in the tree with a.