A new collaborative semantic proximity measure

What is it about?

Search engines, continually exploring the Web, are a natural source of information on which to base a modern approach to semantic annotation. A promising idea is that it is possible to generalize the semantic similarity, under the assumption that semantically similar terms behave similarly, and define collaborative proximity measures based on the indexing information returned by search engines. In this work PMING collaborative proximity measure is introduced as a basis to extract semantic infos.

Featured Image

Why is it important?

The use of search engine-based statistics to extract semantics, instead of an ontology, provides an automated evolutionary method that doesn't need manutention, updating, domain experts. PMING distance worked very well compared to the main proximity measures used for semantics. Results were evaluated both by clustering and ranking both by human evaluation, compared with well known data sets.