SIMTERM calculates the degree of similarity or synonymy between two given words, using as source of information the WordNet thesaurus. It incorporates two types of relational similarity models:
- an approach based on the sets of synonyms (synsets) of each word using distance measures such as Jaccard or Dice coefficients.
- an approach based on the hypernyms and hyponyms of each term (building a graph) and applying distance measures on graphs, such as Path Length, Wu & Palmer, ...
Previous models result in six different measures that model different similarity criteria.
SIMTERM is a project in the Citius under the Talentum program sponsored by Telefónica.
Authors
-
- Researchers
- Alberto Bugarín Diz
- Yago Fontenla Seco
- Martín Pereira Fariña
-
- Associates
- Telefónica
