GermaNet-based Semantic Relation Pairs involving Coherent Mini-Networks
Silke Scheible und Sabine Schulte im Walde
We compiled a large-scale dataset containing German concept pairs related via paradigmatic semantic relations. The dataset is based on GermaNet and consists of two parts: A representative selection of lexical units drawn from the three major word classes adjectives, nouns, and verbs, which are balanced according to semantic category, polysemy, and type frequency ('SemrelTargets'); and a set of semantically coherent GermaNet subnets consisting of semantic relations pairs clustering around the selected targets ('Semrel- Nets').
The database contains 99 SemrelTargets for each of the three word classes, and a total of 1,623 relation pairs distributed across the respective subnets.
The SubCat-Extractor is freely available for education, research and other non-commercial purposes. Please contact the SemRel group to obtain the tool.