Towards the Methodology for Extending Princeton WordNet

Abstract

The paper presents the methodology and results of the first, pilot stage of the extension of Princeton WordNet, a huge electronic English language thesaurus and lexico-semantic network based on synsets, ie. sets of synonymous lexical units, or lemma sense pairs. The necessity for such extension arose in the course of mapping plWordNet (Polish WordNet -Słowosieć) onto Princeton WordNet, which produced a large number of inter-lingual hyponymy links signalling differences in the structure and lexical coverage of the two networks. The proposed strategy uses I-hyponymy links as pointers to presumed gaps in the lexical coverage of PrincetonWordNet and offers strategies of filling them in with new lexical units and synsets.

Equivalent sentiment measures for cross-language analysis of corporate communications

“…synsets, which carry information about shared meanings of particular words [ 6 ]. National Wordnets (French WOLF, GermaNet, Polish plWordNet, Spanish/Italian MCR and more) are mapped between each other using an interlingual index (ILI) that references synsets from one language to another [ 7 ]. Our mapping procedure yields lists of words in two languages that can be used to generate comparable sentiment measures.…”