A Strategy of Mapping Polish WordNet onto Princeton WordNet

Abstract

We present a strategy and the early results of the mapping of plWordNet ‐ one of the largest such language resources in existence ‐ onto Princeton Wo rdNet. The fundamental structural premise of plWordNet differs from those of most other wordnets: lexical units rather than synsets are the basic building blocks. The addition of new material to plWordNet is consistently informed by semantic relations and by various analyses of large corpora. The mapping is difficult because of the subtly distinct structures and because of WordN et’s focus on synsets. We have designed a set of inter-lingual semantic relations and an effectiv e mapping procedure. In the course of mapping, we have discovered a range of systematic difference s between plWordNet and WordNet, and proposed ways of accounting for such differences.