POS Tagging and Less Resources Languages Individuated Features in CorpusWiki

摘要

CorpusWiki (http://www.corpuswiki.org) is an online tool for building POS tagged corpora in (almost) any language. The system is primarily aimed at those languages for which no corpus data exist, and for which it would be very difficult to create tagged data by traditional means. This article describes how CorpusWiki uses individuated morphosyntactic features to combine the flexibility required in annotating less-described languages with the requirements of a POS tagger.