Acquiring NLP Data by means of Games
Zuzana Nevěřilová, Marek Grác · RASLAN · 2010
For some of the NLP tasks, obtaining appropriate data is very difficult. In this paper we concentrate on acquiring NLP data by means of games. Two different projects are presented and various aspects of these games are discussed. First, we discuss public-made collections of linguistic data generally: the quality and reliability of contributors and collections, and the high dependence of number of contributions on motivation and contribution policy. Second, we describe creation of games for acquiring NLP data in detail. As example, two existing games are presented. Finally, evaluation techniques for both projects are discussed.