Enriching SerbianWordNet and electronic dictionaries with terms from the culinary domain

Abstract

In this paper we present three lexical resources for Serbian that are crucial for the development of applications in the culinary domain based on natural language processing. The first two of them - Serbian WordNet and morphological e-dictionaries - have already been in development for some time, while the third one -a corpus of culinary recipes-has been developed specifically for this purpose. In this paper, we present how we use each of these resources to correct and enlarge the other two. We use various automatic procedures, but manually check all the results.