Expert‐Built and Collaboratively Constructed Lexical Semantic Resources

Abstract

AbstractKnowledge about words, their meanings, and their relations to other words contained in lexical semantic resources is of particular interest for the automatic processing of human language. In the last decades, expert‐built lexical semantic resources such as WordNet have been utilized in a vast number of natural language processing (NLP) tasks. Recently, collaboratively constructed resources such as the online encyclopedia Wikipedia with its exceptional scale and Wiktionary, a combination of dictionary and thesaurus, have been discovered as valuable substitutes for expert‐built resources. In this study, we first introduce diverse types of lexical semantic resources with respect to their content and structure. We identify the differences between expert‐built and collaboratively constructed resources and compare WordNet, Wiktionary, and Wikipedia. We provide a comprehensive overview of the lexical semantic knowledge therein and discuss their assets and drawbacks. Finally, we review work on orchestrating different resources in order to combine their strengths and explore their use in major NLP applications.