Towards knowledge about causal agents in wordnet
David B. Bracewell, Fuji Ren, Shingo Kuroiwa · Annual Conference on Computers · 2006
WordNet is widely used in Natural Language Processing (NLP). It however lacks certain information, such as functions and attributes for nouns. In this paper a method for obtaining this knowledge from the glosses for a subset of causal agents is given. This extra knowledge can help in Question & Answering, Machine Translation, etc. Different classes of glasses were created and based on these classes we attempt to gain knowledge in a frame-based representation about actions, attributes, and domains.