Abstract
This paper describes a Natural Language Processing (NLP) Program called FUNES which can learn the meaning of proper nouns (PNs) it encounters in its processing of news text. FUNES reads short newspaper stories and produces a case-frame based output which represents the events described. It is tolerant of unknown words occurring in its input and is able to build definitions for unknown PNs it encounters. The paper shows that PNs are almost always defined within the text where they occur. This means that to completely understand a text containing such definitions we must understand the definitions. The various ways that PNs can be defined are described and we show how FUNES utilises these definitions to update its lexicon. This approach offers a solution to the problem of poor proper noun coverage in Machine Readable Dictionaries.
No comments:
Post a Comment