|
| Speaker: | Adam Kilgarriff |
| ITRI, University of Brighton, UK | |
| Date: | Tuesday 26th October 1999 |
| Time: | 11:00-12:30 |
| Place: | Room E6A 357, Macquarie University |
Abstract: As dictionaries tell us, most common words have more than one meaning. When a word is used in a book or in conversation, generally speaking, just one of those meanings will apply. This is not a problem for people. We are very rarely slowed down in our comprehension by the need to work out which meaning of a word applies. But it is for computers. It results in a lot of misinterpretation.
For forty years now, people have been writing computer programs to do Word Sense Disambiguation (WSD). It has always been hard to say how well these program do, and how good they are at the task, because success rates are dependent on the dictionary, the words chosen, and the text to be disambiguated, and also because it depends on a person going through to say what the right answer is. In this paper I describe the first ever open evaluation exercise for WSD programs, in which 25 systems, from three continets, took part.
About the Presenter:Adam Kilgarriff obtained his DPhil from the University of Sussex in 1992. The title was "Polysemy", and the issues explored (What does it mean for a word to have more than one meaning? How can related meanings be identified, distinguished (by people or by computers), represented?) have been central to his research since.
Dr. Kilgarriff then worked as Computational Linguist for Longman Dictionaries, developing the dictionary-writing environment and corpus provision for lexicographers, and preparing dictionary databases for NLP use.
Since 1995 Dr. Kilgarriff has been Research Fellow/Senior Research Fellow at ITRI, University of Brighton where he has been investigating foundational issues in the use of language corpora, various aspects of dictionary development, and Word Sense Disambiguation software. In 1998 he co-ordinated SENSEVAL, the first open evaluation exercise for disambiguation programs, and he is currently developing a lexicographer's workbench that will support high-accuracy disambiguation.
Enquiries: sals@mri.mq.edu.au
| Last modified: 13th October 1999 |