Abstract
Dependent bigrams are two consecutive words that occur together in a text more often than would be expected purely by chance. Identifying such bigrams is an important issue since they provide valuable clues for machine translation, word sense disambiguation, and information retrieval. Minimum sensitivity is one of the proposal made to identify these lexical pairs. It is simple to compute and is free from the underlying distributional assumptions made by significance tests. Experimental results show that minimum sensitivity results in the identification of bigrams that are largely made up of content words. The tendency of minimum sensitivity to filter out bigrams containing non-content words is an important quality in language processing applications.
Original language | English (US) |
---|---|
Title of host publication | Innovative Applications of Artificial Intelligence - Conference Proceedings |
Editors | Anon |
Publisher | AAAI |
Number of pages | 1 |
State | Published - Jan 1 1998 |
Event | Proceedings of the 1998 10th Conference on Innovative Applications of Artificial Intelligence, IAAI - Madison, WI, USA Duration: Jul 26 1998 → Jul 30 1998 |
Other
Other | Proceedings of the 1998 10th Conference on Innovative Applications of Artificial Intelligence, IAAI |
---|---|
City | Madison, WI, USA |
Period | 7/26/98 → 7/30/98 |