arrivals, dinner (8.00pm)
| AM | The evaluation framework | |
| 9.30 | Welcome, framework for English SENSEVAL: Adam Kilgarriff | |
| 9.50 | ROMANSEVAL framework for French and Italian: Jean Veronis | |
| 10.10 | Inter-tagger agreement for: | |
| French (10 mins, Jean Veronis) | ||
| Italian (10 mins, Nicoletta Calzolari) | ||
| English (20 mins, Adam Kilgarriff) | ||
| 11.00 | TEA/COFFEE | |
| 11.30 | Lexicographer's experience of tagging: Ramesh Krishnamurthy and Diane Nicholls | |
| 12.00 | Results for: | |
| French (15 mins, Jean Veronis) | ||
| Italian (10 mins, Nicoletta Calzolari) | ||
| English (30 mins, Joseph Rosenzweig) | ||
| 1.00 | LUNCH | |
| PM | Participating systems | |
| (24 participating systems for English, French or Italian) | ||
| 2.30 | 10 x 5 mins presentations + half-hour discussion | |
| 4.00 | TEA/COFFEE | |
| 4.30 | Debate: "Computational criteria for consistent sense distinctions" | |
| Chair: Martha Palmer | ||
| Panelists: Nancy Ide, David Yarowsky, Rosamund Moon | ||
| 7.30 | RECEPTION / BUFFET |
| AM | Participating systems (cont.) |
| 9.30 | Report on ARCADE (evaluation for text alignment): Jean Veronis |
| 9.50 | 8 x 5 mins presentations + half-hour discussion |
| 11.00 | TEA/COFFEE |
| 11.30 | 7 x 5 mins presentations |
| 12.10 | Plenary discussion: what worked well, where and how? |
| 1.00 | LUNCH |
| PM | There will be a meeting of the ARCADE project, for details see: http://www.lpl.univ-aix.fr/projects/arcade/ |
| Other options include: | |
| * cycling, walking etc in vicinity of Herstmonceux (including Roman ruins of Pevensey castle) | |
| * trip into Brighton to sample delights of UK's foremost seaside resort! |
| AM | Pressing Issues |
| 9.30 | Is Word Sense Tagging just one more task, or is it special? Yorick Wilks |
| 9.50 | More than one sense per discourse: Bob Krovetz |
| 10.10 | Topics listed below form a preliminary list: suggestions for additions
and other changes are welcome. We are canvassing levels of interest to
work out what should be plenary, what should be in parallel etc. Feedback
please!
* What lexical resources/sense inventories to use * Selection of corpus data * Word selection; lexical-sample and all-words tasks * Interaction with POS-tagging * Scoring * What sorts of lexical representations were used by participating systems: implications for, e.g., EAGLES working group on lexical semantics standards * Tools for manual sense-tagging * SENSEVALs for further languages * Mapping senses between dictionaries |
| 1.00 | LUNCH |
| PM | The way ahead |
| 2.30 | Invited Speaker: Patrick Paroubek, ELSE (Evaluation in Language and Speech Engineering) ELSE and future SENSEVALs |
| 3.00 | * Should there be a next SENSEVAL, and if so, when, where and how (to include: who should bid for funding, and from whom) |
| * Was the pilot SENSEVAL model a good one? Which tasks did it miss out, which systmes did it unfairly disadvantage? | |
| 4.00 | TEA, DEPARTURES |
This page is maintained by Carole
Tiberius and was last modified on 14
July 1998.