Dear participants,
The table of results by team/run is available in the following address:
http://nlp.uned.es/weps/weps2/weps2_clustering_ranking.html
The teams/runs are sorted according to F-measure of B-Cubed Precision and
Recall with alpha set to 0.5. Note that F-measures are macro-averaged over all test sets.
Legend:
BEP: BCubed Precision
BER: BCubed Recall
FMeasure_0.5_BEP-BER: F-measure of B-Cubed P/R with alpha set to 0.5
FMeasure_0.2_BEP-BER: F-measure of B-Cubed P/R with alpha set to 0.2
P: PurityIP: Inverse Purity
FMeasure_0.5_P-IP: F-measure of Purity and Inverse Purity with alpha set to 0.5
FMeasure_0.2_P-IP: F-measure of Purity and Inverse Purity with alpha set to 0.2For more details about the evaluation metrics please refer to:
A comparison of extrinsic clustering evaluation metrics based on formal constraints.
E. Amigó, J. Gonzalo, J. Artiles, F. Verdejo.
Information Retrieval Journal. 2008.
The baselines and the rationale for F-measures with alpha 0.2 are explained in the WePS1 task
description paper:
The SemEval-2007 WePS Evaluation: Establishing a
benchmark for the Web People Search Task.
Javier Artiles, Julio Gonzalo and Satoshi
Sekine.
Proc. Int'l Workshop Semantic Evaluations (SemEval 2007).Best regards (and happy holidays!),
Javier Artiles (on behalf of the WePS organizers).
--
------------------------------
------------------------------------------------