thank you

[Jakub Mareček]

Dear all, 

I would like to thank many of you for your support in our work on

ensemble methods for temporal-relations classifiers. Our F1 scores 

on SemEval-2013 TempEval-3 (Temporal Annotation) and 

SemEval-2016 Task 12 (Clinical TempEval) are 0.3915 and 0.595 

respectively. These are the present best, as far as I know, but rely

crucially on the performance of the individual classifiers whose 

output you have kindly provided. Thank you!

The associated paper now appears in the most recent issue of Data 

Mining and Knowledge Discovery as:

Integer programming ensemble of temporal relations classifiers 

Catherine Kerr, Terri Hoare, Paula Carroll & Jakub Mareček

https://link.springer.com/article/10.1007%2Fs10618-019-00671-x

The submitted version is attached. Our code is at:

https://github.com/jmarecek/ip-ensemble

We cannot redistribute data kindly provided by the individual participants,

but may be able to re-run on those, should there be further interest

in ensemble methods or similar. 

Please do ping me with any questions or if I can help in any way.

Thank you!

Jakub Marecek