TEES analysis

[Negacy Hallu]

Hi,
My team is planning to make use of the already existing TEES analysis for task9.2. I can see that TEES analysis is available for both the test and training data. Are we supposed to assume there might be another test data coming or is this one the only test data that the challenge will provide us? Does anybody know if the TEES people will again provide TEES analysis if we got a new test data for some reason?

Thanks.

Neg.

[Isabel Segura]

There is not any another dataset. 

[Kevin B. Cohen]

But it doesn't violate the rules to make use of features that come
from someone else having "looked at" the test data? Just want to
clarify.

Kev

--
Kevin Bretonnel Cohen, PhD
Biomedical Text Mining Group Lead, Computational Bioscience Program,
U. Colorado School of Medicine
303-916-2417 (cell) 303-377-9194 (home)
http://compbio.ucdenver.edu/Hunter_lab/Cohen

[Isabel Segura]

Please let me know if I'm not understanding you correctly. TEES analyses are provided by the University of Turku, which is also participating in the task. Of course, this team only has access to the unlabeled test dataset.

Moreover, we trust that this team did not use any kind of human examination to analyze texts.

Here you can find an email sent by this team about the TEES analyses:
https://groups.google.com/forum/?fromgroups=#!topic/ddiextraction_semeval/6JXtbcwrQ0g

Please, let me mention the followint extract:

"It's important to note that the University of Turku is not affiliated with the SemEval-2013 workshop and that the TEES 2.1 analyses are not in any way part of the official training or test data. All of these analyses have been computed automatically and you use them entirely at your own risk. That said, hopefully they prove useful, and can help you in building the best possible DDI extraction system!"




Best,

Isabel

--
Isabel Segura Bedmar
Despacho 2.2.A.10
Telf: 91 624 99 88
Departamento de Informática
Universidad Carlos III de Madrid,

http://www.inf.uc3m.es/component/comprofiler/userprofile/isegura

[Jari Björne]

This is indeed the case, as mentioned in the TEES analyses announcement. All the analyses are generated automatically, with no access to the test labels. There's been no human examination of the test dataset.

The TEES data (parses, predictions etc.) is simply what anyone can produce using the open source and publicly available TEES 2.1 system, but since the program is somewhat time-consuming to set up, precalculated results are also provided.

Regards,
Jari