search

tarsqi / ttk

Tarsqi Toolkit
Apache License 2.0
25 stars 10 forks source link

Incomplete output from the pipeline (upto BLINKER) #26

Closed sanjibksaha closed 7 years ago

sanjibksaha commented 8 years ago

wsj_0169.txt

The processing did not complete and generated an incomplete output file. WSJ_0169.TXT

The latest TTK pipeline (upto BLINKER module, Python version) has been run on this file. Here is the console output:

C:\Users\sahask\workspace\TTK-2-UIMA-Fit\Tarsqi\resources\ttk\code>python tarsqi .py --source=timebank --pipeline=PREPROCESSOR,GUTIME,EVITA,SLINKET,S2T,BLINKER P :\workspace\ValidationData\TimeBank_dataset\TimeBank_unannotated\wsj_0169.xml C: \Users\sahask\workspace\TTK-2-UIMA-Fit\Resources\out\test1.out reading parameters ... tagging ... ERROR printing output finished.

We also tried to run the entire pipeline, provides the same result.

marcverhagen commented 8 years ago

Fixed in https://github.com/tarsqi/ttk/commit/bab615ca2b5f41317d90cb1cf9e0894eb6886ce8.

However, this issue does bring up the fact that the behavior of Evita when there are already event tags in the input is not well-defined (see issue https://github.com/tarsqi/ttk/issues/3). This is probably true for other components too.

Update (3/31/17). It is actually clear what happens when there are EVENT tags in the input: they are put in the source_tags repository and will not influence any subsequent Tarsqi processing.

sanjibksaha commented 8 years ago

wsj_0169.txt

I tried with this Timebank file and got the same problem. I uploaded the wrong file in my previous comment.

marcverhagen commented 7 years ago

I can run the code on wsj_0169.txt just fine.