Missing entity recognition component for prediction

[mikelkl]

Hi there,

Thanks for your kindly code releasing.

when I was reading the code, I found that the golden-entity-mentions data is provided in advance. Also, in forward function of EDModel, batch_golden_entities argument is required.

Does them mean when I wish to predict data without entity label, an extra NER system is needed to recognize entities before and then feed results into JMEE?

Thanks & Regards, Mike

[airkid]

I believe so.
In ACE2005 Dataset, the entity mention are pre given.
If you want to apply this code to a plane text data set there should be a NER result before this project.

[mikelkl]

Hi @airkid, yeah, if our idea is correct, it's not a completed joint model I think.

[airkid]

Hi @mikelkl , joint means predict trigger and argument in one model in this task.
The opposite model structure is called pipe line model, which extract event mention in model 1 and send the result (text result) into model 2 for argument extraction.
In the source paper of this project, you can find DMCNN in the experiment part, which is a famous pipe line model as baseline method.

[mikelkl]

Hi @airkid , does DMCNN also need pre given entities? If so, maybe both the "joint" approach and the "pipeline" approach assume entities is known, then maybe the argument extraction phase is actually the argument role classification for each entity-trigger pair.

[airkid]

Hi @mikelkl , you are right they both need per given entities.