Closed iuria21 closed 4 years ago
Can you train this project at standart dataset CoNLL (YAGO) and the TAC KBP 2010 challenge
now?
@basque21 are you trying to create a model trained just for a subset of the possible entities? Currently DeepType first does classification among "types" (e.g. person, animal, etc.) and you can switch that out for a different family of types. Then disambiguation uses the strings in the text to find matches in a KB (stored as a trie). If you want only to match against "Mortage Law" for instance, then making sure that KB contains just Mortage Law is the way to go.
Can you train this project at standart dataset
CoNLL (YAGO) and the TAC KBP 2010 challenge
now?
Currently DeepType is trained without a particular dataset (just using Wikipedia), and then you run it on the evaluation sets for TAC KBP and others. This is how the paper results were obtained. Not sure if that was your question?
@JonathanRaiman hi, I just realized that this model doesn't fit for my problem as my disambiguation can't be done finding string in the text in the KB, so I changed my point of view.
Hi, first thanks for your repo and the docs. I want to use the library for the following case and want to know if you think it could work: I have legal text with labeled references to laws and codes for these laws. I want to train the model to link the reference of the law with it's code.
I understand that I'll have to create a dataset like:
would it be possible to train with only this features?
Thanks again!