wtangdev / UniRel

released code for our EMNLP22 paper: UniRel: Unified Representation and Interaction for Joint Relational Triple Extraction
Apache License 2.0
82 stars 17 forks source link

Extended to Multi-token Entity Setting #3

Closed UnkownSomeone closed 1 year ago

UnkownSomeone commented 1 year ago

Thanks for your awosome work, but I cant't find the Extended to Multi-token Entity Setting code.

wtangdev commented 1 year ago

Hi, we will update the extension later. You can also implement it by following the instructions given in the paper. All you need is Adding implementations of head-head and head-tail for each relation, which is very same as the implementation of tail-tail. Files need to be edited is data_processor.py, dataset.py and model_transformer.py.

UnkownSomeone commented 1 year ago

Thanks a lot.

bablf commented 1 year ago

Are you still planing on releasing the extension? Also I am quite sure that data_extractor.py would need to be modified as well since only tail labels are evaluated.

The description of what to do is also a bit vague. It would be great if you could either contribute a detailed description or just publish your code (even if it currently does not work out of the box)

wtangdev commented 1 year ago

Hi. Codes are updated to add the multi-token recognition implementations. Check it now~ @UnkownSomeone @bablf