Creating counterfactual examples for English

[vidhig]

I am working on creating counterfactual examples for English but once the sentences are generated, there are some spacing issues, for example: Previously at other Nintendostores, there have been a broader selection of Ethernet switches. where Nintendostores should have been 2 words Nintendo and stores. Could someone help me solve this bug?

[xijiz]

Thank you for your attention to this work!

The code of this work is designed to handle Chinese datasets. There are some places that need to be modified to adapt datasets in other languages, such as token splitting and entity replacement. The most different handling logic is that Chinese text is split at the character level, while English text is split at the word or sub-word level. The rest logic is as same as the common NER pipeline. You only need to modify the above places.

Currently, I am sorry that I have no time to add new features and do tests. Hence, you can try to improve the logic for adapting to new languages according to the difference that I have mentioned above.