Source of data

[CAROLZXYZXY]

Hi,

Thanks for the code. I've got a question. Where does the files data/WebQSP.RE.train(test).with_boundary.withpool.dlnlp.txt comes from. More specifically, what are the pre-preprocess step from the natural language questions. I noticed that the negative relations are only a part of the whole relations, then how to select them?

Thank you very much!

[LayneIns]

@CAROLZXYZXY It is the original form of the data, which means that when releasing, the data has been processed into such form for the convenience of QA tasks, so I don't need further cleaning works. If you want more information about this dataset, you can search "web question dataset" on google or github, there are actually some papers using this dataset, including the one releasing using it.

Best