Minor issues

[danyaljj]

Hello there,

Nitpicky issues:

• I think the "readable" files are not proper "tsv"s. The question-answer space is not a tab (and sometimes its just a single space) which makes it hard to read (with a program or a tsv processor).
• There are repeated questions; for example "Who did she replace?" in the test file: https://raw.githubusercontent.com/uwnlp/qamr/master/data/readable/test.tsv

[julianmichael]

Thanks for the feedback—you're right, the .tsv extension is not technically right. But I did it that way so you can read them directly in a text editor or browser, e.g., in the link you gave above.

To read them in automatically, just chop off the first 50 characters for the question and trim whitespace. (50 characters was the length limit for the questions during annotation.)

Yes, some questions are repeated. In dev and test, we had multiple annotators write QA pairs for each set of target words, so duplicates will be especially common in those partitions. The example you give is one of those.

Hope that resolves your questions. I'll leave this issue open until I get around to updating the readme with clarifications.