Several issues I find in the entity/number extraction functions

[wwt17]

Here are several issues I find in the rule-based entity/number extraction functions in data_utils.py. I'm still working on the dataset and will probably update more once I observe them. These issues are generally sorted by the order of their impact in my impression.

1. Implicit number words, such as (a|an|a pair of|a trio of) followed by nouns like (rebound|turnover|assist|block|steal|board|three-pointer|three pointer|free-throw|free throw|dime), are ignored. This probably results in the largest number of omissions.
2. Besides aliases like {'Los Angeles', 'LA'}, there are other aliases, like {'76ers', 'Sixers'} and {'Mavericks', 'Mavs'}.
3. The NLTK tokenizer did not separate suffixes containing ’ (looks similar to ', but with Unicode 2019), which makes some entities cannot be identified.
4. A player name with initials, such as J.J., may be present in another form, like JJ.
5. A player name ending with Jr., may also end with Jr, , Jr. or , Jr.
6. A player name with a hyphen, in the form A B-C, may be present in another form A-B C.
7. Other minor issues, such as Oklahoma city.

You can search for these cases in the dataset.

Hopefully, my findings could help people in the future. Actually, I found these issues when I was working on Lin et al, 2020 some years ago. These issues make the dataset quite noisy.

I do have my own version of data_utils.py which fixes these issues to some extent.

Thanks for your pioneering work!