Enhance String Intersection Function for More Accurate Answer Evaluation in qa_evaluate.py

[yarikama]

Thank you for your excellent work!

In the current implementation of qa_evaluate, the has_intersection function used for comparing predicted answers with gold standard answers splits strings on whitespace and checks for exact matches between words.

This can lead to false negatives when comparing strings with minor differences in punctuation.

The current method may incorrectly evaluate answers as different when they are semantically the same but differ in punctuation. For example, "Yes," and "Yes" are currently treated as different answers.

I suggest implementing an improved has_intersection function with removing punctuation.

[yixuantt]

Thanks for your interest in this work!
Yes, you raise a good question. I will optimize the evaluation code when I have time. I would also greatly appreciate any suggestions regarding the semantic evaluation of the answer, excluding LLM as a judge.