search

yunx-z / COMBO

Merging Generated and Retrieved Knowledge for Open-Domain QA (EMNLP 2023)
22 stars 1 forks source link

Question: About perfect matching in Compatibility-guided Optimal Matching #1

Open hoangnv735 opened 10 months ago

hoangnv735 commented 10 months ago

Thank you for your interesting work. I recently read your paper and have a small question. I wonder why we need perfect matching (1-1) between 2 sets of context, other than allowing 1-many (or many-1) maching. For example: (L1, L2, L3); (R1, R2, R3) can be mapped like (L1 -- R1), (L1 -- R2), (L2 -- R3) and (L3 -- R3)

yunx-z commented 8 months ago

Hi,

Perfect matching results in the least number of passage pairs that can cover all the input passages, which saves training and inference time. 1-many (or many-1) matching also makes sense to me, but we did not try it out.

Thanks, Yunxiang