[good first issue] Create evaluation topic benchmark dataset

[federicodeponte]

Create a reproducible benchmark dataset of 20 diverse research topics for evaluation.

What to do:

• Create data/eval_topics.json with 20 topics
• Cover diverse domains: CS, medicine, economics, social science, physics, biology
• Include expected canonical papers for each topic
• Document evaluation criteria in EVALUATION.md

Skills needed: Research background, JSON editing

Related: EVALUATION.md

[keshavsharma0614-blip]

Hi @federicodeponte,

I'd like to work on this issue. I can create the benchmark dataset with diverse research topics, include canonical papers for each topic, and document the evaluation criteria in EVALUATION.md.

Thanks!