分享一波LLM辅助论文筛选的骚操作

下面是prompt:

Now you are going to help me as a paper analysis assistant, I will provide you with a csv composed of columns 'title', 'year', 'tldr', where the information of tldr is Auto-generated short summary of the paper from the SciTLDR model, I need you to answer for each line, that is, each paper.

the form like [id],[problem1answer],[problem2answer]

the [id] mean that row count

[problem1answer] mean you read tldr and the result of answer 'What problem is the paper trying to solve? '

[problem2answer] mean you read tldr and the result of answer 'What is the key to the solution mentioned in the paper? '

And at the end, make a statistical general summary of the problems and key technologies solved by these papers.

You must finish this work for all row in this csv file, the row count is [row count]! never miss any row!

Please use well formatted markdown output

----

有钱的也可以用gpt4的code interpreter