3 ms·

This is great! I just tried some queries and the results were pretty decent, in terms of semantics. But, just thinking of it as a user, if this were to be part

This is great! I just tried some queries and the results were pretty decent, in terms of semantics. But, just thinking of it as a user, if this were to be part of my daily workflow (instead of say something like Google Scholar), I would like: 1. The option to somehow see _how_ the paper was reviewed and/or cited, if at all. There are things like OpenReview, see example [1] 2. The ability to "tell me a story to get up to speed" about a collection of papers. Generative models could help here -- but essentially, I want this thing to be able to write a paragraph for what one might find in the literature review / related work of a paper, with citations. :-) All the best! [1] https://openreview.net/forum?id=jhKbnNhwhc https://openreview.net/forum?id=jhKbnNhwhc

Quizzical4230 2y ago
1. I was not aware of OpenReview. I love the transparency and would definitely look into integrating it. 2. This is good feedback, making models write the Introduction section! I was planning to keep this search engine a little more traditional, however if the results are good, then it should be the way forward. Thank you, Happy Holidays! :D
- odyssey7 2y ago
  I have to second the idea, having hacked together something similar myself, to help me complete a literature review——a literature review that I wasn’t planning to publish. Simply generating summaries or pulling key quotes, paper by paper, wasn’t sufficient to be able to understand the topic in the way I wanted to for writing the literature review. In the end, the system would process a collection of hundreds of PDFs that might be related, generate summaries of what they mentioned about the topic in question, and, importantly, was also prompted to note anything about how the insights built upon or were related to insights from previous research, and the motivations behind developing that insight / the challenge it was attempting to solve and whether it was successful. This worked well enough to reduce what might have been weeks worth of work to just a few hours. Genuinely, I believe that research in the near future could look a lot different from what it looks like today.

Quizzical4230 2y ago

1. I was not aware of OpenReview. I love the transparency and would definitely look into integrating it. 2. This is good feedback, making models write the Introduction section! I was planning to keep this search engine a little more traditional, however if the results are good, then it should be the way forward. Thank you, Happy Holidays! :D

odyssey7 2y ago
I have to second the idea, having hacked together something similar myself, to help me complete a literature review——a literature review that I wasn’t planning to publish. Simply generating summaries or pulling key quotes, paper by paper, wasn’t sufficient to be able to understand the topic in the way I wanted to for writing the literature review. In the end, the system would process a collection of hundreds of PDFs that might be related, generate summaries of what they mentioned about the topic in question, and, importantly, was also prompted to note anything about how the insights built upon or were related to insights from previous research, and the motivations behind developing that insight / the challenge it was attempting to solve and whether it was successful. This worked well enough to reduce what might have been weeks worth of work to just a few hours. Genuinely, I believe that research in the near future could look a lot different from what it looks like today.