37

Half a year ago I stumbled across a noteworthy peer-reviewed article that was very relevant to my current research area. Being a noteworthy idiot, I didn't save it nor take a note (to be fair, I wasn't doing any relevant research at that point).

Fast forward six months, I really need to find this article again. I want to reference it and check out for more relevant information.

However, the problem is... it vanished.

It's been a week of non-stop trying, and I just can't find it. I've tried all possible combinations of keywords, google, scholar.google. I've checked every reference in all related articles that I could find. Not a trace. It even made me question my sanity: I'm starting to believe that I imagined or dreamed of it.

This isn't the first time I've run into such a problem. I've similarly lost at least two articles before. Same story: stumbled upon it at some point, found it interesting but didn't make a note (as it didn't seem important at that point) — only to never find it again.

Any advice on how I could find it again? I don't remember the title, the author, or the exact quotes from it. I only know what kind of point it made, and that I'll recognize it if I see it.

3
  • Comments have been moved to chat; please do not continue the discussion here. Before posting a comment below this one, please review the purposes of comments. Comments that do not request clarification or suggest improvements usually belong as an answer, on Academia Meta, or in Academia Chat. Comments continuing discussion may be removed.
    – Bryan Krause
    Oct 24, 2023 at 20:10
  • 4
    Your own Google search history might still have the link.
    – RonJohn
    Oct 25, 2023 at 14:01
  • You could try a different search engine, like Bing or Duckduckgo. These use different indexing rules.
    – Jim
    Oct 25, 2023 at 19:27

7 Answers 7

50

If you have noticed it, other researchers in your field are likely to have noticed it too. I'd ask other researchers in your field if your description of the paper reminds them of anything, and then possibly ask the question on social media, giving as many details as you can remember.

To avoid a repeat of this situation, I'd also strongly recommend using a reference manager with a dedicated folder where you'd collect all 'interesting but not read yet' papers.

7
  • 11
    +1, especially for the long-term suggestion in the second part. Unfortunately, that won't offer a surefire solution, either: I have been doing this for decades now (plus jotting down a thought or two on papers I read in my by-now-massive journal, which I can quickly CTRL-F through - but my literature DB now contains over 5,000 entries, and it starts getting hard to find "that one highly relevant paper I read sometime last year" in all that... Oct 24, 2023 at 9:00
  • 1
    To help navigating your reference database create a few keywords of your own that all begins with an exclamation mark, e.g. "!sunflower", "!statistics", "!whale", etc. Such keywords will never interfere with the given keywords. Usually I will add just one of those. You can create subsets for the literature cited in your works by adding another set, e.g., "$Paris 2018", "$Nature 2015", etc. Both methods are superior to splitting you database into several databases according to topic or task. Oct 24, 2023 at 10:54
  • 2
    @StephanKolassa a partial solution I've used for this, at least in Zotero, is to assign at least one tag (such as "Bayesian optimization") to the paper as soon as it is stored in the DB. I can then filter my collection of papers by clicking on tags. In theory, you should then be able to filter first by tags before attempting a search. Another possibility is to split the library into collections, where each collection has a unifying topic.
    – mhdadk
    Oct 24, 2023 at 14:13
  • 2
    Good tips. I use both tags and collections (/folders). Finding papers would be much less efficient without them, I can’t imagine having all my papers in one giant folder! A nice feature (at least in Zotero) is also that you can associate the same paper to different folders.
    – user126108
    Oct 24, 2023 at 14:35
  • 1
    I rather meant something like Word2Vec, NLTK or even ChatGPT not just on headings, but on the full text. Would be a great addition for any bibtex tool. ... just looked it up. it is called Doc2Vec
    – Summer-Sky
    Oct 24, 2023 at 21:42
43

Ask the librarian at your institution. They are absolute wizards at finding information, and are also usually very happy to help.

4
  • 27
    Research librarians are worth their weight in gold. And are often more than happy to help because so many people think they know how to perform searches better than a trained experienced librarian. The more difficult the search is the more excited they get. Please, think of the librarians!
    – Jon Custer
    Oct 25, 2023 at 13:38
  • But OP might not be affiliated with any institution with librarians that she can ask
    – user103496
    Oct 26, 2023 at 2:01
  • 2
    @user103496 that's fine, not all stackexchange answers have to apply to every single person.
    – Christian
    Oct 26, 2023 at 14:25
  • 1
    @user103496 depending on the place, it may just be a matter of entering a university library and asking for help. No need to be formally affiliated.
    – WoJ
    Oct 26, 2023 at 14:29
9

As an adaption of @leonos answer, try looking at the references in papers on a similar topic. They may reference this paper, or be cited by this paper.

Also can you remember anything distinctive about the paper's formatting, which may also help narrow down the journal/age? Details like was it a very long or very short article? did the text look a bit odd, perhaps being a scanned copy of an older work?

9

Have you tried describing the paper to ChatGPT (version 4)?

I haven't used it to search for papers specifically, but I continue to be amazed at its awareness of obscure books and its ability to identify them based on incomplete/inaccurate information.

3
  • 16
    In contrast, I continue to be amazed by ChatGPT's (or Bard or any other similarly hyped AI) mistakes, hallucinations, and capacity to waste my time with false information.
    – user103496
    Oct 25, 2023 at 3:45
  • 19
    @user103496 ChatGPT has a host of problems but using it as an intelligent search like this is playing to its strengths and avoiding its weaknesses. The only downside if it suggests the wrong paper is a loss of time, but I suspect our answer has already wasted a substantial amount of time looking for it. Oct 25, 2023 at 8:38
  • 2
    This is definitely worth a try. I even recently used it to identify a couple of obscure shows. I’ve never really experienced any hallucinations to a large degree in GPT 4. I think the key is to properly formulate your prompts. Oct 25, 2023 at 17:53
5

As humans we're not great at remembering keywords, much better at remembering concepts. And useful material can use different terminology in adjacent fields. Google of course tries to help, by including synonyms in search results, but they're overly broad, so instead of finding the needle, it merely increases the size of the haystack. So quote marks around the keywords narrow things down - but you may have misremembered, and you narrow down to the wrong results. A manually-generated set of synonyms helps. I prefer a normal Google search over Scholar in this sort of case.

When you do search, you're likely to find near misses. If these near misses are recent (try to remember how old your target paper was) look at their references. If your near misses are old, look at what cited them

I suspect this happen to all of us at some point. Quite often it's not interesting enough at the time to file in the "interesting, for later" folder/database, then happens to be of use to another project in the future.

5

This might be too simple, but if you remember around what time you accessed the paper and if you were not using private browsing mode or disabled your history, you should be able to simply search your browser history around the relevant date or even manually go through it.

I've done it a few times and it has found me information I couldn't reproduce again by a search

1
  • 1
    Sometimes the downloads folder can be a helpful place to search, too.
    – jpa
    Oct 26, 2023 at 16:34
3

Post a question on the relevant Stackexchange site, listing all the relevant details and clues you can remember.

(If there's no particularly suitable specialized Stackexchange site, I believe you can post here on Academia too.)

Try also other forums (e.g. the relevant Reddit forum).

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .