5

Although there are plenty of search engines which allow you to identify citation links - paper Y cites paper X - are there any which identify what statements and summaries are attributed to paper X? You type in the DOI for paper X, and in addition to a list of which papers cite X, the list of particular sentences in each of these papers which refer to X are also received, e.g.

  • "The world is round" [paper 1]
  • "The curvature of the world has long been debated" [paper 2]
  • "Both Flat Earth alternate theories have been proposed as to the shape of the world" [paper 3]
  • "It is widely assumed that the world is round" [paper 4]
  • etc etc.

If a citation search tool like this exists, can it also be used to link from declarative statements, to references cited in support in that statement?

For example, Statement == "The world is round" - 215 hits, of which:

  • 139 unsupported by citations
  • 24 supported by reference X
  • 12 supported by reference Y

and so on.

Do any academic search engines with these capabilities already exist?

Improve this question
1
  • If I didn't correct the title properly, please do a rollback.
    – aparente001
    Oct 14, 2016 at 17:39

1 Answer 1

Reset to default
1

I pretty sure this does not exist. It would require a level of sophisticated text parsing that has not been developed by computers. After all, not all statements that the earth is round are going to phrased the same way, and for such a tool to be useful, it would be necessary to identify statements with the same semantic meanings. This is not possible with the current language processing techniques; it would be huge advance if a computer could reliably parse and identify logically equivalent statements. So you are not going to find what you are looking for.

Improve this answer
4
  • Thanks. I don't think this would be a very tough text parsing challenge though: look for a text string in a series of texts, then look for particular combinations of symbols (e.g. [x], (1), something superscripted, a non-standard word followed by a four digit number, perhaps with 'et al' in the middle), then look for something that corresponds with this series of documents at the end of the text block.
    – JonMinton
    Oct 15, 2016 at 10:05
  • The hard part is not picking out references, it is identifying the claim that the reference is being used to support (or whatever).
    – Buzz
    Oct 15, 2016 at 13:06
  • It doesn't need to be anything as complex as you described. All it has to do is to automatically index the sentence in which the reference was made, and maybe the sentences before and after as well. Computers can do that very easily; they don't need to understand any meaning--they only need to be able to identify the reference. It's up to the researcher to decide if that context is useful or not. That said, I agree with you that even something on this simpler level probably doesn't exist, not yet anyway.
    – Tripartio
    Oct 15, 2016 at 14:25
  • Thanks both for your comments. I'd have thought that, given we've got self-driving cars and Quiz-willing AIs, something like this should be with us already. The two main uses I could think are: 1) to identify whether something you've written has been interpreted in the way you intended, or whether it's become shorthand for stating something you didn't intend it to; 2) (more contentiously) to quickly identify a way to cite something, where in the field it's become an expectation that it should be cited, in such a way that it's unlikely to be contested.
    – JonMinton
    Oct 16, 2016 at 9:21

You must log in to answer this question.

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