6

Having spent quite a lot of time going through the literature on my research topic, I've been disappointed by the search features offered by the dominant academic databases and search engine (Web of Science, Scopus, Google Scholar etc...).

The search feature remains pretty basic and allows us to search by common theme.

We cannot search for ideas shared between papers, how one paper contradicts or supports another, or create nodes graphs by citations etc... I've thus relied on doing Knowledge Extraction by myself, but I'm sure someone has encountered these issues before.

Has anyone found a better tool or approach ?

4
  • 3
    What advice did your librarian give when you consulted them?
    – shoover
    Aug 11 at 17:27
  • Nothing helpful, the tasks I'd like to do (paragraph level topic modeling etc...) were not supported by any search engine they knew of
    – Matthieu
    Aug 11 at 17:38
  • 3
    This is impossible with current technology. You're essentially expecting the search engine to understand your area of research! Aug 11 at 18:56
  • 3
    I've built this myself on a smaller corpus - topic modeling is standard NLP practice. Scite.ai for instance (mentioned below) makes use of it, albeit at a restricted scale. I'll try scaling it up and share more if the results are good
    – Matthieu
    Aug 11 at 20:29

2 Answers 2

9

... how one paper contradicts or supports another

The tools to go for this feature are:

... or create nodes graphs by citations

You may be interested in:

1

I ask new researchers to generate a conceptual model by identifying 3-7 key terms, authors, and journals before implementing any review of literature through a search engine. This prepares them for either a narrative or systematic review of literature. This also allows them to quickly begin the process of understanding how their future work "fits" within the larger corpus of literature.

You must log in to answer this question.

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