Recently I flagged a question as more suitable for Data Science Stack Exchange and should be migrated there. However it was declined. To me this is clearly a Data Science question (no ambiguity). Did I make a mistake in my judgement or are we entertaining Data Science questions also?


I personally feel like most Data Science questions would be just fine on AI too, Data Science and AI simply are very closely related. The only argument against having Data Science questions on AI.se that I'm aware of basically boils down to trying to avoid as much overlap as possible.

From my point of view, that kind of overlap really isn't too much of a problem. The topic in the question (entity recognition from text) is certainly a topic that could be described as being a part of "Artificial Intelligence", and it would be just as correct as saying it's a "Data Science" topic. So I personally really wouldn't mind if it's allowed on either site, I can see it fitting in either just fine. I understand that StackExchange as a complete network might find it more problematic if there's too much overlap, and if that's the case their opinion is probably more important than mine, I just don't experience it as problematic personally.

The only sentence in the question you linked to that is maybe a bit questionable in my opinion is the following:

I have tried Spacy and NLTK for entity extraction but that doesn't suffice above requirements.

That sentence is describing specific tools/frameworks, and implies the question-asker might be looking for more names of similarly specific tools/frameworks. I do feel those kinds of questions would be a better fit for Data Science.

But the same question, especially if you ignore that one sentence, can easily be interpreted as being of a more conceptual nature, asking more generally about techniques/algorithms that would be applicable. It looks to me like both of the current answers also interpret the question in that way. Such "conceptual" questions would be just fine here in my opinion.

  • 1
    $\begingroup$ Yep, the last paragraph is why I didn't migrate the question. Well said! $\endgroup$
    – Ben N
    Aug 19 '18 at 17:46
  • 2
    $\begingroup$ @BenN my thoughts were that the question would have been better served by a DS.SE...the user probably would have got better answers there $\endgroup$
    – user9947
    Aug 20 '18 at 2:19

To second Dennis Soemers' answer:

Much, but not all, of Data Science relies on AI tools. When the question relates to AI, we should answer it.

Some examples of Data Science topics that are not about AI, and which we should migrate are:

  • Questions about scraping data from the web.
  • Questions about hypothesis testing or other conventional techniques from statistics (unless about the evaluation of ML methods).
  • Questions about programming languages or toolkits within those languages, that focus on syntax or programming rather than AI/ML algorithms.

You must log in to answer this question.