0
$\begingroup$

I know there are significant development about sentiment analysis in text e.g. "x is awful", "x is bad" or "y is fantastic". But are there similar approaches to compare hypotheses?

e.g. given a hypothesis say "gene A causes disease B", any approaches that examine a corpus of text generate evidence to validate or invalidate that hypotheses?

Are these approaches that work for "analysis of competing hypotheses"?

Edit: meant to say by analyzing text corpora (a la hypothesis testing in clinical (numerical) data)

$\endgroup$
  • 1
    $\begingroup$ Hypothesis testing (en.wikipedia.org/wiki/Statistical_hypothesis_testing) is a rather large part of statistics and has been an active area of research for over 100 years; do you mean to limit the domain of your question solely to ML - i.e., not statistical - approaches? $\endgroup$ – jbowman Nov 10 '18 at 17:00
  • $\begingroup$ I meant to say,hypotheses testing using text corpora. I am aware of hypotheses testing approach using clinical data etc. But can ML approaches be leveraged to learn from text and extrapolate? $\endgroup$ – Santino Nov 10 '18 at 19:28
  • $\begingroup$ I don't see how you could validate or invalidate a hypothesis about "gene A causes disease B" by looking at a corpus of text. I also don't understand why you are limiting yourself to machine learning techniques for hypothesis testing. There is a very large field of work that's been done on textual analysis, dating back (at least) to a famous work of determining the authorship of the various Federalist papers: books.google.com/books/about/…. What, specifically, are you trying to test? $\endgroup$ – jbowman Nov 10 '18 at 22:10
  • $\begingroup$ Hmm not sure what "limiting to ML" is, but I'd consider text analysis (NLP?) an ML technique. I was thinking approaches such as word embeddings, Bayesian approaches etc. might be helpful in discerning causal relationships. Curious about active research in this space. $\endgroup$ – Santino Nov 10 '18 at 22:45

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Browse other questions tagged or ask your own question.