So I want to start out by saying that ML (as well as NLP more specifically) are things that are really quite new to me. I have read a lot about different approaches over the years and I feel I have at least a somewhat passable theoretical understanding of them, which makes me fairly confident that what I want to achieve is possible.
Essentially, I want to develop what I I feel would be most accurately labeled a "recommendation engine" of sorts that will parse through text looking for phrases such as eg "looking for an hvac provider" or other statements of that sort that indicate that a user is looking for a specific service within a certain domain, so that I can potentially facilitate recommendations to clients of some given service in order to try and drive more "organic" (sic) discovery of said client brands and services they offer.
I've been digging around through a lot of NLP libraries and it seems that the main way that I may have to do this is annotate a corpus of text of my own with different sorts of forms that such information requests may take, and while I will happily do that if I have to, I suppose what I really would like to know is:
- Are there any libraries out there, ideally for nodejs, that would assist to facilitate this business need, and,
- If not, are they any corpora out there that are annotated with the sort of information that I might need to identify such requests for recommendations in text?
I am going to continue to do some research and will update here if I end up finding anything, but as of yet I am not having much luck! Thanks in advance, I am hoping that I have more luck posting this here as I was informed that this community is a more appropriate one for this question after posting on SO. To sum everything up: I am more looking for a library or a corpus to augment a library rather than the theory and praxis behind such a library (though I certainly am interested in those as well but since I am seeking recommendations that'll be an entirely different bridge to burn some other day).