JavaScript algorithm to match text as closely as possible (but not identical)

embraer170 · 18-02-2015 11:12PM #1

Hi there,

Do you have any advice for a JavaScript algorithm, which searches through a database of strings of text to find multiple matches as closely related as possible to one particular string.

For example:
I ate a chicken

Database:
I think I ate bad chicken for my lunch
I had chicken salad for breakfast
Had I eaten breakfast, I would have felt better.

It would be a large database and will have to repeat hundreds of times for different searches.

Giblet · 18-02-2015 11:13PM

You would use an indexer of some sort for this, such as Solr or Elastic Search (both based on Lucene).

bpmurray · 20-02-2015 09:36AM

I've never hear of fuzzy matching written in JavaScript - that sounds rather ambitious. Instead, use something like Lucene as Giblit suggested - FuzzyQuery might be what you want.

colm_c · 20-02-2015 10:30AM

Depends on the overall solution, but I don't think a client side JS will be up to the job, here's a node module which might help:
https://github.com/mattyork/fuzzy

embraer170 · 20-02-2015 11:51PM

Thanks all for your help, I will be doing some digging based on these references.

JavaScript algorithm to match text as closely as possible (but not identical)

Comments