Name variation search (Pro Omega, ProOmega, Pro-Omega)

Elastic Stack Elasticsearch
1

I use elastic to search for products in an index and have an issue where I have a product called "ProOmega" and the issue is when "Pro Omega" is searched I get no results. Ideally if I search for "ProOmega", "Pro Omega", "Pro-Omega" all these queries would give me back the "ProOmega" product.

I have a pretty basic whitespace analyzer on my name field.

I've tried playing around with some fuzziness but couldn't get a good solution. I think the issue comes from the fact that "Pro Omega" is broken into two tokens but the actual product "ProOmega" is a single token.

I think one solution could be to add a character filter to convert spaces to "nothing" but I fear it might affect other search results.

Is there any better solution to this problem?

2

You can define synonyms for these but that can get tedious to maintain as the list grows long.
An alternative is to index using "spaceless shingles" - pairs of words without a space between.

This works for regular match queries but has an issues with match_phrase query types.
See this issue for example code and details of problems with phrase searches.

3

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.