Optimizing search relevance using Querqy for Elasticsearch

United States and Canada Virtual
Thu, Dec 3, 2020, 2:00 PM (EST)

36 RSVP'ed

About this event

Increasingly, applications use open source technologies such as Elasticsearch and Solr to implement search. However, when migrating an application from a commercial search software to an open source solution, there often is a feature gap, especially in the search management capabilities. While commercial solutions normally provide convenient ways to configure search relevance and searchandising rules, in open source solutions these features are either completely missing or, under certain conditions, have scoring and recall problems.

In my talk, I will give an introduction to Querqy, an open source library for search query preprocessing that solves many of these problems. Querqy was originally started six years ago for Solr. It has been available for Elasticsearch since December 2019. I shall also introduce Querqy as a general query rewriting framework which allows to plug in further rewriters - for example, for handling decompounding or implementing custom query rewriters. I will also explain how this library simplifies query building for Elasticsearch in the context of search relevance optimisation. Querqy is now being used by top-tier e-commerce sites and recently a web UI was developed to facilitate search management.