17–19 October 2016: Conference & Tutorials
19–20 October 2016: Training
London, UK

Open sourcing learning to rank in Solr

Diego Ceccarelli (Bloomberg LP)
14:25–15:05 Monday, 17/10/2016
Open data & data science
Location: Windsor Suite Level: Intermediate
Average rating: ****.
(4.50, 2 ratings)

Prerequisite knowledge

  • A basic familiarity with Lucene, Solr, and Elasticsearch (useful but not required)

What you'll learn

  • Explore how learning to rank works
  • Learn how to use it in a production environment

Description

Learning to rank is a technique used by all the big search engines (Google, Bing, Yandex, etc.) that allows you to apply a machine-learning model in the construction of a ranking model for information retrieval systems. In fact, it is well known that sophisticated models can make more nuanced ranking decisions than a traditional ranking function. At the moment, there is not an open source solution available, but Bloomberg is working on implementing an open source plugin for Solr (an open source search engine), enabling others to easily build their own learning-to-rank systems and access the rich matching features readily available in Solr. Diego Ceccarelli presents learning-to-rank key concepts and explains how the Solr plugin works.

Photo of Diego Ceccarelli

Diego Ceccarelli

Bloomberg LP

Diego Ceccarelli is a software engineer at Bloomberg LP working on the News Search R&D team, where he focuses on improving search relevance for financial news. Before joining Bloomberg, Diego was a researcher in information retrieval at the National Council of Research in Italy, while completing his PhD in the same field at the University of Pisa.