BM25 is So Yesterday: Modern Techniques for Better Search Relevance in Solr - Grant Ingersoll

BM25 is So Yesterday: Modern Techniques for Better Search Relevance in Solr - Grant Ingersoll, Lucidworks

Modern relevance in search engines has come a long way since the early days of information retrieval, when the likes of TF-IDF and BM25 scoring models first came on the scene. And, while those core models are still good for a first pass retrieval, more and more search engines are employing machine learning, natural language processing and sophisticated re-ranking techniques to fine tune relevance. This presentation will provide a review of current best practices in relevance tuning, including what to measure and how to measure it. CTO of Lucidworks, Grant Ingersoll will then give details on how to use techniques like learning to rank and query intent classification to improve results, with examples in Apache Solr. We’ll finish with a sneak peak into using deep learning and word2vec in a search context.

About Grant Ingersoll
Grant Ingersoll is the CTO and co-founder of Lucidworks, co-author of Taming Text, co-founder of Apache Mahout and a long-standing committer on the Apache Lucene and Solr open source projects. Grant’s experience includes engineering a variety of search, question answering, and natural language processing applications for a variety of domains and languages.