Apache Solr, Elasticsearch and Vespa are three of the most popular general purpose search engines that are used in search, recommendations and analytics based applications. Machine learning is a critical aid to improving relevance beyond the native ranking algorithms (e.g. TF-IDF, BM25 etc.) by leveraging editorial judgements and user behaviour and factoring them into ranking of results. This is known as "learning to rank" (LTR) or "machine learned ranking" (MLR) etc.

In this talk, the speaker presents a comparison of machine learning frameworks across all of these search engines [1]. The comparison is followed by a quick demonstration on how to use all of these frameworks. This talk is aimed at those who are looking to decide upon which search engine to use in their applications based on the machine learned ranking capabilities available for it, and the ease of using such features.

[1] 1. LTR based re-ranking module in Solr, https://lucene.apache.org/solr/guide/7_2/learning-to-rank.html       2. Elasticsearch's LTR plugin, https://github.com/o19s/elasticsearch-learning-to-rank       3. Vespa's MLR functionality, http://docs.vespa.ai/documentation/ranking.html