Neural Search Improvements with Apache Solr 9.1: Approximate Nearest Neighbor and Pre-Filtering - Alessandro Benedetti, Sease Ltd.
Neural search is an Artificial Intelligence technique that allows a search engine to return documents that are relevant to the user information need without necessarily containing the input query terms; it learns the similarity of terms and sentences through deep neural networks and numerical vector representation. This solves the vocabulary mismatch problem, a limitation of many search engines where search results have to contain the keywords that the user typed in the search box in order to be retrieved. For example, the query “leopard” won’t retrieve documents containing only the terms “panthera pardus”. The focus of this talk is to describe the Apache Solr neural search module introducing the new improvements coming in Apache Solr 9.1: - indexing time and memory improvements to build the HNSW graphs data structures - pre-filtering capability with a dynamic switch between exact and approximate nearest neighbor search - Deep Learning integrations (e.g, BERT) Join us as we explore this new exciting Apache Solr feature and learn how you can use it to improve your search! It's going to be the first time this topic is presented in Asia and this talk is substantially different from its predecessors as it presents for the first time the new developments in Apache Solr 9.1.
Ещё видео!