Advanced Document Similarity with Apache Lucene

Being your core domain involving real world entities ( such as hotels, restaurant, cars …) or text documents, searching for similar entities, given one in input, is a very common use case for most o …

Being your core domain involving real world entities ( such as hotels, restaurant, cars …) or text documents, searching for similar entities, given one in input, is a very common use case for most of the systems that involve information retrieval. This presentation will start describing how much this problem is present across a variety of different scenarios and how you can use the More Like This feature in the Apache Lucene library to solve it. Building on the introduction the focus will be on how the More Like This module internally works, all the components involved end to end, BM25 text similarity metric and how this has been included through a cospicuos refactor and testing process. The presentation will include real world usage examples and future developments such as improved query building through positional phrase queries and term relevancy scoring pluggability.