Did you ever come across terms like BM25 similarity model, KL divergence model, the language model, and the translation model in ElasticSearch or Solr documentation? Did you wonder what these models do and whether they can improve your product?
In this talk we revisit the classical information retrieval approaches and explain the thinking and intuition behind the models so that you can decide whether they fit your use case. We also show some useful extensions handling common cases such as matching the query across multiple fields, handling spelling errors, synonyms and personalized context.

Stefan works as a Senior Software Engineer at ResearchGate, where he focuses on developing their recommendation system. Previously, Stefan worked as a Software Engineer for relevance in Microsoft. He holds a PhD in Information Retrieval from Northeastern University, Boston and a MSc degree in Natural Language Processing from RWTH, Aachen. Stefan loves to discuss topics related to algorithms, search engine implementation, functional programming languages and machine learning.