We are excited to announce our next Berlin Search Technology Meetup.
With just a few days to go until the Haystack EU conference, our talks will be all about search relevance again this time.
The meetup will be hosted and sponsored by SoundCloud – a big thank you!
18:30 - Doors open, Snacks and Beers
19:00 - Introduction
19:05 - Conversion Models: A Systematic Method of Building Learning to Rank Training Data (Doug Turnbull, Open Source Connections)
19.50 - 20.00 break
20:00 - Query relaxation - a rewriting technique between search and recommendations (René Kriegler, Freelance Search Consultant)
20:45 - 21:30 Networking, Snacks and Beers
Conversion Models: A Systematic Method of Building Learning to Rank Training Data (Doug Turnbull)
When using user signals to improve relevance, what should you use? Clicks are more frequent, but really only correspond to a search result looking attractive. A conversion is a powerful signal of true relevance but occurs less frequently. Can we combine shallow "this looks interesting" click events along with strong, but rare conversion signals in a robust fashion to generate learning to rank training data? In this talk, we introduce click models, an industry-proven way of measuring search result attractiveness from clicks, and propose a systematic way of incorporating conversion data into click models. Whether your industry is conversion heavy (like e-commerce), or lacking in any clear conversion signal (like publishing) you'll take away from this talk a system for turning any search analytics into robust judgments and training data. Because, after all, there is no AI-based Search without good training data!
Author of ‘Relevant Search’. CTO of OpenSource Connections (http://opensourceconnections.com) . As a consultant, Doug's mission is to help teams understand the intersection of AI-driven search technology, organizational challenges, and user needs. Towards this mission, Doug enjoys training teams on Solr Relevance and Learning to Rank, helping teams organize to be "relevance-centered enterprise", and guiding teams to a cohesive relevance strategy.
Doug loves learning, writing about search on OpenSource Connection's blog, and working on tools that enable Solr relevancy delivery, including Quepid, Splainer and Elyzer.
Query relaxation - a rewriting technique between search and recommendations (René Kriegler)
In search quality optimisation, various techniques are used to improve recall, especially in order to avoid empty search result sets. In most of the solutions, such as spelling correction and query expansion, the search query is modified while the original query intent is normally preserved.
In my talk, I shall describe my experiments with different approaches to query relaxation. Query relaxation is a query rewriting technique which removes one or more terms from multi-term queries that would otherwise lead to zero results. In many cases the removal of a query term entails a change of the query intent, making it difficult to judge the quality of the rewritten query and hence to decide which query term should be removed.
I argue that query relaxation might be best understood if it is seen as a technique on the border between search and recommendations. My focus is on a solution in the context of e-commerce search which is based on using Word2Vec embeddings.
René has been working as a freelance search consultant for clients in Germany and abroad for more than ten years. Although he is interested in all aspects of search and NLP, key areas include search relevance consulting and e-commerce search. His technological focus is on Solr/Lucene & Elasticsearch. René is the main organiser of MICES (Mix-Camp E-Commerce Search, http://mices.co). He maintains the Querqy open source query rewriting library (https://github.com/renekrie/querqy)