Lucene/Solr Revolution 2017 has ended
View analytic
Thursday, September 14 • 1:30pm - 2:10pm
Integrating Clickstream Data in Solr for Ranking and Dynamic Facet Optimization
Ranking with clickstream data: Clickstream data provides implicit guest feedback on search results. In this talk, I will cover extracting clickstream events to derive ranking score for a given search term and item based on previous click history. We will discuss:

Offline data computation and indexing:
- Defining clickstream events: clicks, conversion
- Spark implementation for clickstream batch processing
- Reverse indexing search term and rank score to SOLR

Query time ranking:
- Apply click scoring using SOLR function and boost queries
- Merge click score with default lucene tf/idf similarity score

Facet Optimization:
Faceted search increases conversion. Showing the right facets is key to improved user engagement and a better search experience. In this talk, I will discuss how click stream data can be used to derive function to reorder facets to optimize user engagement. Also, I will go over learning model for discovering the right facets for a query and filtering irrelevant facets. The model incrementally learns from previous iterations to dynamically adjust the ranking of facets.

avatar for Ilayaraja Prabakaran

Ilayaraja Prabakaran

Lead Engineer, Target
Ilay is Lead Engineer for Search relevance and ranking at Target. His prior experience included indian web and e-commerce startups and Yahoo!. He has been SOLR user for several years and a big data enthusiast. Ilay has master in computer science from IIIT, Hyderabad where he specialized... Read More →

Thursday September 14, 2017 1:30pm - 2:10pm
South Seas C