Loading…
This event has ended. Create your own event → Check it out
This event has ended. Create your own
View analytic
Thursday, September 14 • 3:10pm - 3:50pm
User Behavior Driven Intelligent Query Re-ranking and Suggestion

With the advancements in machine learning, Solr is becoming more intelligent. In this talk, two user behavior driven features are discussed. Search result straight coming out of index may not be optimal. When those results are re-ordered by price, review score, arrival time etc, results are further degraded. An L2R (learn to rank) method is developed to learn from user behavior and the learnt model re-ranks search result in an optimal way. Both text features and image features are converted into vector forms and combined to train ensemble models. The method can be applied to queries with or without user behavior and can also be used to intelligently sort search results by criteria other than search relevancy. The second user behavior driven feature is type-ahead, also known as auto-complete, autofill, auto-suggest. Traditional methods are based on statistics from query log and product catalog. The new method leverages users’ selection of the predicted queries along with the location and time of the interaction to provide more relevant and accurate type ahead suggestions. Both features showcase how machine learning meets solr to render a more powerful overall search experience in an e-Commerce website.  


Speakers
avatar for Rajdeep Mondal

Rajdeep Mondal

Senior Software Developer, The Home Depot
Rajdeep is a Senior Software Engineer at The Home Depot where he researches and implements new statistical and machine learning algorithms in Search Relevance and Personalization domains. Rajdeep is keenly interested in application of deep learning techniques in text and images t... Read More →
avatar for Rongkai (Alfred) Zhao

Rongkai (Alfred) Zhao

Software Engineering Manager & Architect, The Home Depot
Rongkai Zhao (Alfred) is a Software Engineering Manager and Architect at The Home Depot where he oversees the research and development of system components in search, personalization, and call center intelligence. He has worked on e-commerce search engine since 2010 and has a wid... Read More →


Thursday September 14, 2017 3:10pm - 3:50pm
South Seas A