Jonathan Strahl & Etsimo

Etsimo is a search engine that helps users explore results beyond their initial search. 

Type in search terms, and you will see your terms as part of a larger search space in a radar-like view. This search space consists of keywords that are close to your original search terms. 

You can bring interesting keywords closer to the center of the radar to discover results relevant to those keywords and help the system learn what you are looking for.

What’s under the hood?

The system uses machine learning to model user intent. A probabilistic model returns relevance predictions with uncertainty estimates for keywords, based on the user feedback. The prediction and uncertainty estimate are used to score a keyword’s potential relevance. The potentially most relevant keywords represent the user’s search intent. This is combined with information retrieval by using weighted query expansion to collect new sets of documents based on user feedback.