Dec 14th, 2021: [en] Leverage your data to build better search faster with curations powered by adaptive relevance

The building blocks of building great search experiences

Building a powerful search experience is a necessity for any applications that are built on top of data. Ensuring that the best results are delivered to users based on their queries can make or break the success of the app. When building search solutions, having an understanding of what results users are hoping for brings an immediate satisfaction to them. This is where some of the enhancements in Enterprise Search, specifically around adaptive relevance, can ensure your search solutions bring great value to users.

The current curations experience

App Search operators are well aware of what benefit curations can provide. Curations allow operators to customize search results for specific queries. Operators can “promote” documents to ensure they are bubbled to the top of queries that match. For example, below is a curation that ensures that when a user searches for “mountains,” the two promoted documents that will be at the top of the results for that query are “Rocky Mountain” and “Glacier.”

This will lead to a search result like below

Curations are extremely powerful, but there is a challenge in determining the results to promote. Users want the best results based on what they expect to see, and if documents are promoted that they aren’t expecting, they might lose faith in the search experience they're using.

Introducing curations powered by adaptive relevance

In an effort to bring more automation to the curations process, we are excited to announce curations powered by adaptive relevance in release 7.16 (released December 7th 2021). Adaptive relevance is a new area of focus to bring automated suggestions to core areas of App Search, with curations being the first area to be released. The experience is driven by actual user interaction. For instance, when a user makes a query, and clicks on a particular result, adaptive relevance tracks this. It then builds a model over time with queries and their most-clicked results. If a trend is detected, adaptive relevance will make suggestions for specific documents to promote for those queries. After that, the experience is the same as creating a curation manually. Here is an example of the entire experience from within Kibana.

You can manage the entire adaptive relevance experience in Kibana or via the API. More information on doing that is located in the Curations docs. This is just the first step in bringing more automation to tasks in App Search, and we encourage folks to give it a try and provide feedback if they have any.

1 Like

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.