Skip to main content

Getting Started with Clavis

Clavis is a recommendations engine that processes user behavior data through machine learning algorithms and returns content of interest to a user. Leveraging the latest in personalization technology, Clavis makes content recommendations and trending articles available to end users by analyzing click activity. By understanding what content is clicked on and by whom, we can determine articles that are relevant to the user’s interest to deliver engaging, relatable content.

Clavis allows you to surface items of interest to users, which in turn recirculates your content and increases engagement and time on site, making your content more discoverable. Readers can easily and quickly find new content they are interested in.

How Clavis works

Clavis is a set of APIs that power list-type blocks on pages. Primarily, you can use Clavis to create the following items: 

  • Popular/Trending - With Clavis, you can create a "most read" section on a homepage or section front. You can display the most popular content on your site over the last 24 hours (popular) or content that is gaining traction (trending) over the last 8 hours. If used on a section front, you can filter the Clavis responses to a specific section, for example, "Most Read Sports Articles". Clavis' popularity logic is based on number of pageview events on an article. 

  • Similarity - You can create a "Recommended Next" section on an article page. Clavis considers the article the user is currently viewing and serves up additional articles that others' have viewed. The events endpoint collects clickstream data from your site, and then the similarity endpoint considers that interactions dataset (not article metadata about topic or keywords) to determine similar articles to surface. 

Implementing Clavis

Clavis consists of an SDK to be installed on your website and configured to use the two available endpoints:

  • the events endpoint collects clickstream data from users

  • the SIMS Recommendations endpoint uses event data to generate recommendations based on popularity or similarity

To get started:

In this section: