Evri intended to be a semantic map of the web based on content analysis of millions of documents, images and video published online every day. The goal of the product was to reduce the amount of time to find related content online based on what you were reading, and allow you to discover content and connections to other people, products, and things while viewing the most current and relevant information.
As the Senior Interaction Designer, I worked with a small team to migrate an existing query-based search and natural language parsing engine into a new interface for the Evri web app and content recommendation widgets for our partners' web pages. Our navigation model was displayed and controlled by an interactive hub and spoke diagram of the content showing all the main entities as balls, with verbs as connection lines between - clicking on them pivoted the data to relevant content or drilled deeper into related content. The web app was comprised of individual pages for people, products and things. Each page was dynamically updated to show a timely description, current content including news, blogs, images, video, products, and a detailed graph that showed the top relationships based on natural language parsing of current information using the sentence structure of noun + verb + direct objects as entities.
The content recommendation widgets were to be placed on any web page and show the top people, products and things found in an article or an entire web page, including links to the portal site, related content, and a brief description of the selected item. A large percentage of my work was focused on ideation to prove the technology and find the best solution for these widgets; which were powerful applications that existed in a small amount of page real estate, implemented in Air. This was possible by collaborating with Jose Hernando, an Action Script expert, to create Flash prototypes that could be realized, tested and refined iteratively. Since this product was all about data, the ability to demonstrate the experience using a working model was key.
Shown here are the initial concepts using an article about an HBO series about surfers living in Hawaii - a diagram of the story shows the main entities, the widget from the article opens related content that again dives deeper into detail when launching the web app. The final product shown here showcases Radiohead in the web app format, and content widgets injected into the pages of Hype Machine and Slate show related content for the reader's perusal intended to keep them digging into content, not leaving the site.
Medium: OmniGraffle, Photoshop
Role: Interaction Design, Visual Design