Spurious Recommendation Identification and Serendipitious Discovery

This example came from user study March, 2022 using Vis dataset. It will demonstrate how map view (global explanation) helps identify spurious recommendation and how scatterplot encourages serendipitious discovery.

After an Abstract Search with her abstract, our participant quickly identified a spurious recommendation from the map view - one that is far from every other recommendation. Scanning through the titles confirms that this recommendation is spurious. Participant was also able to justify how the recommendation were picked up due to the common terms “color”, “map” used.

Although it is a spurious recommendation, common term usage between this recommendation and input abstract made participant curious about its neighbors in the scatterplot view. Hovering over its neighbors in the scatterplot view helped our participant discover a related work serendipitiously.