Magitti Context-Aware Leisure Guide (2004-2006)

Many other PARC employees participated in this work, including Bo Begole, Ed H. Chi, Nicolas Ducheneaut, Ji Fang, Ellen Isaacs, Tracy King, Mark W. Newman, Kurt Partridge, Bob Price, Paul Rasmussen, Michael Roberts, Diane J. Schiano, and Alan Walendowski.

Think back to 2004, before the advent of the iPhone and Android smartphones. I led the innovation process, UX research and UX design of the Magitti leisure guide for Dai Nippon Printing.

Intensive fieldwork in Japan informed the concept and innovation brainstorms were its birthplace. The original concept addressed target users desire to be able to find interesting venues in the city, but it was a highly ambitious augmented reality solution around the concept of publishing city guide info to context. It's the sort of thing that would have worked well with Google Glass, but at the time the platforms weren't there to make something that advanced feasible.

So, in order to get something developed quickly, we consulted with target users about what their top priorities were and then simplified the concept to a minimal viable product that still addressed their needs, providing contextually sensitive leisure activity recommendations to its user for things to do in the local area.

The screenshot shows an interface optimized for thumb-based interaction as fieldwork showed that Japanese target users needed a one-handed interaction UI (this is how they use their smartphones today).

Magitti Screenshot

Magitti System Overview

The system was way ahead of its time, using both online and sensed context and nine different modeling approaches with a model weights learner to make its inferences. It predicted that the user would be engaged in one of five activity types, while at leisure; eat, shop, see, do or read.

Fieldwork from intercept interviews by students in various locations in Tokyo was used to derive population priors on the probability of each activity during time of day and day of week. But as the user used the device it would begin to develop a personal profile and personalize its recommendations.

The figure on the left gives an overall schematic of the system.

CHI publication available