(I Can Get) Satisfaction: Preference-Based Scheduling for Concert-Goers at Multi-venue Music Festivals

With more than 30 million attendees each year in the U.S. alone, music festivals are a fast-growing source of entertainment, visited by both fans and industry professionals. Popular music festivals are large-scale events, often spread across multiple venues and lasting several days. The largest festivals exceed 600 shows per day across dozens of venues. With many artists performing at overlapping times in distant locations, preparing a personal schedule for a festival-goer is a challenging task. In this work, we present an automated system for building a personal schedule that maximizes the utility of the shows attended based on the user preferences, while taking into account travel times and required breaks. Our system leverages data mining and machine learning techniques together with combinatorial optimization to provide optimal personal schedules in real time, over a web interface. We evaluate MaxSAT and Constraint Programming formulations on a large set of real festival timetables, demonstrating that MaxSAT can provide optimal solutions in about 10 seconds on average, making it a suitable technology for such an online application.