Smartphone technology is becoming increasingly sophisticated and ‘now-casting’ software routinely uses activity history to make suggestions about what a user might like to do ‘now’. At the University of Melbourne’s School of Computing and Information Systems, Professor Rui Zhang is leading a project to further this predictive capacity.
He is working with Microsoft Research to develop software that predicts a person’s destination, using the real-time path of travel as tracked by the GPS on a user’s smartphone. Underpinning the predictions is not just one user’s journey, but the history of millions of trips made by others in the same city.
Professor Zhang says the software can potentially create new service and commercial opportunities. It will allow recommendations for goods and services to be made based on where the software predicts a person is going. This might include a restaurant nearby a concert venue, for example, or tickets for a similar performance.
The algorithm he has developed to do this advances technologies for location-based service recommendations. Currently, services are recommended based on the location of the customer at the time, but customers in transit are unlikely to be able to act on these. Professor Zhang’s technology may predict the final destination, enabling recommendations more relevant to the customer’s intention, which are more likely to be acted on.
The technology behind the predictions was developed based on big data models and tested using millions of taxi trips recorded in Beijing, China. In other cities, data from sources including individual smartphones, buses, taxis and traffic records could be combined.
With sufficient supporting data, Professor Zhang says by the time a person is halfway to their final destination, the software can predict the destination as one of its ‘top five’ selections with more than 60 per cent accuracy.
Telephone: +61 3 8344 1345