The problem is defined in the context of crowdsourced taxicabs searching for customers to pick up.
• A road network (map);
• A training dataset (list of resource locations and timestamps);
• A test dataset (list of resource locations and timestamps);
• The number of agents (i.e., agent cardinality) and their initial locations. The agent cardinality may be 5000, 6000, 7000, 8000, 9000, or 10000.
Search path for each agent.
Minimizing (in order of priority) the average search time, average wait time, and expiration percentage. The average search time will be the primary evaluation metric. The average wait time and then the expiration percentage will be used to break ties.