Scalable Public Transportation Queries on the Database

Recent scientific literature focuses on answering Earliest Arrival (EA), Latest Departure (LD) and Shortest Duration (SD) queries in (schedule-based) public transportation networks. Unfortunately, most of the existing solutions operate in main memory, making the proposed methods hard to scale for larger instances and difficult to integrate in a multi-user environment. This work proposes PTLDB (Public Transporation Labels on the DataBase), a novel, scalable, pure-SQL framework for answering EA, LD and SD queries, implemented entirely on an open-source database system. Moreover, we formulate four new types of queries targeting public transportation networks, namely the Earliest Arrival and Latest Departure k-Nearest Neighbor (kNN) and Oneto-many queries and propose novel ways to efficiently answer them within PTLDB. Our experimentation will show that the proposed solution is fast, scalable and easy to use, making our PTLDB framework a serious contender for use in real-world applications.