Modeling and Querying Multimodal Urban Transportation Networks

This dissertation presents a comprehensive and coherent approach to modeling and querying multimodal urban transportation networks. The time-dependent graph model captures the spatio-temporal aspects of real world transportation systems needed in order to effectively plan trips through the network. The temporal aspects of the model (e.g., transit schedules, current speed of roads) are modeled probabilistically. From the graph model, a relational model is developed in order to facilitate the generation of a new query language—the TRANsportation QUerY Language (TRANQUYL). This language allows for succinct and intuitive expression of complex trip planning queries. In order to facilitate use of TRANQUYL by lay-users, a natural language interface was developed to translate English language queries into TRANQUYL. The approach was shown to be robust with regards to grammatical and vocabulary variation, and capable of translating long, complex queries. The system incorporates self referential language, allowing the user to make references to previously known places of interest as well as generic resources. Much of this functionality is accomplished through the use of ontologies. The efficacy of the system was shown through an intrinsic study.