A natural language interface to timetable and journey planner services

OData support
Supervisor:
Recski Gábor András
Department of Automation and Applied Informatics

Tervezz Velem (TeVe) is a publicly available chatbot on Facebook Messenger Platform, which responds to questions written in natural language and displays information about Budapest public transport. Using the chatbot the user can get the schedule of the nearby stops. After the user initiates trip planning with text messages, the application provides detailed information (e.g. duration, estimated arrival time, number of transfers) about the possible itineraries.

The textprocessing module of the application allows the chatbot to determine user intent based on fragmented input with great precision. Thanks to the integration of the e-magyar Hungarian language processing pipeline, the system can detect various word suffixes. The ranking logic enables us to find the most appropriate analisys even for ambiguous input. Currently, the system has more than 180 active users, and is accessible by the public on https://m.me/tevepp, and the full source code is available under an MIT licence at https://github.com/kebou/tevepp.

Downloads

Please sign in to download the files of this thesis.