Abstract

Finding quickly the relevant information is essential in many application scenarios. In the past years, huge data collections have been created, but for most users it is still very difficult to find the information relevant for a specific, often complex problem. With the advances in automatic language processing chatbots have been developed to simplify the information search providing an intuitive user interface that gives the user the needed information in a natural dialog. In this work we present a chatbot framework that answers questions related to services offered by the public administration. The framework enables complex dialogs and supports the user with giving hints and recommendations. Based on the framework, public chatbot services have been deployed for 2 major German cities designed to answer questions related to offered service, locations, and appointments. The paper discusses the architecture of the system and explains the developed algorithms. We report experiences running the systems as well as discuss the strengths and weaknesses of the developed approach.

@inproceedings{Chatbots2019,
author = {Andreas Lommatzsch and Jonas Katins},
title = {An Information Retrieval-based Approach for Building Intuitive Chatbots for Large Knowledge Bases},
booktitle = {Proceedings of the LWDA conference 2019},
year = {2019},
isbn = {ISSN 1613-0073},
numpages = {10},
pages = {343-352},
location = {Berlin, Germany},
publisher = {CEUR Workshop Proceedings}
}