Classification of Public Transport Information Dialogues Using an Information-Based Coding Scheme

The goal of a recently started research project of OVR is to develop a system to automate part of its dialogues now held by human operators at its call centres. To achieve a well thought-out design of such a system both an analysis of the current dialogues and an experiment determining which form of dialogues would be favoured by humans were started. In this paper we will emphasize the analysis. Over 5000 dialogues were recorded and transliterated. Our coding scheme based on the information components has been derived from this set. The coding scheme was applied to a sample of nearly 500 dialogues. Not only are we able to describe the semantics of the utterances in a dialogue, but we are also able to classify the dialogues with respect to their diiculty for an automated speech processing system. With a perfect speech recogniser and parser, in theory 88% of the dialogues could be handled by a simple system that would work with question forms. This is not realistic since speech recognition is a serious bottleneck. Dialogues about train information only are preferable over dialogues about other transport modalities for such a system, since these dialogues are shorter and less complex.