Étude de l'utilisation de conversations en langage naturel pour la prise de rendez-vous par SMS

Abstract

Appointment scheduling often leads to heavy workload and resources spending (temporal, monetary or human) for enterprises. Although several products provide online appointment scheduling software using a graphical interface, the phone call is often preferred by customers. Recently, with the growing interest in artificial intelligence and dialog systems, conversational commerce has emerged. It refers to enterprises that provide services through messaging. Businesses tend to replace graphical interfaces by conversational ones to strengthen their Customer relationship. This project was conducted with an enterprise that provides an appointment scheduling software. According to an innovation process, the long term goal is to propose an automatic system to book an appointment by phone. This thesis was focused on communication through SMS. A dialog system prototype has been developed in order to identify, the impact of interface change in an appointment scheduling process and the special nature of an appointment scheduling dialog. First, data gathering was performed, where a set of text messages were collected, classified and modeled. Then, we proposed two approaches to develop the prototype. The first experiment used a state machine to manage the conversation and the entity extraction. The second experiment focused on representing the user text as an intention and studying methods and Tools used for Intelligent Personal Assistant (IPA) development. We conclude that the appointment scheduling process between a customer and an enterprise is more complex than those managed by IPA. This study showed two particular characters of the appointment scheduling. First, an appointment could be incomplete and is constructed during the conversation. Then, the use of a conversational interface involves an appointment proposal that is closely related to the business strategy modeling.

Date	28 Nov 2016
Original language	French
Awarding Institution	École de technologie supérieure
Supervisor	Sylvie Ratté (Supervisor)