Assistentes Virtuais para Comunicação Empresarial

• Autor(a) principal: João Carlos Gonçalves Costa
• Data de Publicação: 2017
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: https://hdl.handle.net/10216/107201
• topic: Engenharia electrotécnica, electrónica e informática; Electrical engineering, Electronic engineering, Information engineering

In recent years, the number of devices with Internet access as well as the efficiency of remote communications have increased. Despite the fact that today there are numerous communication channels such as chat, email and video, people still consider voice calls as the communication channel that offers a more personal experience in real time. However, in the business world, people are constantly busy with meetings and tasks, failing to respond immediately to all contacts they receive. On the other hand, more and more companies choose to outsource certain tasks as well as allow their employees to work remotely from their home. This made an increase in telephone audioconferences, telephone calls with more than just two participants. The audioconferences allow meetings between work teams located in different locations. Following the presented context, this dissertation was developed starting from two initial objectives. The first focused on the development of a system which, controlled exclusively by voice, would be able to manage several functionalities necessary to perform a telephone conference call, such as scheduling it or establishing its connection. The system takes advantage of a virtual assistant which communicates with the user in a way similar to a human. Together, and this was the second objective of this dissertation, a prototype of a system was developed to take advantage of an Android application for responding to all the native voice calls received by a busy user. A native call consists of a "regular" telephone call using the public switched telephone network (PSTN). The developed application detects a native voice call attempt and promptly starts a communication session with the person initiating the call through a virtual assistant (bot). The virtual assistant asks the user for the reason for the call and interactively offers some options to the person they tried to contact. It should be noted that at the end of this dissertation, both developed virtual assistants work independently. Nowadays there are tools and devices for speech recognition as well as tools for speech synthesis. The development of the virtual assistants took advantage of the mentioned tools, namely Amazon Lex and Amazon Alexa for understanding natural language and creating conversational flows, and the Google Cloud Speech API for converting audio to text. The Amazon Echo Dot device, a wireless speaker which allows the user to receive voice controls, was used to take advantage of Amazon's capabilities and enable voice communications with the user. The Twilio communication platform was also used, which allows, through an API, to receive, control and monitor telephone calls. The evaluation of the efficiency of the realized prototypes was achieved based on user satisfaction surveys and usability tests. With both developed systems, the questions that arose were able to be answered. However, the telephone conference management system has not been as effective and responsive to users as the system created to respond to calls received by a busy user.