Veronica: Multi-Channel Conversational Agent to support the programmer’s reasoning

Conversational agents also known as chatbots are more than ever part of our everyday life from some years. Depending on the application’s environment it is possible to find those kinds of agents more or less involved to aid the user. Their spreading has led to a continuous research to improve even more their utility and to help the automation of more complex tasks. The major objective of this work is to analyze the capabilities of the chatbots inside a different and less common environment and to understand if there is some potential application for the future. Specifically, during this thesis work a prototype of a chatbot, named Veronica, has been created. Its purpose is to help the programmers, on a daily basis, to improve their quality of life in tasks such as code development and keeping track of reasoning documentation. The tool implemented consist of a conversational agent integrated as an extension for Visual Studio Code. The core components of Veronica are the following: the Visual Studio Code IDE in which the chatbot resides, a small headless google chrome browser to perform Speech Recognition and Speech Synthesis and a small google chrome extension to track the user navigation. First, a literature review has been done to understand the existing proposals that rely on conversational agents to support code evelopment without much success. Because of that scarce adoption of conversational agent on that environment, we decided to think about useful tasks that could have been performed for our use case. Then, a second wave of researches has been done concerning the general users preferences when using conversational agents and the best type of chatbot that could have fit for our needs. Follow the reasoning behind the design of the prototype and the relative implementation. Both has been performed by following the most significant insights from the previous step. Subsequently, the description of the evaluation phase performed on the tool is presented. It has been conducted on a group of university students with the final objective to evaluate the capabilities and usefulness of the chatbot abilities and to listen to possible new implementations to improve it even more in future. The thesis work is concluded with a discussion on the results obtained from the evaluation phase.