Spidious: Improving User's Efficiency on Email with Machine Learning and a Better Visualization Approach

Abstract

Communication is present in human life since the beginning, initially at a small scale using the cave paintings and gestures and later, viz. in the twentieth century, at a much larger scale through the use of technology. In 1969, by the virtue of this constant need, came into existence on of the most durable and more used tools of communication: the email. This, in the way we know it, is deeply rooted all over the world to the point that it is estimated that about 196 billion emails are sent on a daily basis, and it is also estimated that the number of email accounts to be approximately 5.2 billion by 2018. At the time of its projection, the email was thought to solve a problem: to communicate with people at a distance. And the truth is that it was solved effectively. However, as this solution has become more popular, associated with a very high pace checked at work and in people's lives, it afforded the creation of another problem, more serious, not solved yet: the lack of efficiency in the use of email. This same problem enhances the existence of high levels of entropy in e-mail accounts and, consequently, it leads to a very considerable loss of time for the user, and in business situations materializes in high costs. Note that this is a problem with an unsurmountable importance and understudied, that certainly cannot continue without a reliable solution. It is estimated that the email consumes about 25% of the working time of its users. And if, in the past, the human being improved its face-to-face communication, once that it was essential, being the communication through email important in the present, why we can't improve that too? In fact, currently, we have seen a great interest from technology companies, in relation to email subject, that it has led to the appearance of some attempted solutions to the problems posed by email. Among these we can highlight, for example, the Inbox by Google. The ActiveInbox, or even the Boomerang for Google da BaydIn. However, and unfortunately for all users, each one of these solutions solves only small details or parts of the problem and, for these reasons, none of them is an alternative feasible to resolve the scientific question under study. Thus, this MSc thesis aims to be an assertive solution that solves the problem of current email efficiency. To this, it will go through two distinct phases. The first involves the creation of a new way of viewing for email, based on new paradigms, which will provide the user essential tools to easily manage a large volume messages. And the second, gives the email a certain level of intelligence using statistical models will shaping with their own user actions, causing the email work for the user and not the user for the email, making it a good servant and a good master.