Hidden Content Unlocked!

Event Identification from the Live Web
The Live Web (or Now Web) is that part of the Web Ecosystem that is constantly updated with information and content, submitted by the users. Therefore, users are content consumers and producers simultaneously. Twitter and Facebook are distinctive platforms of the Live Web.
Because of the user engagement, and the topics that users post information about, the Live Web serves as an important data source for event detection, news reporting, disaster management, etc. So, we developed an architecture, shown below, to help us identify events coming from microblogs.

Architecture of the TwInsight Project

As explained in our paper, we use emotions as the basic input to our event identification technique. We formalise the problem as one of finding outliers in an incoming stream of data points, and use an online event detection algorithm to solve our problem. For us, events are the main cause behind significant, unexpected alterations of the emotional state of a large portion of the platform's users. By monitoring the aggregate emotional state of users, across time and space, we have access to monitoring their emotional state.

To facilitate event identification even further, we have built a front end for our system, that exploits the information that we use to identify the events: location, time, emotions and, finally, textual description. Our UI paper discusses how we are able to present all of the information in a contextualized manner, while coping with the large volumes of data, multilinguism, content diversity and other challenges that are prevalent in the setting that we consider.

GUI to Present Identified Events From the Live Web