You are here

Automatic detection of event factuality in Italian

Event date: 
Friday, 3 October, 2014 - 14:30
Location: 
HLT room
Speaker: 
Federico Nanni
Abstract: 

Federico Nanni will present the work he has done on event factuality detection in Italian news during his 2-month internship in the HLT group.

The internship aims at implementing a system for the detection of the polarity of an event (positive or negative), the certainty of an event (certain, probable or possible) and the relative time of an event (FUTURE or NON-FUTURE). These three aspects describe the factuality of an event, i.e. the fact that an event is factual, counter-factual or non factual. In order to detect these aspects of the factuality of an event he has used a method based on machine learning.
He will present the corpus used for the experiments, the method and the features used and the results obtained.

During the seminar he will also introduce his PhD research. His research deals with a methodological analysis of the issues that arise when using born digital documents, especially websites, as primary sources; in particular his purpose is to better understand how the processes of finding, selecting and analyzing sources change when dealing with “born-digital” ones. The case study of his research is a historical reconstruction of the online presence of Italian universities.