Euclid Seminar

Αρχείο παρουσιάσεων του Euclid Seminar. Το σεμινάριο πραγματοποιείται στο εργαστήριο Τομέα Λογισμικού, Ισόγειο κτιρίου Α.

Πέμπτη, 20 Δεκεμβρίου 2018

Ομιλητής
Γεώργιος Τσουλουχάς (Μεταδιδάκτορας)
Θέμα
Αναγνώριση ψευδών ειδήσεων (fake news) σε αρθρογραφία του διαδικτύου και σε κοινωνικά δίκτυα.
Περίληψη

Τρίτη, 27 Νοεμβρίου 2018

1ος Ομιλητής
Γεώργιος Σταματελάτος
Θέμα
“Deriving the Political Affinity of Twitter Users from Their Followers”
Περίληψη
In this work, we show that Twitter users can reveal valuable political information about particular Nodes of Interest (NOIs) they opt to follow. More precisely, we utilize an interesting graph projection method and a series of algorithmic approaches, such as modularity clustering, a minimum linear arrangement (MinLA) approximation algorithm and the DeGroot opinion update model in order to reveal the political affinity of selected NOIs. Our methods, which are purely structure-based, are applied to a snapshot of the Twitter network based on the user accounts of NOIs, consisting of the members of the current Greek Parliament along with their respective followers. The findings confirm that the information obtained can portray with significant precision the political affinity of the NOIs. We, furthermore, argue that these methods are of general interest for imprinting the political leaning of other NOIs, for example news media, and potentially classifying them in respect to their political bias.
2ος Ομιλητής
Ελένη Μπριόλα
Θέμα
“Privacy Leakages about Political Beliefs through Analysis of Twitter Followers”
Περίληψη
In this paper, we focus on privacy leakages about Twitter users and show that simply establishing follower and friend connections in the Twitter network might be enough to reveal sensitive information about the political beliefs of a user. More precisely, we create a Twitter dataset containing the twitter nodes of Greek news media and politicians, as well as their followers and present two approaches to predict the political beliefs of a Twitter user and the politicians the user will potentially follow. The intuition behind both approaches is the saying “Show me who your friends are, and I will tell you who you are”. In the first approach, the politicians followed by the followers of the user are shown to be correlated to the politicians that the user follows. In the second one, the media that the user follows are used to predict the politicians that the user follows. The experimental results of both approaches confirm our claims and show that significant privacy leakages are possible.

Πέμπτη, 22 Νοεμβρίου 2018

Ομιλητής
Ελένη Μπαμπαλή
Θέμα
Fake News
Περίληψη

Πέμπτη, 15 Νοεμβρίου 2018

Ομιλητής
Μαγδαλινή Αναστασιάδου
Θέμα
Αντικειμενικότητα των ΜΜΕ
Περίληψη
Μορφές παραπληροφόρησης και λανθασμένης πληροφόρησης, κατηγορίες μεροληψίας των ΜΜΕ και ιδέες για αλγοριθμικό υπολογισμό της μεροληψίας.