Μεθοδολογίες πρόγνωσης ακμών σε κοινωνικά δίκτυα με χρήση νευρωνικών δικτύων γράφων
Link prediction in social networks with graph neural networks
View/
Keywords
Link prediction ; Graph neural networks ; Social networksAbstract
The purpose of this dissertation is to study the problem of link prediction on social media using graph neural networks.
Graph Neural Networks are connection models that use the transmission of messages between graph nodes to reflect graph dependency. Graphic neural networks, unlike normal neural networks, maintain a state that allows them to represent information from their immediate environment with arbitrary depth. Link prediction is a major problem for network-structured data. Link prediction heuristics use certain scoring functions, such as common neighbors and the Katz index, to measure link probability. They have received wide practical uses due to their simplicity, interpretability, and for some of them, their scalability. Initially, a reference is made to social networks, their structure, and their analysis as well as other issues related to the research of social network analysis. Then we describe the problem of link prediction, but before moving on to the link prediction techniques, a more general but at the same time extensive reference is made to neural networks in graphs, their general design and how these are applied in the real world.
Postgraduate Studies Programme
Προηγμένα Συστήματα ΠληροφορικήςDepartment
Σχολή Τεχνολογιών Πληροφορικής και Επικοινωνιών. Τμήμα ΠληροφορικήςNumber of pages
52Language
GreekCollections
Except where otherwise noted, this item's license is described as
Αναφορά Δημιουργού-Μη Εμπορική Χρήση-Όχι Παράγωγα Έργα 3.0 Ελλάδα
Αναφορά Δημιουργού-Μη Εμπορική Χρήση-Όχι Παράγωγα Έργα 3.0 Ελλάδα