Tesi etd-07072020-181703 |
Link copiato negli appunti
Tipo di tesi
Tesi di laurea magistrale
Autore
PANSANELLA, VALENTINA
URN
etd-07072020-181703
Titolo
Modeling Algorithmic Bias in Opinion Dynamics: simplicial complexes, and evolving network topologies
Dipartimento
INFORMATICA
Corso di studi
DATA SCIENCE AND BUSINESS INFORMATICS
Relatori
relatore Rossetti, Giulio
relatore Milli, Letizia
relatore Milli, Letizia
Parole chiave
- algorithmic bias
- complex network
- network analysis
- opinion dynamics
- peer pressure
- simplicial complexes
Data inizio appello
24/07/2020
Consultabilità
Completa
Riassunto
Ad oggi, il modo in cui le persone interagiscono è influenzato dai meccanismi di filtraggio che agiscono sugli online social network, che selezionano le informazioni e gli utenti da mostrare in base a qualche criterio di similarità, creando un bias algoritmico. Inoltre, gli utenti stessi tendono a creare legami con persone che sono più simili a loro e a romperli con chi è troppo diverso. Infine, anche in questi contesti le persone interagiscono anche in gruppo, creando le condizioni per la peer-pressure. Il presente lavoro parte dallo studio di un modello di Opinion Dynamics che implementa l'algorithmic bias su reti random e scale-free. Inoltre, implementa e studia sulle stesse reti modelli che includono meccanismi di peer-pressure e la possibilità di rewiring degli archi, secondo un criterio di omofilia.
Nowadays, the way people interact is influenced by the filtering mechanism in place in online social networks, which selects informations and users to show based on some similarity criterion to enhance platform usage. This creates an algorithmic bias which affects the reality people can see and interact with. Moreover, users themselves tend to create ties with who is more similar to them and to break those with who is too different. Lastly, people interact also group-wise, creating the conditions for peer-pressure. The present works starts from studying an Opinion Dynamics model which implements algorithmic bias on random and scale-free networks; starting from this, models which implement peer-pressure mechanism and the possibility of edge rewiring according to an homophily criterion were implemented and studied on the same networks.
Nowadays, the way people interact is influenced by the filtering mechanism in place in online social networks, which selects informations and users to show based on some similarity criterion to enhance platform usage. This creates an algorithmic bias which affects the reality people can see and interact with. Moreover, users themselves tend to create ties with who is more similar to them and to break those with who is too different. Lastly, people interact also group-wise, creating the conditions for peer-pressure. The present works starts from studying an Opinion Dynamics model which implements algorithmic bias on random and scale-free networks; starting from this, models which implement peer-pressure mechanism and the possibility of edge rewiring according to an homophily criterion were implemented and studied on the same networks.
File
Nome file | Dimensione |
---|---|
Tesi_Val...Final.pdf | 4.66 Mb |
Contatta l’autore |