Nicolo Pagan


Last Name

Pagan

First Name

Nicolo

ORCID

0000-0003-0071-7856

Organisational unit

Filters

2019 - 201912020 - 20234
Reset filters

Search Results

Publications 1 - 5 of 5
  • Game theoretical inference of human behavior in social networks
    Item type: Journal Article
    Pagan, Nicolo; Dörfler, Florian (2019)
    Nature Communications
    Social networks emerge as a result of actors’ linking decisions. We propose a game-theoretical model of socio-strategic network formation on directed weighted graphs, in which every actors’ benefit is a parametric trade-off between centrality measure, brokerage opportunities, clustering coefficient, and sociological network patterns. We use two different stability definitions to infer individual behavior of homogeneous, rational agents from network structure, and to quantify the impact of cooperation. Our theoretical analysis confirms results known for specific network motifs studied previously in isolation, yet enables us to precisely quantify the trade-offs in the space of user preferences. To deal with complex networks of heterogeneous and irrational actors, we construct a statistical behavior estimation method using Nash equilibrium conditions. We provide evidence that our results are consistent with empirical, historical, and sociological observations on real-world data-sets. Furthermore, our method offers sociological and strategic interpretations of random networks models, such as preferential attachment and small-world networks.
  • Modeling, Analysis, and Inference in Social Network Formation
    Item type: Doctoral Thesis
    Pagan, Nicolo (2021)
    Human beings are social creatures that are naturally looking for, and in fact need, social interactions to maintain a healthy life and mindset. In the past couple of decades, the advent of social media platforms has facilitated maintaining relationships with family and friends even across long distances, but it has also affected people's behavior in their social activity, e.g., allowing them to connect with others they do not even know in real life. On the other hand, the growing amount of available data is giving a unique opportunity for the analysis of the social network structures as well as of the underlying human social behavior. Social networks constitute highly complex systems because of their size, the inner complexity of their components (human beings), and the feedback and cascade effects due to the inter-dependency between the individual behavior and the network structure. Furthermore, social networks are strongly coupled with other network systems, e.g., financial markets, technological systems, infrastructures. With the objective of advancing our understanding of such a complex ecosystem, this thesis focuses on the modelling, analysis, and inference of the social network formation process. At a high level, we distinguish between online social networks, based on real-world friendships, and online social networks, corresponding to our virtual identity on the social media platforms. In the fi rst part of the thesis we analyze online social networks, whose topological structure is fundamentally the result of the social behavior of the agents that locally strive to optimize their position in the network. According to a number of socio-economic theories grounded on extensive empirical research, improving one's own network position can increase the individual's social capital in different forms, e.g., in terms of social influence, brokerage opportunities, or social support. Hence, we consider a network formation process in which actors' interest is defi ned as a parametric combination of different socio-strategic incentives based on the network topology. We study this individual networking behavior through the lens of game theory: following a utility maximization principle, actors rationally choose their set of followees. While a more common objective in strategic network formation games is to study the equilibria resulting from predefi ned payoff functions, our goal is to rationalize the observed network structures in terms of the unknown individual behavior of the actors. Our approach proves to be mathematically tractable and statistically robust, and one of its main advantages is that it allows for empirical validations on real-world historical data on well-studied online social networks, as well as for comparison with well-known random networks models. In the second part of the thesis, we turn our attention to today's most popular online social networks. These platforms paved the way for the so-called Web 2.0, giving people the opportunity to build a virtual user profi le and to share different forms of User- Generated Content. Some of these online social networks, e.g., Facebook or LinkedIn, require mutual consent for establishing a friendship, hence they mimic the characteristics of our online social networks. Conversely, Twitter, YouTube, Instagram, TikTok, and many other popular platforms are all directed networks, where social media users can virtually follow real-life strangers. By means of the integrated search engines, they can navigate others' user pro files, and their User-Generated Content, e.g., Tweets or Instagram posts, which is typically catalogued by hashtags that facilitate its discovery to the interested users. While the number of monthly active users in these platforms has dramatically increased in the last decade, the scienti c literature on social network formation models has not considered the User-Generated Content as main driving factor. Our main contribution consists in formulating a network formation model based on the attractiveness of the User-Generated Content. The individual linking decision is rooted in a meritocratic principle that rewards those users that provide higher quality content. The comparison between theoretical results and empirical data from Twitch, a popular platform for online gamers, proves that our quality-based model captures a number of realistic features of User-Generated Content-based online social networks.
  • A Classification of Feedback Loops and Their Relation to Biases in Automated Decision-Making Systems
    Item type: Conference Paper
    Pagan, Nicolo; Baumann, Joachim; Elokda, Ezzat; et al. (2023)
    ACM International Conference Proceeding Series ~ EAAMO '23: Proceedings of the 3rd ACM Conference on Equity and Access in Algorithms, Mechanisms, and Optimization
    Prediction-based decision-making systems are becoming increasingly prevalent in various domains. Previous studies have demonstrated that such systems are vulnerable to runaway feedback loops, e.g., when police are repeatedly sent back to the same neighborhoods regardless of the actual rate of criminal activity, which exacerbate existing biases. In practice, the automated decisions have dynamic feedback effects on the system itself – which in ML literature is sometimes referred to as performative predictions – that can perpetuate over time, making it difficult for short-sighted design choices to control the system’s evolution. While researchers started proposing longer-term solutions to prevent adverse outcomes (such as bias towards certain groups), these interventions largely depend on ad hoc modeling assumptions and a rigorous theoretical understanding of the feedback dynamics in ML-based decision-making systems is currently missing. In this paper, we use the language of dynamical systems theory, a branch of applied mathematics that deals with the analysis of the interconnection of systems with dynamic behaviors, to rigorously classify the different types of feedback loops in the ML-based decision-making pipeline. By reviewing existing scholarly work, we show that this classification covers many examples discussed in the algorithmic fairness community, thereby providing a unifying and principled framework to study feedback loops. By qualitative analysis, and through a simulation example of recommender systems, we show which specific types of ML biases are affected by each type of feedback loop. We find that the existence of feedback loops in the ML-based decision-making pipeline can perpetuate, reinforce, or even reduce ML biases.
  • A meritocratic network formation model for the rise of social media influencers
    Item type: Journal Article
    Pagan, Nicolo; Mei, Wenjun; Cheng, Li; et al. (2021)
    Nature Communications
    Many of today’s most used online social networks such as Instagram, YouTube, Twitter, or Twitch are based on User-Generated Content (UGC). Thanks to the integrated search engines, users of these platforms can discover and follow their peers based on the UGC and its quality. Here, we propose an untouched meritocratic approach for directed network formation, inspired by empirical evidence on Twitter data: actors continuously search for the best UGC provider. We theoretically and numerically analyze the network equilibria properties under different meeting probabilities: while featuring common real-world networks properties, e.g., scaling law or small-world effect, our model predicts that the expected in-degree follows a Zipf’s law with respect to the quality ranking. Notably, the results are robust against the effect of recommendation systems mimicked through preferential attachment based meeting approaches. Our theoretical results are empirically validated against large data sets collected from Twitch, a fast-growing platform for online gamers.
  • The Impact of Recommendation Systems on Opinion Dynamics: Microscopic versus Macroscopic Effects
    Item type: Conference Paper
    Lanzetti, Nicolas; Dörfler, Florian; Pagan, Nicolo (2023)
    2023 62nd IEEE Conference on Decision and Control (CDC)
    Recommendation systems are widely used in web services, such as social networks and e-commerce platforms, to serve personalized content to the users and, thus, enhance their experience. While personalization assists users in navigating through the available options, there have been growing concerns regarding its repercussions on the users and their opinions. Examples of negative impacts include the emergence of filter bubbles and the amplification of users’ confirmation bias, which can cause opinion polarization and radicalization. In this paper, we study the impact of recommendation systems on users, both from a microscopic (i.e., at the level of individual users) and a macroscopic (i.e., at the level of a homogenous population) perspective. Specifically, we build on recent work on the interactions between opinion dynamics and recommendation systems to propose a model for this closed loop, which we then study both analytically and numerically. Among others, our analysis reveals that shifts in the opinions of individual users do not always align with shifts in the opinion distribution of the population. In particular, even in settings where the opinion distribution appears unaltered (e.g., measured via surveys across the population), the opinion of individual users might be significantly distorted by the recommendation system.
Publications 1 - 5 of 5