| dc.description.abstract | Given the ubiquity and significance of social network systems, comprehending the network topology is essential for a deeper understanding of these networks and can help guide online user interactions. One notable phenomenon in social networks, Friendship Paradox (FP), has been extensively studied and has led to the Generalized Friendship Paradox (GFP), which states that an individual’s neighbors, on average, have more of some measurable characteristic or quantity than the individual (e.g., friends/degree in the original FP). However, most of the existing works on FP and GFP only focus on positive relationships on an unsigned network setting. However, users in online social networks have negative relationships just as they do in the real world (i.e. reporting, blocking, leaving a negative review). Thus social networks can be more accurately modeled and information rich with signed and directed edges. To bridge this crucial gap, we investigate (G)FP in signed networks which contain both positive and negative relationships (e.g., friends and foes). Specifically, we propose the Signed Neighbor Paradox metric and its generalized version based on the traditional (G)FP that not only considers homogeneous link relations but also heterogeneous link relations. The Signed Neighbor Paradox aims to capture how users’ representative positive and negative relations compare to that of their friends and foes. To further understand this paradox and the relationship between a node and its neighborhood, we analyze the impact that a node’s local topology, such as degree and local clustering coefficient, have on the signed neighbor paradox. We additionally illustrate how the Signed Neighbor Paradox evolves with respect to time and increasing relationships. Furthermore, we develop a Signed Neighbor-Neighbor Paradox metric to study the relationship between an individual’s positive and negative neighborhood sets. Our analysis is performed on five representative signed social networks and we conclude with discussing numerous future directions. | |
