Modeling Uncertainty in Social Network

In the study of human connection, scientists have long relied on maps that look deceptively simple. These maps, known as social networks, treat relationships as binary switches: two people are either connected or they are not. If they are, a line is drawn between them; if not, the space remains empty. This approach has been useful for counting friends or tracking how quickly a rumor might spread, but it struggles to capture the messy reality of human life. In the real world, a relationship is rarely a simple yes or no. It is a spectrum of feeling. One colleague might view a partnership as a deep, vital alliance, while the other sees it as a casual, occasional exchange of ideas. Traditional maps force these complex, ambiguous feelings into rigid categories, effectively erasing the uncertainty that defines how people actually interact.

Researchers at the University of Tehran have proposed a new way to draw these maps, one that embraces this ambiguity rather than ignoring it. By using a mathematical approach called fuzzy logic, they created a system that allows relationships to exist as "fuzzy" connections, where the strength of a bond is not a single fixed number but a range of possibilities. This method was tested on a real group of university professors, revealing that when we stop pretending human relationships are precise, we see a different, more accurate picture of who holds power and influence in a community. The study suggests that the most important people in a network are not always the ones with the most connections, but often those who serve as the most reliable bridges between others, a role that only becomes visible when we account for the uncertainty in human perception.

The core of this new approach lies in how it treats the data collected from people. In a standard survey, a professor might be asked to rate their collaboration with a colleague on a scale from "very low" to "very high." In a traditional analysis, a researcher might simply turn the word "high" into the number 4 and "very high" into the number 5. This turns a subjective feeling into a hard fact, as if the difference between a 4 and a 5 is as exact as the difference between one meter and two meters. The University of Tehran team argued that this is a mistake. Instead of forcing a single number, they translated these words into "triangular fuzzy numbers." Think of this as drawing a small hill on a graph rather than placing a single dot. The peak of the hill represents the most likely value, while the slopes on either side represent the range of doubt or hesitation the person might feel. This preserves the nuance of the original answer, acknowledging that human judgment is rarely absolute.

To test this idea, the researchers distributed a questionnaire to twenty professors in the Department of Engineering Sciences at the University of Tehran. The survey asked each professor to evaluate their relationships with every other colleague on two distinct fronts: scientific collaboration, such as co-authoring papers or advising students, and social interaction, such as friendly or cultural exchanges. The team then built two separate networks based on these answers. The first network mapped the flow of academic work, while the second mapped the flow of social bonds. Crucially, they did not convert the survey answers into simple numbers. Instead, they kept the data in its fuzzy form, allowing the mathematical analysis to work with the full range of uncertainty from the start.

Once the fuzzy networks were built, the researchers applied three different ways of measuring importance, known as centrality measures. The first measure, degree centrality, simply counts how many connections a person has. The second, closeness centrality, looks at how quickly a person can reach everyone else in the network. The third, betweenness centrality, identifies who acts as a bridge or a gatekeeper, sitting on the shortest paths between other people. In a fuzzy network, these calculations are more complex. Instead of just counting lines, the system calculates the "best" path between two people by weighing the strength and reliability of the connections along the way. It asks not just if a path exists, but how strong and certain that path feels.

The results of this fuzzy analysis offered a strikingly different view of the department compared to traditional methods. In the academic network, one professor stood out as the most connected, acting as a hub who was deeply involved in many collaborations. However, the fuzzy analysis revealed another professor who, while not the most connected, was the most critical bridge. This individual sat on the optimal paths between almost every other pair of colleagues. In the fuzzy model, this professor had a perfect score for betweenness, meaning they were the indispensable link holding the academic structure together. Traditional methods, which forced the data into rigid numbers, had largely missed this structural importance, ranking the professor much lower.

A similar pattern emerged in the social network. While one professor was clearly the most socially active, a different individual emerged as the key connector for the group's social cohesion. The fuzzy approach showed that this person played a vital role in linking different social circles, even if they did not have the highest number of direct friends. The study found that the rankings produced by the fuzzy method were often quite different from those produced by the old, rigid methods. For the measure of betweenness, the correlation between the fuzzy results and the traditional results was so weak that they were almost unrelated. This suggests that the fuzzy approach is uncovering hidden roles that standard analysis simply cannot see.

The researchers also looked at the distribution of these scores to understand the overall health of the network. They found that in the academic world, the network relied heavily on a few key individuals to function, creating a structure where the loss of one person could severely disrupt communication. In contrast, the social network was more evenly balanced, with fewer critical bottlenecks. This distinction is vital for university administrators. If a department relies on a single "bridge" professor for academic collaboration, that person becomes a single point of failure. Identifying these roles allows leaders to make better decisions about where to allocate resources, who to support in interdisciplinary projects, and how to strengthen the network against potential disruptions.

By keeping the uncertainty in the data rather than discarding it, this study demonstrates that the most accurate map of a social network is not the one that looks the cleanest, but the one that admits the most about the complexity of human relationships. The fuzzy approach does not just add a layer of mathematical sophistication; it changes the story the data tells. It reveals that influence is not just about having many friends or being very close to everyone, but about being the most reliable path between people who would otherwise be disconnected. For the first time, researchers have a tool that can measure the strength of a relationship not as a fixed fact, but as a living, breathing uncertainty, offering a clearer view of the invisible architecture that holds communities together.