Abstract
News of disruptive events in history such as terror attacks often appear to change the behaviour of a society and influence how people react to future events of a similar kind. In the first part of the thesis I will introduce a model of a society represented as a collection of agents which exchange opinions under the influence of the pressure of the society and external news. Interactions between agents encode a combination of human tendencies including assimilation and differentiation and depend on the past history of their mutual agreement and disagreement. Our study shows how the exposure to external news can shape the connections between the individuals and create a collective memory of opinions. The opinions formed in the society in response to these external events can be ‘stored’ in the society’s interpersonal interactions and recalled when similar triggering signals arrive. We will show how presenting news with different intensity and frequency of appearance, such as persistently, periodically, intermittently or randomly, allow weaker or stronger memory formation. In most of the cases analysed we are able to study analytically the long time behaviour of the society. When news appear in an intermittent way, we calculate the minimum fraction of time the society needs to be exposed to a news item to form a stable memory of it. We also describe the role of the frequency and strength of signals in the memorization of the opinions and we show how the latter can be influenced by the mutual interaction between concomitant news. When the society is exposed to a long series of news, our model exhibits properties of a forgetful memory, allowing the society to remember most recent news and forget older ones. Finally we show that even distorted versions of previously presented news can trigger the recall of the corresponding originally stored memories.While the first part of the thesis concerns the modelling of opinion formation, the second part is focused on the study of a more specific realization of the society as virtual community which exchanges ideas through online discussions. Online conversations are much larger in scale than offline ones, with many users interacting and readers often forced to sample a subset of the arguments being put forth. Since such sampling is rarely done in a principled manner, users may not get all the relevant arguments needed to get a full picture of the debate. The second part of the thesis is thus dedicated to answering the question of how users should sample online conversations to selectively favour logically justified arguments. Using a combination of argumentation theory and complex networks we are able to obtain the distribution of justified arguments along simulated and real online discussions extracted from a debating platform called Kialo. We observe that the probability of finding justified arguments at a certain point of the discussion depends on factors such as the degree distribution of the discussion graph and the fraction of ‘supporting’ comments in the debate. We will also show that the location and the quantity of unrebutted arguments (i.e. the “last words”) have a strong influence on the location of the justified arguments in the debate.
In this thesis I will thus present the study of two interdisciplinary problems concerning physics, computation and society. The approach used to solve them is both theoretical and data-driven, with a major focus on statistical mechanics, probability theory and complex networks.
| Date of Award | 1 Aug 2021 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Chiara Cammarota (Supervisor) & Reimer Kuhn (Supervisor) |