A Framework For Experimental Evaluation of Opinion Diffusion Models
| Thesis title in Czech: | Framework pro experimentální studium modelů šíření názorů |
|---|---|
| Thesis title in English: | A Framework For Experimental Evaluation of Opinion Diffusion Models |
| Key words: | šíření názorů|teorie sociální volby|experiment |
| English key words: | opinion diffusion|computational social choice|experiment |
| Academic year of topic announcement: | 2021/2022 |
| Thesis type: | Bachelor's thesis |
| Thesis language: | angličtina |
| Department: | Computer Science Institute of Charles University (32-IUUK) |
| Supervisor: | Mgr. Martin Koutecký, Ph.D. |
| Author: | hidden - assigned and confirmed by the Study Dept. |
| Date of registration: | 17.07.2021 |
| Date of assignment: | 17.07.2021 |
| Confirmed by Study dept. on: | 16.08.2021 |
| Date and time of defence: | 07.09.2023 09:00 |
| Date of electronic submission: | 20.07.2023 |
| Date of submission of printed version: | 20.07.2023 |
| Date of proceeded defence: | 07.09.2023 |
| Opponents: | prof. Adrian Vetta |
| Guidelines |
| An important goal in the study of opinion diffusion is to propose realistic models. The "realism" may be roughly divided into two parts: one, the mechanism of the process should be sufficiently rich, and two, the parameters of the process need to be calibrated to achieve higher levels of realism. The result should be a process which corresponds with real-life observations.
Specifically, we are interested in models where we have a social graph where vertices are endowed with (partial) ordinal preferences which are vectorized with either Kendall tau distance or Spearman's footrule distance. A diffusion step involves either an opinion change based on the influence of peers, or a creation or removal of an edge, based on preference (dis)similarity. The hyper-parameters of the model control the relationship of a preference or edge change to the preference similarity and the open-mindedness and stubbornness of every individual voter. The topic of the thesis is to develop a framework which allows experimental evaluation of the impact of parameter choices on real-world data. This involves constructing the social network based on existing voting data and the opinion diffusion process itself, running the process, and visualizing its impact on the society. The visualization task in particular is quite complex: for example, one might need to cluster the voters and monitor how the opinion diffusion affects the clusters in graph representation or dimensionally reduced voters space. Additionally, we would like to plot the changing political preferences based on popular voting rules, and also plot various parameters such as the degree distribution of the graph and heat maps of where opinions change rapidly in the network. |
| References |
| [1] Shakarian, P., Bhatnagar, A., Aleali, A., Shaabani, E., & Guo, R. (2015). Diffusion in social networks (pp. 47-58). Cham, Switzerland: Springer International Publishing.
[2] Brandt, F., Conitzer, V., Endriss, U., Lang, J., & Procaccia, A. D. (Eds.). (2016). Handbook of computational social choice. Cambridge University Press. |