Universiteit Leiden

nl en

LUCDH Lunch Lecture

Modelling Social Dynamics on Social Media: Networks and NLP

Date
Friday 9 May 2025
Time
Address
Johan Huizinga
Doelensteeg 16
2311 VL Leiden
Room
Humanities Hub, Huizinga 0.09 or 0.11

Modelling Social Dynamics on Social Media: Networks and NLP


The dynamics of online interactions is an extremely complex process. Existing models have offered empirical evidence into how information cascade grow, e.g. on online discussion boards such as Reddit, and how fast they grow. However, with the increasing availability of data mining techniques, such as Natural Language Processing (NLP), it becomes possible to uncover the mechanisms that drive user engagement (e.g., number of comments) that are beyond the structural and temporal characteristics of information cascade. For example, emotion and stance detection can be used to infer emotions expressed by users and disagreement between them as nodal attributes and edge signs respectively. These features are crucial for understanding the patterns and behaviours of user engagement, given that users’ interactions cannot be solely understood as an outcome of tree structures, but also depend on the actual contexts where they encounter content and their attitudes towards this information. The key questions addressed in this talk will explore questions such as (1) how popularity grows, (2) what may (not) explain the growth and (3) whether there exists a universality of growth dynamics across topics.

This website uses cookies.  More information.