Clustering User-Behavior in a Collaborative Online Social Network : A Case Study on Quantitative User-Behavior Classification
Andreas Johansson · KTH Publication Database DiVA (KTH Royal Institute of Technology) · 2016
This thesis investigates how quantitative user data, extracted from server logs, and clustering algorithms can be used to model and understand user-behavior. The thesis also investigates how the results compare to the more traditional method of qualitative user-behavior analysis through interviews and observations. The results show that clustering of all user data, as opposed to interviewing only a small subset of users, increases the reliability of findings. However, the quantitative method has a risk of missing important insights that can only be discovered through observation of the user. The conclusion drawn in this thesis is that a combination of both is necessary to truly understand the user-behavior.