Give an application example where global outliers, contextual outliers and collective outliers are all interesting. What are the attributes, and what are the contextual and behavioural attributes? How is the relationship among objects modelled in collective outlier detection?