Nowadays, advanced analysis of data streams is quickly becoming a key area of data mining research, as the number of applications demanding such processing increases. Online mining when such data streams evolve over time, that is, when concepts drift or change completely, is becoming one of the core issues. At the same time, closure-based mining on relational data has recently provided some interesting algorithmic developments as well as practical uses. In this chapter we show how to use closure-based mining to reduce drastically the number of attributes in XML tree classification tasks. Moreover, using maximal frequent trees, we reduce even more the number of attributes needed in tree classification, in many cases without losing accuracy. We show a general framework to classify XML trees using subtree occurrence, composing a Tree XML Closed Frequent Miner with a classifier algorithm. We present specific methods that can adaptively mining closed patterns from data streams that change over time.