Event-driven (one tag at time) parsers, e.g. expat. These are good for big XML document, but one needs to hand code quite a bit of the tag and content handling logic. You can easily ignore most tags and only retrieve what you need. (We will study this type of parsing in this session) Parse everything to a DOM XML tree, e.g. Gnome's libxml. All the parsing is done for you, but it can be slower for big files. It also uses more memory and the tag filtering needs to be done post-facto. Parse-on-demand parsers (aka "pull parsers"), that internally behave like the event-driven ones, but return a branch of the DOM tree _for the node (or nodes) requested_.