The core diagram uses actors, use cases, associations, include and extend. Four more notations appear in larger systems, each shown here on a small example, with a note on what it is and when to use it.
A specialised actor is a kind of a more general one, and it inherits all of the general actor's use cases. Reach for it when two roles overlap heavily, one can do everything the other can and a little more, so the shared use cases are drawn once on the general actor instead of being repeated.
One use case can be a specific variant of another, inheriting its behaviour and adding its own steps. Use it when several use cases are the same task done in different ways, to capture the shared behaviour once and let each variant specialise it.
A composite use case stands in for a group of finer ones and is expanded in its own lower-level diagram. Use it when a system has too many use cases for one readable page: the top diagram stays a map, and each composite opens into its own detail.
An extension point is a named place in a base use case where an extension can attach. Use them when a base has more than one optional or exceptional path, so each extend states exactly where it fires rather than vaguely somewhere inside.