Specifically, the server and client agree on a set of defined 'objects' (in this case including text boxes, drop-down menus, and images). The client then displays said objects to the user by following instructions from the server. The claimed innovation seems to be that the server and client talk about the aforementioned objects rather than sending exact descriptions of the client's display back and forth.
Alcatel-Lucent intended to apply the patent very broadly - consider a web server as the 'host' and a web browser as the 'terminal', for example.
From a quick read, it appears to be about a server keeping track of menus and clicks on those menus that happen on some client (e.g. smartphone?) app. That seems like absolute nonsense to me, but I may be getting tripped up by the use of the term "host computer."