Dialogue Management: Questions to Session 8

• In the TRINDIbook, they introduced the two concepts of "information state" and "dialogue state". What exactly is a "dialogue state" consisting of? What are the finiteness restrictions on "dialogue states" that makes it hard to transform "information states" into "dialoue states"?
• The dialogue systems, presented in the book, all consists of DMEs (Dialogue Move Engines). What is the formal definition of a dialogue move? What are these moves made of? Why are they necessary? And, could you show a concrete example?
• I read the paper "Coding Instructional Dialogue for Information States" from the Trindi publications and they classify for example a core speecch act "move" with DRI-scheme "forward-looking, backward-looking". I think this is from a multi-level conversation acts theory. So if this theory uses some forward- or backward-looking or several layers, it needs to store layers or context or something to look back on. I just wanted to know what this storage system looks like, if it is a database or a stack.