Softwareteknologi DTU - Project No. 0166:  A graphical tool for computing action effects in dynamic epistemic logic
Danmarks Tekniske Universitet DTU
Bachelorprojekt - Softwareteknologi
Project No. 0166:  A graphical tool for computing action effects in dynamic epistemic logic
Aktuelle Tidligere  

Description:

To create an intelligent agent (e.g. robot or softbot) that can efficiently communicate, collaborate and interact with other agents, it is necessary for it to have the ability of reasoning about the beliefs and goals of these other agents. An agent can not be helpful to other agents unless it is able to take the perspective of those other agents. Dynamic epistemic logic (DEL) is a logical formalism for agents that can reason about the beliefs of other agents. In this logic, states are represented by a certain type of labelled graph called a Kripke model, and actions are represented by similar types of labelled graphs. When one wants to compute the result of executing an action in a state, one has to take the "product" of the two graphs. It is possible to do this by hand, but from experience we know this to be quite hard to do, even for experienced researchers in DEL. This has even led to incorrect results in published papers on the subject. This project concerns the development of a software tool in which states and actions represented by graphs can easily be drawn and the product of them can be automatically calculated and displayed. Such a tool could be very useful to researchers applying DEL in various areas of artificial intelligence (e.g. automated planning).

If time allows it, or the number of students working jointly on the project is sufficiently large, the project should include the implementation of an algorithm that can minimise the size of the resulting graphs/states (by implementing so-called bisimulation contraction).

Prerequisites: 

The project is quite mathematically demanding, as the involved structures are rather complex. It is however not beyond what can be achieved in a bachelor project if you are strong in mathematics and algorithms. Furthermore, it requires some familiarity with logic, as can be achieved by the course 02156 Logical Systems and Logic Programming. Since the overall area of this project is artificial intelligence, it would be an advantage to have completed 02180 Introduction to Artificial Intelligence, but this is not a mandatory prerequisite.

Supervisor(s) Thomas Bolander

Sidst opdateret: Oct 30, 2015 af Hans Henrik Løvengreen