Date of Award


Publication Type

Doctoral Thesis

Degree Name



Computer Science


Artificial Intelligence.


Tawfik, Ahmed (Computer Science)




This dissertation embodies a study of the concept of surprise as a base for constructing qualitative calculi for representing and reasoning about uncertain knowledge. Two functions are presented, kappa++} and z, which construct qualitative ranks for events by obtaining the order of magnitude abstraction of the degree of surprise associated with them. The functions use natural numbers to classify events based their associated surprise and aim at providing a ranking that improves those provided by existing ranking functions. This in turn enables the use of such functions in an a la carte probabilistic system where one can choose the level of detail required to represent uncertain knowledge depending on the requirements of the application. The proposed ranking functions are defined along with surprise-update models associated with them. The reasoning mechanisms associated with the functions are developed mathematically and graphically. The advantages and expected limitations of both functions are compared with respect to each other and with existing ranking functions in the context of a bioinformatics application known as ''reverse engineering of genetic regulatory networks'' in which the relations among various genetic components are discovered through the examination of a large amount of collected data. The ranking functions are examined in this context via graphical models which are exclusively developed or this purpose and which utilize the developed functions to represent uncertain knowledge at various levels of details.