Dynamics of belief : epistemology, abduction, and database updates | |
| Author | Aravindan, Chandrabose |
| Call Number | AIT Diss. no.CS-95-2 |
| Subject(s) | Expert systems (Computer science) Logic programming |
| Note | A dissertation submitted in partial fulfillment of the requirements for the degree of doctor of Engineering |
| Publisher | Asian Institute of Technology |
| Abstract | We live in a constantly changing world, and consequently our beliefs and knowledge on the state of the world change over time. This notion of "change" manifests itself in various applications such as knowledge base revision, database updates, view maintenance in databases, counterfactuals, non-monotonic reasoning, abductive reasoning, theory of actions etc. This raises two major questions to be answered: When are we sure that we carry out the change rationally? How this can be implemented for a specific application? So far, these questions have been dealt with separately: various philosophical works on belief dynamics giving the postulates to be satisfied by an rational chang; and specific algorithms for specific applications without explicit discussion on the rationality of the algorithm. The purpose of this dissertation is to ling these two tracks together. To apply the rationality results of belief dynamics to various application, we first observe that belief dynamics should be generalized in two respects: first of all, it should allow certain part of the knowledge to be declared immutable; and secondly, the belief state need not be deductively closed. In this dissertation, we achieve this in two different ways. In the first approach, we simply generalise the existing notions to incorporate the above requirements, without any major changes on the underlying logic or the consequence operator. The second one is a radically different approach, where the immutable part of the knowledge is treated as a logical system, defining its own logical closure and consequence operators. Later we explore how the results of this generalized belief dynamics, referred to as knowledge base dynamics, can provide an axiomatic characterization for "change" in various applications such as view updates in databases. To this effect, first of all, relationship between knowledge base dynamics and abductive reasoning is brought out, We show how abductive procedures can be used to compute change in a knowledge base, and present a general algorithm for the same. Later, we demonstrate how rationality postulates for knowledge base dynamics provide an axiomatic characterization for view deletions in databases. |
| Year | 1995 |
| Type | Dissertation |
| School | School of Engineering and Technology (SET) |
| Department | Department of Information and Communications Technologies (DICT) |
| Academic Program/FoS | Computer Science (CS) |
| Chairperson(s) | Dung, Phan Ming |
| Examination Committee(s) | Phien, Huynh Ngoc;Vilas Wuwongse, |
| Scholarship Donor(s) | Government of Japan |
| Degree | Thesis (Ph.D.) - Asian Institute of Technology, 1995 |