Modelling objects, knowledge and learning in distributed object-based systems.
Item
-
Title
-
Modelling objects, knowledge and learning in distributed object-based systems.
-
Identifier
-
AAI9315493
-
identifier
-
9315493
-
Creator
-
Ndjatou, Gilbert.
-
Contributor
-
Adviser: Rohit Parikh
-
Date
-
1993
-
Language
-
English
-
Publisher
-
City University of New York.
-
Subject
-
Computer Science | Artificial Intelligence
-
Abstract
-
Our model is intended to provide a means of specifying and reasoning about the behavior of systems consisting of active, autonomous and interacting components, as well as those of the constituting components while abstracting the detailed mechanisms involved. We also provide mechanisms to describe object classes and inheritance within our model.;Unlike previous models of reactive systems, we would like ours to capture both the data and the process views of objects, on the same lines as the objectcharts of S. Bear et al (BACH90), or L. Lamport's (L89) transition axiom method. Objects are thus viewed as reactive systems with some intelligence and a decision mechanism or the organ of will of J. McCarthy and P. J. Hayes (MH69). To capture the intelligence of objects, we introduce and reason about objects' knowledge within our framework. We also introduce the notions of objects' procedural knowledge and learning in distributed object-based systems. It will then be possible to provide a knowledge-based characterization of object "similarity", inheritance and rational behavior.;An object's knowledge corresponds to facts about its task-domain situations that it should be said to know according to its behavior specifications or protocol. It is static and does not depend on system computations. It is introduced in a modal and dynamic logics framework, and leads to a logic of knowledge and actions. An object's procedural knowledge corresponds to the notion of "laws of ability" in (MH69), and expresses the ability of an object to use and transform its knowledge of facts about its task-domain situations. It is specified in terms of objects' knowledge and actions.;Our approach to learning corresponds to the knowledge obtained and transferred in distributed object-based systems. It also corresponds to object instances' awareness of their own knowledge or properties of computations in distributed object-based systems, and is closely related to the intuitive notion of learning by experience. We introduce a modal logic of knowledge and learning/awareness, and also prove some properties of learning/awareness and forgetting in distributed object-based systems. This approach to knowledge and learning/awareness also gives us a semantical basis to the distinction and relationship between these two aspects of knowledge in distributed object-based systems that we refer to as static/local knowledge and dynamic knowledge. (Abstract shortened by UMI.).
-
Type
-
dissertation
-
Source
-
PQT Legacy CUNY.xlsx
-
degree
-
Ph.D.
