Please use this identifier to cite or link to this item: http://localhost:8081/xmlui/handle/123456789/9299
Authors: Singh, Tejpal
Issue Date: 1995
Abstract: Design is a process of producing a description of' a system or process to satisfy .a set of requirement. To support the designer in the identification and composition of components of a design solutions requires both synthesis and evaluation methods. Such methods can provide a systematic approach to design allowing the designer to pursue more alternatives and evaluate the alternatives based on a discourse of criteria and value. The use of knowledge based techniques for the exploration of synthesis and evaluation methods maintains a separation of method and knowledge, allowing the designer to gui-de the methods with qualitative or empirical knowledge without sacrificing the benefits of a systematic approach. In this work an expert system shell (ESS) named SHELL has been developed which will', help the user in taking decisions and provide him an expert advice based on the facts provided by the user during the interaction. The ESS have a rule based inference engine (IE) which provides decisions based on the rules applicable for the session, which in turn depend' upon the facts that hold for the project.. The.knowledge .r--epr_es.ent.ation (KR) in the ESS is also with the semantic networks which basically form a fact base for the inference engine module. However it can be used independently to provide answers to the queries which can be answered with the help of a semantic network exploiting the inheritance properties of the object ,present in it. Two ESS modules were developed sharing a common 'semantic network. The modules are rule based With a difference in their learning abilities. One is a non- learning module which requires the complete set of rules to provided beforehand to get complete decisions or to cover all sorts of situations. This module assumes that all the knowledge in the form or rule base or semantic net is complete and correct. The following cases are envisaged in this module The system cannot reach a decision because the facts provided't.o it were conflicting ; The system cannot provide a decision because the rule base has no rule(s) to reach that decision ; The system provides conflicting decisions because the facts provided to it were not coherent ; The system is able to provide the decisions assuming that rules provided to it are complete and correct and so are the facts provided. No confirmation is taken from the user i.e. the system is closed for any feedback. In all these cases the system never tries to learn any new knowledge because it was never made to do so. A new situation has to handled explicitly by adding new rules in it's rule base. The other module capable of acquiring,, automatically & by explanation, the know how of an Architect: to help him in the design process. The module -using an incomplete description, provided by the designer, of the construction project to be designed, the system tries to complete the project using the knowledge at it's disposal which the designer has taught it. The following cases are envisaged : The system cannot solve a problem because it has never learnt how to The system cannot solve a problem because it lacks information The system claims to know how to solve a problem and submits a solution to the designer for approval. In all three cases the system learns the knowledge which it will be able to use at a future date. The learning mechanism resembles the learning methodology of a human being who also learns by experience. The two modules share the -semantic net to provide the various facts. The learning module may also be used to generate more facts for the non-learning module or may be used to complete a partial semantic, to be used by both of the modules at a future date and providing a means of interaction between the two modules. SHELL is developed not with the aim of replacing an architect or the designer, but with the aim of providing him a tool capable of simulating the work of an architect or helper who he has trained himself by providing him with rules and knowledge for handling situations. The model of the system developed is built around the High Level Language (HLL) 'C' with an automatic interpretation, knowledge acquisition & explanation system.
Other Identifiers: M.Tech
Research Supervisor/ Guide: Kumar, Padam
metadata.dc.type: M.Tech Dessertation
Appears in Collections:MASTERS' DISSERTATIONS (E & C)

Files in This Item:
File Description SizeFormat 
ECD246773.pdf11.04 MBAdobe PDFView/Open

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.