U-mut:Autolisp'le bilgisayar destekli uzman-mutfak tasarımı

Abstract

Bu çalışmada, tasarım ve bilgisayar konusunda de neyimsiz kullanıcıların verilerinden faydalanarak, oto matik mutfak tasarımı yapabilen bir uzman sistem, umut (Uzman-Mutf ak Tasarımı ) geliştirilmiştir. Uzman siste min geliştirilmesinde, programlama dili olarak LİSP'in bir başka biçimi olan AutoLISP kullanılmıştır. Bu dil aracılığıyla geliştirilen yazılım AutoCAD 10 bilgisayar destekli tasarın paketi içinde çalıştırılmıştır. Birinci bölümde, çeşitli etkenler göz önünde bulun durularak yapılan çalışmanın önemi ve amaçları üzerinde durulmuş ve mimarlıkta bilgisayar kullanımına genel o larak değinilmiştir. İkinci bölüm, bilgisayar destekli tasarıma ilişkin ana temalarla birlikte, bu işlem için önem taşıyan bilgi yapıları ve bilgi tabanı, iletişim, bilgi edinme, höristikler ve protokol analizleri gibi kavramlarları açıklar. Üçüncü bölüm bütün detaylarıyla uzman sistemleri kapsar, yapay zeka ve uzman sistemlerin gelişimi örneklerle açıklanmıştır. Ayrıca uzman sistem lerin uygulama alanları ve. avantajları da bu bölümde tartışılarak uzman sistemlerin yapısı ve uzman sistem dilleri üzerinde durulur. Dördüncü bölümde geliştirilen, uzman sistemin, umut 'un programlama dili, programın kurgusu ve sistem mimarisi açıklanmış, sistem içinde kullanılan bilgi simgeleme metotlarına ilişkin örnekler verilmiştir. Bu bölüm, girdi kolaylıkları, tasarım üni lerinin sunuluşu, yazılım içinde yer alan kurallar ve bu kuralların bilgisayar tarafından değerlendirilmesi alarak açıklanan otomatik tasarım gibi olguları irdele yerek, bilgisayar programının tanıtımını yapar. Prog ramın ve menünün yazılımlarından birer bölüm örnek ola rak verilmiştir. Beşinci bölüm, geliştirilen uzman sis teme özgü eleştirileri değerlendiren bir tartışma bölü müdür. Altıncı bölüm, araştırmada varılan sonuçların, gerek uzman sistemler, gerekse Türkiye açısından değer lendirilmesiyle birlikte, gelecekte yapılabilecek ça lışmalara ilişkin önerileri içerir. Ek olarak, geliştirilen uzman sistem programının kılavuzu sunulmaktadır. vıı U-MuT : A COMPUTER AIDED EXPERT-KITCHEN DESIGN WRITTEN IN AUTDLISP SUMMARY Computers and computing are being an integral part of design- and they are already indispensable in design analysis. Computers are being used in offices for design and job management functions. They are increa singly used to produce.drawings and other graphical rep resentations to communicate ideas. Design is concerned with concepts, ideas, judgements and experiences ;. All of these appear to be outside the realm of traditional computing. Human beings communicate with each other using models of their world, which are mostly unrelated with mathematical descriptions or procedural representations. They make use of knowledge about objects, events and processes, and make declara tive statements about them. These are often written down symbolically.' Traditional computing, as it has limitations, is unable tb represent and manipulate knowledge in an explicit and coherent form, and also unable to farm symbolic computation. More recently, graphics have been produced as a result of design decision making. : Developments in know ledge engineering in the form Df expert systems in com puter aided design have pointed ou:t to increasing capabilities of desing decision making by incorporating some of the designer's knowledge and expertise, directly and explicitly into the computational process. Expert systems are relatively new and can be attrac tive, for designers. An expert system is a useful tDDİ for solving ill-defined problems such as those is architectural design, where intuition and experience are necessary ingredients.

Computers and computing are being an integral part of design- and they are already indispensable in design analysis. Computers are being used in offices for design and job management functions. They are increa singly used to produce.drawings and other graphical rep resentations to communicate ideas. Design is concerned with concepts, ideas, judgements and experiences ;. All of these appear to be outside the realm of traditional computing. Human beings communicate with each other using models of their world, which are mostly unrelated with mathematical descriptions or procedural representations. They make use of knowledge about objects, events and processes, and make declara tive statements about them. These are often written down symbolically.' Traditional computing, as it has limitations, is unable tb represent and manipulate knowledge in an explicit and coherent form, and also unable to farm symbolic computation. More recently, graphics have been produced as a result of design decision making. : Developments in know ledge engineering in the form Df expert systems in com puter aided design have pointed ou:t to increasing capabilities of desing decision making by incorporating some of the designer's knowledge and expertise, directly and explicitly into the computational process. Expert systems are relatively new and can be attrac tive, for designers. An expert system is a useful tDDİ for solving ill-defined problems such as those is architectural design, where intuition and experience are necessary ingredients. vur This dissertation desribes a knowledge-based approach ta computer aided design which focuses on the computer rather than the user as a designer. Most Computer-aided design tools today regüire a considerable amount of learning time in order to be used productively and they are merely drafting tools rather than design tools. Users who are neither computer experts nor pro fessional designers mill therefore have a hard time using such tools ta produce useful designs. In order ta help users, they must be aided with the necessary support during design process. This support will be made available by "construction kits, interactive systems and automated designs. Construction kits are means for constraining the design space by limiting the number of desing building blacks, commands and answers that users can select. Interactive system provided by software, realizes the communication between user and the computer, Automated design is made up of the expert system eva luating the rules in knowledge base. A system, U-MuT (expert - Kitchen Design) which demonstrates these ideas in the area af kitchen design will be described. U-MuT provides a set of domain specific building blacks and has knowledge about haw to combine these building blocks into useful designs. U- MuT is an expert system which generates design solution automatically. Kitchen desing was chosen for this study since it has well established design principles in an ill defined problem domain. This makes it attractive far applying knowledge-based system techniques. The ideas presented here are meant, however, ta extend beyand kitchen design.. In' the first chapter, the importance and the goals af this dissertation are explained and the advantages and the disadvantages of using computers in all sections of architecture are discussed. An abjective af computer aided design is to combine the designer and computer into a problem salving team which is able to achieve the goals of design problems more efficiently than either of them could working alone. In this symbiosis the computer is often an efficient obstacle; efficient because it produces desired results^ but an obstacle in the sence that the designer is required to put same effort into achieving those results. This thesis is an attempt to apply rule-based system techniques in computer aided design to build the communication channel between user and computer in order to overcame this obstacle. To achiye the objective, an expert system which behaves like a human expert, is included. Software and knowledge base are the mast important part of such a system. IX Knowledge-based computer aided design represents a novel -approach to computer aided design which brings together concepts from artificial intelligence, language theory and computation theory. It opens up the possi bility of closer cooperation between the designer and the system. Designer is using by tailoring the knowledge to the his or her needs and preferences. This has impor tant ramifications in terms of the acceptability of computer-aided design systems and in terms of increasing individualization of the design process when using com puters. The second chapter is concerned with the subjects interested in computer-aided design such as knowledge base, knowledge acqusition, communication, designer's behaviour, heuristics and protocol analysis. Computer aided design may be readily broken into two fundamental classes of activities one concerned with interactive graphics and the other with decision making. The designer uses the computer to aid in the production of a design and the above distinction is based on the location of the decision making in the entire process. Knowledge-based and expert systems are important components of integrated CAD systems. Today's CAD sys tems have extensive capabilities, but are currently li mited to problem solving tasks which are algorithmic or deterministic in nature, such as analysis, scheduling, estimating and graphical presentation. However, many design activities, including synthesis, evaluation, modelling, and the creative aspects of the design processes, are ill structured and require heuristic so lutions based Dn judgement and experience. Due tD their nondeterministic nature, these problem solving tasks have not yet been successfully computerized. Knowledge based programming technics (such as an expert system) enable the computer to assist in such ill structured problem solving tasks. A human expert does nut use algorithms to solve problems. In an algorithm a goal (or type of answer) is assumed, and a series of steps carried out in logical order to lead to that goal. Human beings solve problems which are less well defined, and the methods they use are not fully understood. Expertise is reached over a number of years' experience. It consists of much more than just facts and rules. Experts try methods and approaches, developing rules of thumb or heuristics. "Brain problems" do not demand the correct answers, but an adequate answer. This involves weighing up different pieces of evidence in order to choose a path from the several available paths. The potential outcomes of dif ferent paths need to be assessed and compared with the goal. The path with an appropriate outcome is chosen. The obvious method of elicitation is to identify an expert and question him, or to get a group of experts to talk to each other. An interview is the most common method of fact finding in system analysis for computer systems. The trouble is that an expert is not aware of his knowledge and how he uses it. The very act of saying things out aloud or being forced to think about them, may affect the way he views the process. Proto col analysis is based on a transcripted interview, but attempts to structure the process, and produce more meaningful results. Experts find it much easier to talk about specific examples of problems than to talk in abstract terms. The analysis of the transcripts requires a systematic breakdown of the information to produce a structured model Df the expert's knowledge. The knowledge base contains the know ledge specific to the domain of the problem to be solved. The knowledge acqusition module of a system serves as an interface betwen the expert and the expert system. It provides a means for entering domain specific knowledge into the knowledge base and revising this knowledge when necessary. In the third chapter, expert systems are explained. Expert systems are computer programs which attempt to behave in a manner similar to rational human experts. They all share a common fundamental architecture even if the knowledge encoding mechanisms differ. This chapter includes definition of Artificial Intelligence (AI) and expert systems, some examples of these their application areas, advantages and the structure of expert systems. Expert systems have been defined as intelligent computer programs which use symbolic infe rence procedures to deal with problems that are diffi cult enough to require siqnificant human expertise for their solution. XI An expert can analyse a problem, assemble facts, use knowledge to infer other facts, evaluate.postu late, make decisions, give advice, explain his reasoning and learn. Much of his knowledge, acquired over time, is ofthe experiential type. An expert system attempts to emulate on expert but does not necessarily model precisely his processes. All expert systems contain the following three basic campanets; knowledge base, context and inference mechanism. The languages and many expert system tools are also given in this chapter. The crux in using a computer is the software, ie. the programs. Programs are written in programming languages which generally look like a cross between algebra and elementary English. Programs can only be written in when the means for producing solution has been completely explicated. The fourth chapter contains introduction of the expert system called U-MuT (Expert-Kitchen design) for kitchen design and gives information about its program ming 'language, ie. software. Knowledge representation in the system architecture consists of three parts: * User input and presentation of design units * Rules (presentation of expert's knowledge) * Evaluation of the conditions -autameted design. A representation is a set of syntactic and semantic conventions that make it possible to describe things, where the syntax specifies the symbols, that may be used and the way those symbols may be arranged; whereas the semantics specifies how meaning is embodied in the symbols and the symbol arrangements allowed by the syntax. The main forms of knowledge representation are semantic networks, frames (schemata or scripts) predicate logic production systems and decision tables. The determina tion of requirements in carried out by designers before designing. The appropriate requirements thus- serve as performance specifications for the design. In this expert system, rules provide a formal way of representing recommendations, directives, or strategies, they are often appropriate, when the domain knowledge results from empirical associations developed through years of experience, of salving.problems in kitchen design. Rules are expresaed as IF-THEN statements. In a rule -based expert system, such as U-MuT the domain knowledge is