51. Chari, S., Knowledge Representation Using Structured Modeling, Ph.D. Dissertation, John E. Anderson Graduate School of Management, UCLA, 225 pages, 1988. WMSI Reprint 246. Progress reports presented at the Workshop on Structured Modeling, UCLA, 8/86, and at the structured modeling session at the TIMS/ORSA Meeting in New Orleans, 5/87. Co-supervisors: A. Geoffrion, E.B. Swanson.
This dissertation shows how different types of knowledge (especially abstraction concepts, integrity constraints, and intensional rules) can be expressed within the SM framework. This leads to a better understanding of SM, and of how contemporary semantic data models compare with SM. Chari also explains how the primitives of SM can be defined in a clausal form of predicate logic, and implements this successfully in Prolog. This opens the door to doing logic modeling via SML as an intermediate modeling language and also to the incorporation of logic-based inference engines into SM systems.
52. Chiu, C., Automatic Interface for Solver and MMS: Prototype Implementation, part of M.S. Applied Project, Dept. of Decision and Information Systems, Arizona State University, 12/88. Supervisor: R.G. Ramirez. Presented as "Interfacing SML and IFPS" at the ORSA/TIMS International Meeting, Vancouver, B.C., 5/89. See also the companion paper "Automatic Interface for Solver and MMS: Literature Review" and (with R. Ramirez and R. St. Louis) "An Automatic Interface Between IFPS and Model/Data Bases: A Prototype Implementation," Proceedings of the 1989 International IFPS Users' Meeting, May 1989, Austin, TX.
Chiu built a version of the SM-based Information Resource Dictionary System proposed by Dolk in his Comm. ACM paper "Model Management and Structured Modeling," including an interface with Execucom's IFPS/Optimum for solving linear programming problems. He used C and embedded SQL on top of INGRES. Claimed advantages: 1) flexible linkage among models, data, and solvers, and 2) a fairly simple implementation that removes lexical and syntactic burdens from the solver interface. Disadvantages: some redundancy and execution inefficiency associated with the embedded SQL.
52A. Choi, B., A Model Management Systems Approach to Decisions Analyzed Under Uncertainty: A Relational Database Approach, Ph. D. Dissertation, College of Business, Arizona State University, 311 pages, 1995. Supervisor: K.M. Goul.
This dissertation proposes and implements a model management system for decision tree problems based on SML and Dolk’s Information Resource Dictionary System (see item 95 of this Bibliography). The system uses a relational database management system in order to accommodate large models, facilitate model-sharing, and achieve certain model management flexibilities. The implementation, which is in C with embedded SQL, includes the operations of evaluation, sensitivity analysis, information value determination, transportation of the evaluation result of one decision tree to another, and modification of decision trees.
53. Clemence, R.D., LEXICON: A Structured Modeling System for Optimization, Master's Thesis, Naval Postgraduate School, 96 pages, 6/84. Supervisor: G.H. Bradley.
LEXICON was the first prototype implementation of SM. Done in FORTRAN on a mainframe, it was designed to interface with an advanced LP optimizer.
54. Dechter, A., The Design of Hierarchical Planning Systems By Aggregation, Ph.D. Dissertation, GSM, UCLA, 145 pages, 1985. Supervisor: A. Geoffrion.
Uses an early version of SM to attack problems in aggregation in the context of operations management.
55. Derrick, E.J., Conceptual Frameworks for Discrete Event Simulation Modeling, M.S. Thesis, Dept. of Computer Science, Virginia Tech, Blacksburg, VA, 8/88. Supervisor: O. Balci.
This ambitious thesis examines 13 conceptual frameworks applicable to discrete event simulation, including SM. A traffic intersection model is used to exercise all of the frameworks, and as a common frame of reference for making evaluations and comparisons.
A good effort was made to cast the example in SML, but as might be expected -- in the absence of decent tutorial materials on SML at that time -- there are numerous flaws in the formulation. The basic conclusion is that SM supports the design of static simulation model structure, but not (at least explicitly) the design of dynamic structure; nor does it explicitly support computer implementation. The same conclusion was reached for four other conceptual frameworks: the Entity-Relationship, Entity-Attribute-Set, object-oriented, and process graph method approaches. Generally speaking, Derrick was appreciative of elemental, generic, and modular structures, but found SML difficult to use given only the 1987 Management Science article, WP 346 (the working paper version of the 1989 Operations Research article), and an early version of WP 360.
56. Desai, S., Extensible Database Systems as a Platform for Modeling, Ph.D. Dissertation, John E. Anderson Graduate School of Management, UCLA, 178 pages, 6/92. WMSI Reprint 287. Supervisor: A. Geoffrion.
This dissertation examines in detail the potential of extensible database systems as the basis for a modeling environment that supports algebraic modeling languages for different modeling paradigms. Since commercial extensible database systems do not yet exist, 10 postulates and 11 generic commands are presented to capture the essence of what it means for a database system to be "extensible". A data model is then developed that is sufficiently general to represent a broad class of algebraic modeling languages (AMPL, GAMS, and SML are considered in detail). This data model consists of data structures, operators, and structural constraints, and is described both in English and using the generic commands. Attention is given to data access strategies for good performance and to version management. A partial implementation in the experimental system EXODUS has been carried out. Lessons from this effort are given, and also a comparison of extensible versus object-oriented database systems as platforms for modeling environments.
57. Farn, C.K., An Integrated Information System Architecture Based on Structured Modeling, Ph.D. Dissertation, GSM, UCLA, 252 pages, 12/85. WMSI Working Paper 333. Co-supervisors: A. Geoffrion and E.B. Swanson.
Farn develops an information system architecture that combines characteristics of Management Information Systems, Management Science, and Decision Support Systems. Also of interest is the use of first order logic to describe a subset of SM and to prove that it subsumes certain important data models, and the use of universal relations to simplify data queries.
58. Farn, C.K., "Integrated Information System: User Guide" (companion document to Ph.D. Dissertation), 73 pages, 6/85.
IIS was the second prototype implementation of SM. Done in KNOWLEDEGMAN on the PC/XT, it proved the feasibility of the concepts developed in Farn's thesis.
59. Francis, V.E., Aggregation of Network Flow Problems, Ph.D. Dissertation, GSM, UCLA, 295 pages, 1985. Supervisor: A. Geoffrion.
Uses an early version of SM to attack problems in network flow optimization using an aggregation algorithm.
59A. Greenberg, D., A Comparison of Different Modeling Methods -- Regarding MS/OR Community, M.S. Thesis in Industrial Engineering, Technion, Haifa, Israel, 78 pages (in Hebrew), 2/98. Supervisor: Moshe Pollatschek.
This thesis proposes modifications to SML to make it better able to deal with dynamic simulation models: new functionality within the language for describing time-dependent processes, introduction of "virtual records", and incorporation of active database technology. The literature survey includes SML, GAMS, AMPL, MPL, simulation modeling in industrial engineering, graph-based modeling systems including gLPS, ARENA, and DDES.
60. Hamacher, S., Structured Modeling in an Integrated Oil Company: An Evaluation, M. Sc. Thesis in Engineering, COPPE/Federal University of Rio de Janeiro (COPPE/UFRJ), Brazil, 191 pages (in Portuguese), 1991. Supervisors: E.G.M. Jardim (COPPE/UFRJ) and L.J. Lustosa (Pontifical University of Rio de Janeiro). This work was presented at IFORS-SPC1 (Decision Support Systems) in Bruges, Belgium, 3/91 and also at TIMS XXX - SOBRAPO XXIII in Rio de Janeiro, Brazil, 7/91.
This thesis explores the potential of SM for mathematical programming applications to business planning at PETROBRAS, an integrated oil company. In addition, two extensions of the UCLA FW/SM prototype were implemented: a tool for drawing and performing certain operations on genus graphs, and a tool that facilitates hypertext navigation of SML schemas. Both tools are being distributed from UCLA with the FW/SM prototype as of July 1992.
61. Hamacher, S., Modeling Systems for Operations Research Problems: Study and Applications, Ph.D. Dissertation, Industrial Engineering, Ecole Paris Centrale, Paris, 235 pages, 8/95 (in French). Supervisor: P. Dejax. A 59-slide PowerPoint slideshow is available on-line in English, French, or Portuguese.
Drawing mainly on SM and on Semantic Data Modeling ideas, this dissertation develops a conceptual OR modeling framework, including a graphical model representation. The main design goals are to facilitate communication between analysts and end-users, and to achieve OR/IS models that are both efficient and of high quality. To validate this framework, Silvio implemented it as IGOR (Integrated Graphical modeling system for Operations Research) using C, Pascal, VisualBasic, SQL, AccessBasic, and GraphTalk. IGOR's main features are an interactive graphical interface for building and querying models, a generator for the SML representation of the model, automatic structuring of the model's database, an interface to a mathematical programming optimizer, and some model management facilities. A state-of-art survey of OR and IS modeling frameworks and representation schemes is also included. [August 1998: IGOR is being rebuilt with a graphical interface based on Visio and with better integration with Microsoft Office. See also item 110 in this Bibliography.]
62. Hill, D.S., A Prototype for Converting Linear Programming Models to Structured Modeling Graphs, Master's Thesis in Information Systems, Naval Postgraduate School, 137 pages, 3/89. Supervisor: D.R. Dolk.
The focus of this work is the translation of simple LP models stated in algebraic form into an SML schema. The translation rules employed are not fully general. The platform is an IBM PS/2 Model 80 running MS-DOS. LEX, YACC, Microsoft C, and Oracle were used to implement a prototype translator, which also draws the associated genus graph using an algorithm adapted from Wyant's thesis (listed later in this Bibliography).
62A. Hu, Jian, Visual Modeling for Production and Transportation Systems, M. Sc. Thesis, Dept. of Information Systems and Computer Science, National University of Singapore, Singapore, 123 pages, 1996. Supervisor: G.K. Yeo.
This thesis proposes a visually satisfying and very object-oriented way to represent a large class of structured models, and reports on an ambitious prototype implementation called VMS/SM (Visual Modeling System/Structured Modeling) programmed in Visual C++. A detailed comparison with Chari and Sen's GBMS/SM system (see item 87 of this Bibliography) is included.
63. Huh, Soon-Young, An Object-Oriented Model Management Framework for Decision Support Systems, Ph.D. Dissertation, John E. Anderson Graduate School of Management, UCLA, 155 pages, 6/92. Supervisor: R.C. Sprowls.
This dissertation examines in detail the potential of object-oriented database systems as the basis for a modeling environment that simultaneously supports multiple algebraic modeling languages for different modeling paradigms (AMPL, GAMS, and SML are considered in detail). In addition to models in a pure sense -- they are defined in an object-oriented version of the input-output style of the systems approach -- there are also "functional models" that encapsulate solver functionality. A prototype implementation has been completed using C++ and the OODBMS ObjectStore on a Sun-4. See also item 117 of this Bibliography.
64. Kim, J.W., A Modeling Framework for Integrating Decision Models and Information Systems, Ph.D. Dissertation, Dept. of Management Science, KAIST, Korea, 146 pages, 30 November 1994. Supervisor: Prof. Sung Joo Park.
This dissertation formalizes the IMF framework discussed under item 162 of this Bibliography.
65. Krishnan, R., Knowledge Based Aids for Model Construction, Ph.D. Dissertation, Dept. of MSIS, University of Texas, Austin, December 1987. Progress report presented at the Workshop on Structured Modeling, UCLA, 8/86. Supervisor: R. Lee.
This dissertation proposes an architecture for dynamically constructing model structures, and implements it to construct production/distribution/inventory planning LP models. This involves the design of a domain-specific, first-order logic based problem specification language and multiple knowledge bases for use in model construction. The target representation formalism is SM. A prototype system based on the proposed architecture is implemented in Prolog.
66. Lee, Kang-Woo, Integrated Information Management Architecture: Object-Oriented Approach for Problem Representation, Model Formulation, and Model Execution, Ph.D. Dissertation, Dept. of Operations and Information Management, School of Business, University of Wisconsin-Madison, 288 pages, 1992.
This dissertation extends R. Lazimy's Extended Entity-Relationship approach to object-oriented MMS. One of the interesting points about that approach is its use of inheritance, specialization, and instantiation for model building in order to craft models representing the situation of interest from more abstract model classes. Lee gives special attention to model integration (incl. database design, optimization, and dynamic simulation), multi-paradigm support, and certain kinds of support for interactive modeling done by non-specialists. The justification for including this dissertation in this bibliography is that it includes a great deal of material relating Lazimy's approach to SM: see especially Sec. 2.3 and Appendix A.
67. Lin, Suh-Yun Elva, Subscript Free Modeling Languages: An Improved Basis for Integrating Models and Data in Decision Support Systems, Ph.D. Dissertation, Dept. of Decision and Information Systems, Arizona State University, 178 pages plus appendices, 8/93. Supervisor: R.D. St. Louis.
This dissertation studies and implements a dialect of SM/DB, a language for structured modeling that avoids subscripts; an item by Ramirez in Part IIB of this Bibliography presents the original version. The advantages claimed for avoiding subscripts are enhanced readability, enhanced writability, and easier processing by parallel computers. SM/DB models can be stored in RDBMSs, with SM/DB statements translatable (by a compiler Lin built for INGRES) into SQL supplemented by C as necessary via embedded SQL. Lin gives a detailed argument that SM/DB supports the principles of structured modeling even better than SML does. Many SM/DB translations are given for SML models taken from item 11 of this Bibliography, and a formal procedure is given for producing such translations. SM/DB actually consists of three sublanguages: one for defining model schemas, one for specifying model instances, and one for manipulating models.
See also the SM/DB-related items listed under Lin, Schuff, and St. Louis, under Ramirez, and under Ramirez and Lin in Part IIB of this Bibliography. The language has been renamed SFL (for Subscript-Free Language). Lin is now at Motorola.
68. Maturana, S.V., A Translator Writing System for Algebraic Modeling Languages, Ph.D. Dissertation, John E. Anderson Graduate School of Management, UCLA, 179 pages, 9/90. WMSI Reprint 260. Co-Supervisors: A. Geoffrion and J. Mamer. Presented at the ORSA/TIMS Meeting in Las Vegas, 5/90. Portions of have been presented at ISMP 97 and written up as "Towards a Formalism for Algebraic Modeling Languages," Working Paper, Pontificia Universidad Católica de Chile, 23 pages, 3/98.
The number of high-level mathematical programming modeling languages is at least two dozen and growing. The number of commercial quality optimizers is also substantial. In order to couple a particular modeling language with a particular optimizer, it is necessary to have a translator from the high-level language into a low-level language that the optimizer can "understand". Creating such translators requires a great deal of work. This dissertation develops a framework for mechanizing the production of such translators for a large class of algebraic modeling languages.
The approach taken is to split the translator into 3 parts: (1) a front end that translates from the target modeling language to a proposed standard high-level intermediate representation called HIRAM, (2) an invariant part that translates from HIRAM to a proposed standard low-level intermediate representation called LORAM, and (3) a back end that translates from LORAM to the representation required by the solver. A prototype was built with two front ends, one for AMPL and one for SML, and two back ends, one for MINOS 5.3 and one for LINDO.
69. McMillan, J., An Approach to the Design of Model Management Systems Based Upon Structured Modeling, Master of Information Systems, Dept. of Computer Science, University of Queensland, Australia, 77 pages, 11/90. Supervisor: T.A. Halpin.
McMillan views MMS as parallel to DBMS. He develops a conceptual schema for the structured modeling formalism using the NIAM design method, and a 3NF relational schema for this conceptual schema. Thus any SM model class and model instance can be represented in a relational DBMS. Implementation issues are discussed, including (a) the need for a shell external to the RDBMS that provides an object-oriented interface for model manipulation purposes, and (b) support for model integration in a way that preserves the identity of constituent submodels.
70. Neustadter, L., EML Expression Evaluation: A Core Technology for EML-Based Modeling Environments, Ph.D. Dissertation, John E. Anderson Graduate School of Management, UCLA, 237 pages, 1/93. WMSI Reprint 299. Supervisor: A. Geoffrion. Portions appear in "A Formalization of Expression Semantics for an Executable Modeling Language," Proceedings of the Twenty-Seventh Annual Hawaii International Conference on System Sciences, Vol. III, IEEE Computer Society Press, Los Alamitos, CA, pp. 492-504, 1/94.
Expression evaluation -- calculating numeric and logical function values (exact or symbolic) and computing explicit index sets whose populations are given by formula -- is a crucial operation in computer-based modeling environments having algebraic executable modeling languages (EMLs). It is needed to answer even the simplest user requests for basic computations on models, and also for such purposes as loader/editors that understand model structure, model debuggers, solver interfaces, functional style query languages, and report writers.
This thesis defines a generic data model that is representative of algebraic EMLs, defines a compatible extended relational data model, and views EML expression evaluation as queries over data tables associated with this extended relational model. From this viewpoint, it develops and proves correct an expression translation algorithm from the first data model to the second that eventually should enable evaluation to be carried out efficiently for large models through incremental and optimally organized computations much as in RDBMS query optimization. The framework is checked in detail for applicability to two contemporary EMLs, namely AMPL and SML.
Work to extend and validate the analytic results and to study actual computational performance is the subject of a follow-on study (see Part IIB of this Bibliography).
71. O'Dell, D.D., The Design and Implementation of a Visual User Interface for a Structured Model Management System, Master's Thesis in Information Systems, Naval Postgraduate School, 170 pages, 3/88. Supervisor: D.R. Dolk.
This thesis designed and implemented a graphical interface for SM schemas that (a) enables interactive schema input, and (b) produces a relational database representation from the graphic representation. It was developed on an IRIS 2400 dedicated graphics workstation using Lattice C, the HALO Graphics Library, and the Oracle RDBMS. The implementation is, by design, a "throw-away" prototype. See the companion thesis by Wyant.
72. Park, J.Y., A Computer-Aided Tool for Decision Support Model Building Based on Structured Modeling, Master of Engineering Thesis, Korea Advanced Institute of Science and Technology, 57 pages plus appendices, 12/23/87. Supervisor: S.J. Park.
The main point of this thesis is an implementation of SM using the ISDOS meta-system developed at the University of Michigan by Dan Teichroew and others. ISDOS' Information System Language Definition Manager and System Encyclopedia Manager (SEM) were used to define Structured Model Specification Language (SMSL), and ISDOS' Report Specification Interface was used to generate the input file for model checkers and for an evaluator for function and test elements. The latter are coded in C and run on a SUN workstation, whereas everything else runs on a mainframe. Of course, SMSL is only a subset of SML, as Mr. Park had only WMSI Working Papers 338 and 339 to work from. It is based on an entity-relationship model of the core concepts of SM, and is designed in such a way that the SEM query system can be used to perform completeness and consistency checks. The complete system is called SMSS.
A single example, an income statement, is worked using SMSS. It also is worked using IFPS. SMSS is judged superior on two main grounds: a) greater modeling generality, and b) better separation of general model structure from detailed data.
According to Dr. Sung Joo Park, J.Y. Park's advisor, this thesis grew out of a SM application feasibility study done for the largest Korean steel company. Two models were developed, one for a financial application and the other for a production problem. The results were presented to management, and were well received.
73. Patrick, D.J., The Applicability of Structured Modeling to Discrete Event Simulation Systems, Master's Thesis, Naval Postgraduate School, 124 pages, 3/87. Supervisor: D. Dolk.
This is an elaborate modeling exercise built around the ONEC combat simulation model. SM faces more challenges than Odysseus, some of which are outside of its intended domain of application.
There is much food for thought in this work. The conclusion is reached that, while an early version of SML is judged adequate to represent simulation models, it would not be productive to do so until SM syntax is made easier to learn and more conducive to the kind of programming language logic commonly needed for simulation. The author makes four recommendations: (1) more and better tutorial materials on SM; (2) more material like the informal note "Modeling Categorization Hierarchies" which carefully work out important conceptual structures that arise commonly in modeling; (3) allow a high level programming language interface for defining generic rules; (4) find more efficient ways to incorporate time for simulation models. All of these suggestions merit careful consideration. The first has been fulfilled by WMSI Working Paper 378, the third was explored by pre-prototype studies at the FAW in Ulm (Germany), and several people have worked on the fourth.
73A. Ruark, J.D., Implementing Reusable Solvers: An Object-Oriented Framework for Operations Research Algorithms, Ph. D. Dissertation, Dept. of Electrical Enginering and Computer Science, MIT, 352 pages, 6/98. Supervisor: S.C. Graves. On-line here.
This dissertation proposes an object-oriented framework for implementing reusable solvers. Although not mainly concerned with SM, this work is of considerable interest to the SM community and exhibits multiple roots in the SM literature. See especially Section 1.7 (Overview of Related Modeling Literature), which discusses quite a few papers appearing in this Bibliography, and Section 4.2.1 (Monsanto), which uses an SML description of an industrial application as the basis of an example of part of the proposed framework.
74. Samuel, G., Undergraduate Grading Model Using Geoffrion's Structured Modeling, M.Sc. in Business Systems Analysis and Design, City University, London, United Kingdom, 65 pages, 10/93. Supervisor: W. Duncan.
This thesis applies FW/SM to a complex undergraduate grading system, and uses this experience as the basis for an evaluation of SM.
75. Shieh, Chi-Yin, Arithmetic Formula Evaluation for Structured Modeling, Master's Applied Project, Dept. of Decision and Information Systems, Arizona State University, 12/90.
Mr. Shieh wrote a program that translates SM/DB expressions into SQL select statements for evaluation. (See the SM/DB-related item by Ramirez in Part IIB of this Bibliography.)
76. Suh, C., An Approach to Composing a Structured Model from Validated Submodels, M.S. Thesis, Dept. of Industrial Engineering, Pohang Inst. of Science & Technology (POSTECH), Korea, 60 pages, 12/13/89. Supervisor: E. Suh. For a paper-length treatment, see C. Suh and E. Suh, "Composing a Structured Model from Validated Submodels for Effective Model Management," Technical Report IE-TR-90-02, Dept. of Indus. Eng., POSTECH, P.O. Box 125, Pohang, 790-600, Korea, 29 pages, 1/20/90.
A schema integration approach is proposed and prototyped using the FoxBASE+ 2.0 relational database management system. It deals essentially with the Level 1 and Level 2 structure of SML.
77. Tsai, Y. Structured Modeling Query Language, Ph.D. Dissertation, John E. Anderson Graduate School of Management, UCLA, 180 pages, 1991. WMSI Reprint 277. Supervisor: A. Geoffrion. Derivative paper "The Theory of SML Schema-Directed Query" submitted for publication, 11/97.
The general topic is query languages in the context of modeling; not only of data, but also of general structure. This is a significant topic if modeling systems are to gain one of the functionalities that has helped to make database systems so successful. Tsai adopts SML as his model description language, proposes an internal relational representation, and designs an extension of SQL that exploits the semantics of SML. The resulting query language -- actually a trio of related languages -- has a notably easy user interface, can exploit many of the capabilities of available RDBMSs, and has been partially implemented.
78. Vicuña, F., Semantic Formalization in Mathematical Modeling Languages, Ph.D. Dissertation, Computer Science Department, UCLA, 189 pages, 1990. WMSI Reprint 255. Co-supervisors: A. Geoffrion and M. Melkanoff. There is also a 193 page Technical Report, "Specification of a Syntax-Directed Editor for SML."
Most existing languages for mathematical programming have semantics that are only incompletely formalized. This situation -- studied in detail for AMPL, GAMS, and LINGO -- inhibits efforts to achieve a high level of automation for diagnosing errors and generating major components of a computer-based modeling environment (e.g., language-based editors, type inference tools, and immediate expression evaluators). The focus of this dissertation is to demonstrate that attribute grammars furnish a suitable declarative formalism with which to overcome this serious deficiency. SML is the main target language, although a similar development should be possible for most other mathematical modeling languages. The feasibility of the proposed approach is demonstrated by expressing the full syntax and semantics of SML in attribute grammar equations, and by using them in conjunction with the Cornell Synthesizer Generator to generate a language-directed editor for SML on a UNIX workstation. Both the automatic detection of missing language constructs and the immediate evaluation of numerical and logical expressions are addressed.
79. Worobetz, N. Dan, A Computer-Based Management System for Statistical Models, Ph.D. Dissertation, Purdue University, 230 pages, 1992. Supervisor: G. Wright.
Author's abstract: "This dissertation describes an architecture for a model management system that performs statistical analysis on databases. The system allows for the use of several standard statistical procedures resulting in a decision support system that can be uniquely designed for individual needs. The architecture is based extensively on Structured Modeling (Geoffrion 1987). ORACLE is used as the system controller and statistical analysis is achieved using ORACLE's menu system. PC-SAS is used as the primary solver."
80. Wyant, M.A., Design and Implementation of a Prototype Graphical User Interface for a Model Management System, Master's Thesis in Information Systems, Naval Postgraduate School, 98 pages, 3/88. Supervisor: D.R. Dolk.
See the companion thesis by O'Dell. While that thesis covered the translation from graphical representation to a relational database representation (Dolk's), this thesis covers the reverse. The same software tools as O'Dell's are used, and the target hardware is the IBM PC with EGA adapter. The implementation is a "throw-away" prototype, and there is a useful discussion of what was learned from it.