Workshop on Data Abstraction, Databases and Conceptual Modelling 1980: Pingree Park, Colorado
In 1980 there is the Pingree Park Workshop on Data Abstraction, Databases and Conceptual Modelling in Colorado, United States.
The Pingree Park Workshop is dedicated to remove the boundaries between the AI, database management and programming languages to shape research and development of common interests.
One of the first discussions at the beginning of the workshop is the discussion about “What should be modelled?” and the nature of models.
This discussion suffers two problems. First, it is the initial attempt at the workshop by people in different areas to communicate on the same subject. Second, the issue of “What should be modelled?” is frequently confused with “How should it be modelled?“.
Comments to some discussed topics of participants
In the following there are some comments to the discussed topics of participants.
Kinds of Knowledge:
Deutsch: "There are three different kinds of knowledge each of which can be modelled:
Hayes: “Are we discussing the modeling of programs, which are the representations of some world, or the modeling of the world itself?”
Levesque: “Are we talking about what is and what is not allowed in the real world or in the computer system?”
Rich: "We should distinguish features of the world to be modelled from techniques used to represent those features. Examples of modeling problems concerning the real world are what are objects(..)? What objects can be created and destroyed? Can objects overlap? Corresponding to these features we require representations techniques such as message passing, hierarchical and relational database models and pre- and post-conditions for operations”
Sibley: “Are we talking about the real problem or the solution?”
Properties of Models and How much to Model:
Balzer: “What are some desirable features of models? Should a model be operational … support mappings … should a model contain meta knowledge?“
Balzer: “If we ignore implementation details we can focus on what constitutes a model of an application. How broad is the scope of the model: the application itself, the application environment, the system and its environment? These questions depend on the purpose of the model.”
Position Papers of the Workshop
Some of the position papers of the Pingree Park Workshop in 1980 are summarized in the following. These summaries should give a little impression about the current topics and positions at that time.
Heinrich C. Mayr
Make More of Data Types.
This position paper exhibits a way to attack abstraction and specification problems in the database field. The Pingree Park Workshop confirms Mayr in his impression that the database field suffers from a nonuniform and unprecise terminology that is caused by the lack of a fundamental database theory. In this paper he concentrates on types, derive formal notions of ‘data types’ and ‘abstract data types’, and the advantages they have for the database field.
On Conceptual Database Modeling.
The paper describes the current research in the areas of conceptual database modeling, database design and the architecture of database systems. The goal of one project, called “semantic database models and logical database design”, is to devise principles, techniques and tools to support the structured design of databases and the operations. Also the trend of “distributed databases” and techniques to support the design of federated database systems is discussed. All research projects are concerned with database abstraction and conceptual modeling.
A perspective for research on conceptual modeling.
In this position paper Mylopoulos provides a perspective for research on conceptual modeling carried out the past five years at the University of Toronto. Conceptual modeling refers to the activity of constructing abstract models of knowledge about some world and semantic data model as they have been use in AI and Databases. The focus of the research was the development of descriptive tools for describing such models. Descriptive tools for conceptual modeling can be developed within a framework that is object-oriented, uses procedural attachments and pays attention to aggregation, generalization, classification and a context dimension. Examples for the development of such tools within this framework are PSN, a knowledge representation language, and TAXIS, a semantic data model. PSN treats traditional semantic network concepts within a procedural framework. It introduces aggregation, generalization and classification to structure and organize models. TAXIS is a programming language for the design of interactive information systems.
Michael L. Brodie
Data Abstraction for Designing Database-Intensive Applications.
At this time there is a growing exchange of ideas among AI, database and programming language researchers concerning conceptual modeling of complex and object-oriented applications. The complexity is a result of the complicated structural and behavioural properties such as concurrent interactive access by users with different processing needs over a shared database, information locality and update-oriented transactions. So the problems are the managing of intellectual complexity of the design, development and evolution and the definition and ensuring of semantic integrity. In this paper Brodie describes a project, called Active and Passive Component Modeling (ACM/PCM), in which a comprehensive design and development methodology is formulated to address these problems by the integration of recent AI, DB and PL results. ACM/PM is based on semantic database models from the DB area, predicate calculus from AI and data, procedure and control abstractions from programming language. ACM/PCM results from an attempt to take advantages of the distinct perspectives of these three areas and to integrate different tools.
Robert M. Balzer
Dynamic System Specification.
In this paper eight principles of good specification are represented. These principles are:
These principles have been used to determine 18 requirements of good specification languages and they have formed the basis for a new formal specification language.
Gary G. Hendrix
Mediating the Views of Databases and Database Users.
In this paper requirements for a conceptual schema that is general enough to support knowledge bases as well as ordinary data bases are discussed. There are seven features presented that such a schema should support. These features are: generalization hierarchy, functional dependencies, domain roles, definitional mechanisms, conventional schemata, procedural attachment and inference mechanism. The artificial intelligence notations for semantic networks or conceptual graphs are general ones that can support all seven features.
Patrick J. Hayes and Gary G. Hendrix
A logical View of Types.
Hayes and Hendrix want to show how ordinary predicate calculus can be used to talk about the most notations for which the terms “abstraction” and ”type” are used in various areas of computer science. The reason is that people who are working in the areas of data abstraction, databases and conceptual modelling have argued for a long time over the meaning of these terms. Predicate calculus is represented by them not as suitable tool for implementing types, but as a well-understood, uniform conceptual framework and notation for describing and precisely comparing various ideas of typing.
W. Terry Hardgrave and Donald R. Deutsch
Processing Data Model Abstraction.
Hardgrave and Deutsch assume that the Pingree Park Workshop is dedicated to remove the boundaries between the AI, database management and programming languages to shape research and development of common interests. These three fields can benefit from each other only if each one appreciates the diversity of these research viewpoints. In this paper some shared problems, such as that all fields need a common precisely formalism and the need to recognize that data structures and operations on those structures are equally important, are discussed. Also the approach to design and develop a prototype of a Data Model Processor(DMP). The DMP should accept a semi-tabular, parametric description of a data model and emulates the operation of an implementation of the data model. Therefore precise formalisms are needed to describe data models.
Robert Balzer: What Should Be Modelled? (discussion). Workshop on Data Abstraction, Databases and Conceptual Modelling 1980, pp. 40-42