Quaestiones Informaticae Vol 5 No 3

Generating relations using formal grammars

Thu, 01 Jan 1987 00:00:00 GMT

Generating relations using formal grammars von Solms, S. H. Grammars generating 2-dimensional arrays have been studied by many people [1, 2, 3, 4). One effort was Random Context Array Grammars [4], where different types of context conditions placed on the production, were used to control the generating process. Relations, as used in relational data bases, can be considered as rectangular arrays, and therefore it should be possible to generate and manipulate relations using 2-dimensional Grammars. Simple Relation Grammars generate relations, and can simulate some unary operations like select and project on these relations. Extended Relation Grammars also generate relations, but these relations can communicate with each other within a "Extended Relation Schema (ERS)". Within such an ERS binary operations like union and join can be simulated. This paper is a summary of a research project of which certain parts have already been submitted for publication.

Pointers as a data type

Thu, 01 Jan 1987 00:00:00 GMT

Pointers as a data type Philips, N.C.K.; Postma, Stef W. The distinction between an abstract data type and an implementation of a data type is not as clear as is commonly thought. A hash table can be used to implement the type bag but few of our colleagues think of a hash table as itself being a data type. Similarly, pointers can be used to give a linked implementation of the data type list, but few of us seem to think of pointers as themselves forming a data type. However we shall give an algebraic specification of a pointer data type which is an abstraction of the representation of pointers in Pascal. The usual linked structures can then be specified abstractly via mutually recursive type specifications. In addition we shall describe a uniform method of constructing implementations of data type specifications via operations on strings.

Common approach to some informational systems

Thu, 01 Jan 1987 00:00:00 GMT

Common approach to some informational systems Orlowska, M.E. In this paper we consider relationships between a complete system [6], an L-system [3], a stochastic system [13] and an N-system [8]. We introduce a notion of the induction of one system by another. We prove the important fact that an N-system is a common generalization of all the systems mentioned above. This fact is clearly demonstrated using geometric characterization of the set of subcontinua of the internal <0, l>.

Three packaging rules for information system design

Thu, 01 Jan 1987 00:00:00 GMT

Three packaging rules for information system design Mende, J. After identifying the processing functions required in a computer based information system, the designer needs to combine them into an optimal set of load units. Some "packaging" arrangements yield a better system than others, depending upon characteristics of the data collected from external sources and the data extracted for external users. An effective and technically efficient system satisfies three rules. 1. If two user data types are needed at different times, the corresponding extract functions should be separated in different load units. 2. If source data predates the user data derived from it, the corresponding collect and extract functions should be separated in different load units. 3. If two source data types are available at different frequencies, one being less frequent than the user data derived from it, the corresponding collect functions should be separated in different load units.