In [15, 22, 25, 26] Parnas et al. advocate the use of relational model for documenting the intended behaviour of programs. In this method, tabular expressions (or tables) are used to improve readability so that formal documentation can replace conventional documentation. Parnas [23] describes several classes of tables and provides their formal syntax and semantics. In this paper, an alternative, more general and more homogeneous semantics is proposed. The model covers all known types of tables used in Software Engineering. Contents 1 Introduction 2 2 Introductory examples 4 3 Relations 9 3.1 Cartesian Products, Functions, Relations . . . . . . . . . . . . . . . 9 3.2 Direct Products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 3.3 Input-Output Relations . . . . . . . . . . . . . . . . . . . . . . . . . 12 4 Raw Table Skeleton 14 5 Cell Connection Graph and Medium Table Skeleton 15 6 Raw and Medium Table Elements 19 Supported by NSERC of Canada Grant 7 Well Do...

Tabular expressions (Parnas et al. [17, 23, 25, 26]) are means to represent the complex relations that are used to specify or document software systems. A formal model and a semantics for tabular expressions are presented. The model covers all known types of tables used in Software Engineering.

The concept of being a part of for direct products and relations is analysed. Some operations based on this concept are introduced and analysed. 1

Based on very elementary mathematical concepts, SCENATOR allows a short feed-back cycle between stakeholders and specifiers. It also scales up the size of formal software requirements which can be verified for inconsistency in a short period of time.Keywords: automated technology for verification, software requirements, scenario integration, tabular expressions, verification, validation, formal specifications. 1 Introduction The main issues in requirements engineering are (1) how to get the right requirements fromstakeholders, (2) how to correctly document them, and (3) how to keep them correct through the software evolution. In this paper, we propose a prototype tool called SCENATOR1 thatpartially helps to address mainly the o/rst and the third issues.