Signs and Formal Concepts

In this paper we propose a semiotic conceptual framework which is compatible with Peirce’s definition of signs and uses formal concept analysis for its conceptual structures. The goal of our research is to improve the use of formal languages such as ontology languages and programming languages. Even though there exist a myriad of theories, models and implementations of formal languages, in practice it is often not clear which strategies to use. AI ontology language research is in danger of repeating mistakes that have already been studied in other disciplines (such as linguistics and library science) years ago. Just to give an example of existing inefficiencies: Prechelt (2000) compares the implementations of the same program in different programming languages. In an experiment he asked programmers of different languages to write a program for a certain problem. All programmers of so-called scripting languages (Perl, Python, Tcl) used associative arrays as the main data structure for their solution, which resulted in very efficient code. C++ and Java programmers did not use associative arrays but instead manually designed suitable data structures, which in many cases were not very efficient. In scripting languages associative arrays are very commonly used and every student of such languages is usually taught how to use them. In Java and C++, associative arrays are available via the class hierarchy, but not many programmers know about them. Therefore scripting languages performed better in the experiment simply because programmers of Java and C++ were not able to find available, efficient data structures within the large class libraries of these languages. Of course, this does not imply that scripting languages always perform better, but in some cases apparently large class libraries can be a hindrance. These kinds of problems indicate that the challenges of computing nowadays lie frequently in the area of information management. A semiotic-conceptual framework as proposed in this paper views formal languages within a system of information management tasks. More specifically, it identifies management tasks relating to names (namespaces), contexts and (object) identifiers as the three contributing factors. These three management tasks correspond to the three components of a sign: representation, context and denotation. It has been shown in the area of software engineering that formal concept analysis can be used for such information management tasks (Snelting, 1995). Snelting uses formal concept analysis for managing the dependencies of variables within legacy code.