Abstract
The systems envisaged by the life sciences (genetics, agronomy, ecology etc.) are, for the most part, complex systems. In the 1960s, the advent of the computer enabled the capitalization of knowledge gathered in computer programs. The idea of accessing other knowledge by joining programs gave birth to software platforms. The current issue of such software platforms is to answer new questions raised by sustainable development, which involves environmental sustainability, economic viability and social equity. The thesis deals with semantic issues generated by assembly. For our study, we developed an original analysis framework elaborated on the basis of work in cybernetics, software engineering and artificial intelligence. The analysis framework considers morphological, syntactic, and semantic levels of the assembly language. Three generic assembly media (MODCOM, OpenMI, and VLE) and three agronomic software platforms (APES, DSSAT, and SEAMLESS-IF) were analysed. We show that the syntactic rules inherent to assembly language limit its ability to represent some complex systems. We also show that tacit knowledge inherent to programs construction limits the efficiency of metadata and ontology, both used to establish the relationship between the input and the output of programs. To clear up the different aspects of sense, we propose a model for program description established using pragmatic linguistic results and morphosyntactic work. The description is based on the seven pronouns and adverbs that the French language offers. A typology of responses is associated with each of these interrogative pronouns and adverbs using a linguistic bibliography. Our work