Conception des entrepôts de données : Evaluation des besoins flexibles

Abstract

The data warehouse design is very different from database one. In addition, the needs of warehouse users are only as clear as those of OLTP users. Rewriting Requêtes has attracted the attention of many researchers because it is related to several data management issues: query optimization, data integration, data warehouse design. To answer the evaluation and optimization problems of requests in order to specify the needs of the decision-makers (traditional requests) especially in cases where these needs are vague or include inaccurate or indeterminate parts (flexible requests). Fuzzy logic or expert systems or even ontologies can be used to remove indeterminacy. The data source inaccuracies (production base) of decision-making systems can lead to erroneous OLAP analysis. Therefore, the extracted data quality (errors, inaccuracy, etc.) and the precise assessment of the decision-maker needs (Unambiguity, uncertainty and inaccuracy) before choosing the production base has become in the latter years a major focus of decision-making research in general and in the decisional system design in particular. The data imperfections in decision-making systems are due to many causes such as the difficulty of having models close to reality or instruments limitations and acquisition processes. In addition, an inadequate representation of the collected information can also be a source of data quality degradation, the representation becoming too remote from the real world object that we wish to model. The inaccuracy may be due to the needs expression ambiguity of the decision maker. Starting from a misinterpretation of requirements, all DW designs and subsequent OLAP analyzes will be erroneous. As a result, many studies have focused on taking into account vague data in Information Systems but few studies have studied this case in data warehouses. This thesis work comes, along with other works, to fill the void in this context and to initiate a reflection around this problematic, through some contributions allowing the vagueness evaluation of the decisional needs based on requests and leading to the improvement of the data warehouse design quality and the relevance of the OLAP analysis thereafter.