The processing of large, complex and distributed datasets needs to minimize the waste of network resources and maximize interaction with a user. Where to find the use of the targeted application in the field of decision support or data analysis. On the other hand, the more complex multi-criteria query Skyline used in many applications requiring multi-criteria decisions to identify the best results based on user preferences or conditions. In this context, this thesis presents the proposal of two distinct approaches located within the framework of optimizing the time to extract information and increased the accuracy of selecting this information in a large, heterogeneous and distributed database.
In the first approach, a multi-agent system has been proposed which consists of five agents, such that each agent manages a system component which ensures the division of system tasks between the agents. As part of data analysis, we have proposed a database of results that makes it possible to minimize the amount of data in the search engine for association rules using the Pasquier method, which divides the rule generators into two bases: generic base for exact association rules and informative base for approximate association rules. The second proposal presents a new approach for selecting the best elements applied according to user preferences as part of the treatment of the multi criterion problem. This approach is based on the proposal of a Skyline query algorithm (Dynamic join Skyline) in a complex environment through the use of IOT options and has been applied in the context of booking travel services. A series of experiments have been carried out and it gives much better results.