Data is becoming an increasingly decisive resource in modern complex real-world applications in different domains (such as Scientific Experimentations/Observations, Transport, Energy, Surveillance, Climate and Weather, Healthcare, Social Media, etc.). Modern data is data whose scale, diversity and complexity require new approaches, algorithms and analytics to manage and exploit it. On the other hand, in the considered system to which data science is applied, it seems clear that it is not sufficient to consider only data, and the technical and the finalusers must also be put in the loop. The data sources should also be regarded, as well as their interactions in some cases. It is therefore necessary to have a systemic approach of data science, taking into account globally sources, data and users, and managing their characteristics to come to an effective knowledge able to support decisions. Among the major characteristics, it is important to address: uncertainty, complexityand ambiguity. Uncertainty refers to the handling of data subject to a doubt on their validity. Complexity is pertaining to the real world about which data are available only through perceptions, measurements and natural language concepts-based representations. Ambiguity of information can result from the use of natural language, from conflicting sources or from incomplete information. In this presentation, we show how techniques stemming from Computational Intelligence can contribute to provide effective and efficient solutions to data science field for managing and handling of modern big data for decision-making.
Allel HADJALI Full Professor in Computer Science ISAE-ENSMA, Poitiers, FRANCE allel.hadjali@ensma.fr
