Combination of cognitive and HCI modeling for the design of KDD-based DSS used in dynamic situations

• To ensure and improve the human–computer interaction in dynamic decision support systems based on knowledge discovery in data, our research context concerns the modeling of the decision-maker behavior to accomplish complex dynamic decision-making tasks and producing logically valid predictions.
• The methodological contribution consists in designing visual KDD-based Dynamic DSS using a cognitive model. We propose to adapt the well-known Hoc and Amalberti model cognitive model under the KDD specificities.
• The approach was applied in the medical domain for the fight against nosocomial infections in the ICU.
• The evaluation of this proposal under the utility and usability dimensions shows satisfactory results.

Recent work in dynamic decision support systems (DSS) has taken impressive steps toward data preparation and storage, intelligent data mining techniques, and interactive visualization. However, it remains difficult to deal with the uncertainty and complexity generated by the Knowledge Discovery in Databases (KDD). This paper launches the challenge by introducing cognitive modeling for specifying decision-maker behaviors more naturally and intuitively. It consists in introducing cognitive modeling for dynamic situations involving visual KDD-based dynamic DSS. This research work presents an adaptation of a well-known cognitive model under the KDD specificities. We provide cognitive modeling application in visual KDD-based dynamic DSS for the fight against nosocomial infections in an intensive care unit. Finally, we built a series of evaluations verifying the system's utility and usability.

Figure optionsDownload as PowerPoint slide