Comprehensive evaluation method of visual analytics tools based on fuzzy theory and artificial neural network

Abstract

With the development of modern technologies, there are more and more complex Visual Analytics (VA) systems. New challenge of the VA field is to analyze complex, incomplete and inconsistent data. Hence, its evaluation is a primordial task aiming to optimize them, although it is a very difficult task. Even little effort has been made to intelligently evaluate VA artefacts, we proposed to assess the visualization environment and user's interaction by mixing of think aloud protocol analysis and eye tracking that provides an efficient information source where visualization can be used to derive measures about a set of metrics employed to analyze the user performance. As soon as we obtain values from one of these metrics, we use a Neuro-Fuzzy method to intelligently interpret these measures by combining; (1) Fuzzy logic to deal with inaccuracies and uncertainty problems during the evaluation process using the concept of linguistic variables, with (2) Neural network to solve the continuous changes problem in assessment environments with delivery of adaptive learning content. The evaluation results performed by intelligence artificial (IA) are more realistic and accurate than those of traditional methods whereas the lack of ambiguity and uncertainty problems in subjective evaluation.