Attentional characteristics of anomaly detection in conceptual modeling

Abstract

We use eye tracking to better understand the attentional characteristics specific to successful error detection in conceptual models. This phase of our multi-step research project describes the visual comportments associated with successful semantic and syntactic error identification and diagnosis. We test our predictions, based on prior studies on visual attention in an error detection task, or studies comparing experts and non-experts in diverse tasks, in a controlled experiment where participants are tasked with detecting and diagnosing errors in 75 BPMN® models. The results suggest that successful error diagnostics are linked with shorter total view time and shorter fixation duration, with a significant difference between semantic and syntactic errors. By identifying the visual attention differences and tendencies associated with successful detection tasks and the investigation of semantic and syntactic errors, we highlight the non-polarity of the ‘scale’ of expertise and allow clear recommendations for curriculum development and training methods.