Exploring Gaze Data for Determining User Learning with an Interactive Simulation

Kardan, Samad; Conati, Cristina

doi:10.1007/978-3-642-31454-4_11

Cited by 46 publications

(32 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Interestingly, LR consistently achieved the highest accuracies compared to other machine learning models, such as SVM or Decision Trees. Although we do not have an intuitive explanation for this finding, several other works have similarly found LR to perform well with eye gaze data [Kardan and Conati 2012;Bondareva et al 2013].…”

Section: Summary Of Results and Discussionsupporting

confidence: 67%

“…In particular, these analyses have typically consisted of offline processes that require further human analysis (e.g., manually analyzing eye gaze coordinate plots [Iqbal and Bailey 2004]). In terms of actually using raw eye-tracking data for real-time prediction, most research has so far focused on identifying the user's cognitive processes while she is performing nonvisualization activities, such as during exploratory e-learning [Kardan and Conati 2012;Conati and Merten 2007], quizzes [Courtemanche et al 2011], simple puzzle games [Eivazi and Bednarik 2011], or information search tasks (e.g., word search) [Simola et al 2008]. By contrast, our gaze-based work focuses on information visualization, where a user's main activity is to perform simple visualization lookup and comparison tasks.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Inferring Visualization Task Properties, User Performance, and User Cognitive Abilities from Eye Gaze Data

Steichen

Conati

Carenini

2014

ACM Trans. Interact. Intell. Syst.

Self Cite

View full text Add to dashboard Cite

Information visualization systems have traditionally followed a one-size-fits-all model, typically ignoring an individual user's needs, abilities, and preferences. However, recent research has indicated that visualization performance could be improved by adapting aspects of the visualization to the individual user. To this end, this article presents research aimed at supporting the design of novel user-adaptive visualization systems. In particular, we discuss results on using information on user eye gaze patterns while interacting with a given visualization to predict properties of the user's visualization task; the user's performance (in terms of predicted task completion time); and the user's individual cognitive abilities, such as perceptual speed, visual working memory, and verbal working memory. We provide a detailed analysis of different eye gaze feature sets, as well as over-time accuracies. We show that these predictions are significantly better than a baseline classifier even during the early stages of visualization usage. These findings are then discussed with a view to designing visualization systems that can adapt to the individual user in real time.

show abstract

Section: Summary Of Results and Discussionsupporting

confidence: 67%

Section: Related Workmentioning

confidence: 99%

Inferring Visualization Task Properties, User Performance, and User Cognitive Abilities from Eye Gaze Data

Steichen

Conati

Carenini

2014

ACM Trans. Interact. Intell. Syst.

Self Cite

View full text Add to dashboard Cite

show abstract

“…Next, association rule mining is applied to each cluster to extract its common behavior patterns, i.e., rules in form of X c, where X is a set of feature-value pairs and c is the predicted cluster for the data points to which X applies (Table 1 shows samples of these rules, which will be further explained in the next section). We built the user model for the adaptive CSP applet by applying Behavior Discovery to a dataset of 110 users obtained from two previous studies on this simulation [12,13]. From this dataset, the Behavior Discovery generated two clusters of users that achieved significantly different learning levels, labeled as High Learning Gain (HLG) and Low Learning Gain (LLG) groups from now on.…”

Section: Modeling Student Learning In the Csp Appletmentioning

confidence: 99%

Providing Adaptive Support in an Interactive Simulation for Learning

Kardan

Conati

2015

Proceedings of the 33rd Annual ACM Conference on Human Factors in Computing Systems

Self Cite

View full text Add to dashboard Cite

Recent rise of Massive Open Online Courses (MOOCs) with unlimited participants, makes employing learning tools such as interactive simulations all but inevitable. Interactive simulations give students the opportunity to experiment with concrete examples and develop better understanding of concepts they have learned. However, some students do not learn well from this relatively unstructured form of interaction, suggesting the provision of adaptive support as a way to address this issue. This paper presents a formal evaluation of this approach. We describe the process of designing an intervention delivery mechanism for adding adaptive support to an exploratory interactive simulation. The experimental evaluation of the adaptive version of the simulation indicates that the adaptive support provided to students significantly improved their learning performance. Quantitative and qualitative evaluations of users' acceptance of the system are generally positive but pinpoint areas for improvement.

show abstract

“…For example, Qu & Johnson [21] showed that user gaze behaviors can help predict users' motivation during interaction with an intelligent tutoring system. Kardan et al [14] and Bondareva et al [4] showed that eye tracking data can be used to predict student learning with two different educational environments, and that this prediction can be performed early enough to possibly provide adaptive interventions that can foster learning. D'Mello et al [8] evaluated an intelligent tutoring system that both detected and reacted to students' lack of attention based on gaze patterns.…”

Section: Eye-tracking In User Modeling For Adaptive Systemsmentioning

confidence: 98%

“…Following the approach in [14] IUI 2014 • Learning and Skills February 24-27, 2014, Haifa, Israel and [24], we generated a large set of eye-tracking features by calculating statistics upon basic eye-tracking measures (see Table 1 & Table 2). Distance between the two fixations delimiting the saccade (d in Figure 4) Relative Saccade Angles…”

Section: Eye Tracking Measures and Featuresmentioning

confidence: 99%

Towards facilitating user skill acquisition

Toker

Steichen

Gingerich

et al. 2014

Proceedings of the 19th International Conference on Intelligent User Interfaces

Self Cite

View full text Add to dashboard Cite

A key challenge for information visualization designers lies in developing systems that best support users in terms of their individual abilities, needs, and preferences. However, most visualizations require users to first gather a certain set of skills before they can efficiently process the displayed information. This paper presents a first step towards designing visualizations that provide personalized support in order to ease the so-called 'learning curve' during a user's skill acquisition phase. We present prediction models, trained on users' gaze data, that can identify if users are still in the skill acquisition phase or if they have gained the necessary abilities. The paper first reveals that users exhibit the learning curve even during the usage of simple information visualizations, and then shows that we can generate reasonably accurate predictions about a user's skill acquisition using solely their eye gaze behavior.

show abstract

Exploring Gaze Data for Determining User Learning with an Interactive Simulation

Cited by 46 publications

References 16 publications

Inferring Visualization Task Properties, User Performance, and User Cognitive Abilities from Eye Gaze Data

Inferring Visualization Task Properties, User Performance, and User Cognitive Abilities from Eye Gaze Data

Providing Adaptive Support in an Interactive Simulation for Learning

Towards facilitating user skill acquisition

Contact Info

Product

Resources

About