Gaze offset in eye tracking studies biases AOI-based metrics. In UML studies, classes are often operationalized as AOIs and enlarged to mitigate offset, threatening reproducibility. We propose ClassShift, an ensemble-based post-hoc offset-compensation algorithm that improves fixation-to-class assignment in UML class diagrams without enlarging AOIs. We evaluate ClassShift on synthetic data with ground-truth fixation-to-class labels (9332 scenarios) and real eye tracking data derived from 32 participants. Evaluation on synthetic data shows a 19% Macro-F1 improvement relative to baseline containment-based labeling. In the 9.6% of scenarios where ClassShift underperforms, losses are small (median Δ Macro-F1 = -0.033). On real data, visual inspection suggests plausible fixation-to-class assignments. Participant self-reports further indicate which classes were reported as having been inspected particularly long. After applying ClassShift, relative dwell time for these classes increases significantly over the uncorrected data (p =.028, r = 0.50). Together, these results suggest ClassShift improves eye tracking analyses under systematic offset.

Modern education through Learning Management Systems (LMSs) provides learners with personalized learning paths. This is achieved by first querying the learning style according to the theory of Felder and Silverman to recommend suitable learning content. However, a rigid learning style representation is lacking of adaptability to the learners' choices. Therefore, the present study evaluates the idea of providing adaption to the representation of learning styles by using Hidden Markov Models (HMMs). Thus, data is collected from participants out of the Higher Education Area. The Index of Learning Styles questionnaire is used to obtain the learning style based on the theory of Felder and Silverman. Also, a questionnaire that asks the respondents to create a preferred learning path with the sequence length of nine learning elements is provided. From the given data, we initially evaluate the probability relationships between learning styles and learning elements. Then, we use the Viterbi algorithm in HMMs to identify alterations in learning styles from the provided learning paths. The alignment is then quantified by introducing a metric called support value. The findings imply that our concept can be used to adapt the learning style based on the user's real choice of learning elements. Thus, the proposed model also offers a way to integrate a feedback loop within LMSs leading to an improvement of learning path recommendation algorithms.

Most recently, the number of students dropping out of universities or higher education institutions increased dramatically. This might be partly because of the students’ limited capability exploring their own learning paths within a certain course. The introduction of adaptive learning management systems could be a potential solution to this issue. Based on individual’s learning styles, these systems recommend customised and tailored learning paths. These learning styles are commonly identified using questionnaires and learning analytics, but both methods are prone to errors. While questionnaires potentially give superficial answers due to e.g. time constraints, Learning Analytics cannot mirror offline behaviour. This paper proposes an alternative to classify the learning style of individuals through the integration and combination of eye tracking and artificial intelligence algorithms. The eye movement data, which is collected in a study including more than 100 participants, is processed with different methodolgies such as data scaling and subsequently classified using various models ranging from Logistic Regression to Neural Network. Moreover, this experiment setting discovers the interplay between preprocessing and classification techniques based on complex eye tracking metrics in order to determine the most promising solutions for learning style identification. Ultimately, this comprehensive analysis not only enables the understanding of individuals’ subconscious processes, but could also lead to improved educational outcomes.

This study uses holistic models of image perception originating from radiology and psychology to analyze and interpret eye movements during code reviews in the C++ programming language. The study design is based on former experiments, but is supplemented by approaches from expertise research. The study utilizes a sample of 34 subjects whose eye movements are recorded by a Tobii Pro Spectrum 600 Hz. The results show that the holistic models of image perception are suitable for application to source code. In addition, it can be observed that the code reviews are conducted in phases, which are characterized by certain strategies (e.g. scan, error detection,...). Furthermore, experience-related differences can be detected between experts and novices, which emphasize that experts use elaborate strategies and have a comparatively better ability to collect and process information from source code.

In the course Eye Tracking in Engineering Science, students develop a variety of skills. These competencies include both, theoretical knowledge and practical experience: in empirical research (i.e., the design of experiments and the writing of research papers), in eye tracking technology, and in applying computer science and statistics for data analysis. This competence building is done with a teaching concept based on pair-teaching and research-based learning - described in detail in the present article.