In de afgelopen jaren is het steeds makkelijker geworden om eye tracking data te verkrijgen. Om deze data te kunnen gebruiken in onderzoek is het nodig om de, vaak onoverzichtelijke, data goed te visualiseren. Tijdens een cursus op de TU/e heb ik samen met een aantal mede studenten een web-based tool gemaakt waarop data geupload kan worden en waar deze data gemakkelijk en op een begrijpelijke manier wordt gevisualiseerd. Hieronder is de abstract en introductie te lezen. Onder aan de pagina is ons hele research paper te vinden.
Abstract. Eye tracking data has become easier to generate in the last years. However, this data type is large in size and highly complex to analyze without any sort of visualization. In order to visualize this type of data, one nearly always has to install software. EyeVisualize was created to offer a fast, web-based application in which data can be uploaded and easily visualized. This application provides users with four different visualization techniques for eye tracking data (formatted in a specific way): a heat map, a scan path plot, a scarf plot, and a circular heat map. Each visualization technique separately has its benefits and drawbacks, however, combined they can offer a full picture of the data, which can lead to novel insights.
Introduction
Most researchers enjoy doing their experiments and research on the topic of their choice. Yet, the majority of this group does not like the steps afterward as much. These steps include analyzing the data and figuring out different ways to inspect the data to find a conclusion. This is not an easy task, because research usually generates a vast body of data. Most research fields obtain results consisting of raw data as lines of code containing a massive amount of numbers. Using only the naked eye, generally, no sense can be made from these results. Thus, visualization is a necessary step in the process of understanding research findings and interpreting this data correctly. Visualization techniques are there to help researchers transform the data into a clear, sensible, and understandable format. This allows researchers to gain new insights into the data. Furthermore, the use of multiple visualizations could even cause the researcher to find aspects in the data set that were not expected beforehand.
One field that suffers from data that is beyond comprehension at first glance is the field of eye tracking. The use of eye tracking is a long-standing phenomenon in psychology (Smith,1960), as it provides objective data for researchers typically investigating subjective experiences or phenomena. Eye tracking data has become less difficult to collect in recent years; in the past, a complicated device in a computer monitor was necessary to collect reliable data. As technology is continuously evolving at a rapid pace, the technology used for tracking eye movement has also advanced into more flexible devices (Niehorster, 2020; Sugarman, 2016). These new devices make it easier for researchers to utilize eye tracking to gather new insights (Mele, 2012). Despite the continuously developing technologies, eye tracking data remains complex, vague, and vast in size.
These complex eye tracking data sets provide the researcher with objective measures, which are not easy to interpret. A resulting data file from an eye tracking session consists of hundreds or even thousands of lines of code, as eye movements happen extremely fast; visualization of the data is key to understanding it. Numerous visualization techniques have been produced, each one having a different focus (Blascheck, 2017). To visualize eye tracking data in one or more styles, one must nearly always install specific software. This creates a barrier for the researcher to visualize their data, so a web-based platform for visualizing data would be the solution. In this paper, we describe a web-based platform that can transform eye tracking data into four different visualizations. Each visualization provides the user with insights that are impossible to generate from solely inspecting the raw data. Each visualization represents the data differently, which can lead to a variety of possible findings and conclusions. To enhance the usability of the visualizations they have been created to be interactive, as opposed to static. Those interactions allow the researcher to change some aspects of the visualizations according to the demands of the research or his or her preferences. These interactions are based on the report from Yi et al. (Yi, 2007), who described seven categories of interactions in visualizations, which are organized around the users' intent.
Do you want to read the whole article? It is available on ResearchGate. It is also possible to read the publication further from the introduction down below.