The methodology of conducting an experimental study of the perception of the graphical interface using eye-tracking technology

Kheyfits Antonina Evgenievna Assistant, Higher School of Design and Architecture, Peter the Great St. Petersburg Polytechnic University 195251, Russia, Leningradskaya oblast', g. Saint Petersburg, ul. Politekhnicheskaya, 29 antoni.t-h@mail.ru Yanchus Viktor Edmundasovich PhD in Technical Science Associate Professor, Higher School of Design and Architecture, Peter the Great St. Petersburg Polytechnic University 195251, Russia, Leningradskaya oblast', g. Saint Petersburg, ul. Politekhnicheskaya, 29 victorimop@mail.ru

Borevich Ekaterina Vladislavovna Assistant, Higher School of Design and Architecture, Peter the Great St. Petersburg Polytechnic University 195251, Russia, Leningradskaya oblast', g. Saint Petersburg, ul. Politekhnicheskaya, 29 plasma5210@mail.ru

Introduction

In the last decade, there has been an increasing interest in the design of complex technical and biomechanical systems in which a human operator interacts with computer systems. One of the main elements of such systems is the user interface, through which a person communicates with a computer. Great attention is paid to the ergonomic design of new forms of interfaces that ensure effective adaptation of operators to natural and artificial learning environments ^[1]. Ensuring effective interaction of operators with a computer system is an urgent problem, despite the introduction of simulators and computer training technologies ^[2].

To analyze the effectiveness, productivity, and aesthetic appeal of the developed user interfaces, the method of oculography is widely used, which is implemented in practice with the help of AI-tracking software and hardware complexes ^[3].

Based on the above, we can talk about the relevance of the task of studying the conditions for the operator's perception of graphic information in the functioning of a human-computer interaction system using AI-tracking technologies.

When solving the problems of engineering design of user interfaces, the quality of any interface is ultimately determined by the effectiveness of interaction between one person and one system ^[4]. One of the most striking examples of human-computer interaction is the use of an interface when controlling an airplane in the pilot – airplane system. The works of scientists substantiate the concept of "multiplication of possibilities" and a symbiotic approach to the integration of the pilot and the aircraft, according to which, the integration of the pilot with the aircraft is symbiotic. As a result, a new technobiotic self-organizing unity emerges, behaving like a combat unit focused on achieving decisive superiority over the enemy ^[5]. The issue of designing an intelligent interface that helps a person adapt to a remote object management system is relevant in the context of solving effective management problems ^[6,7]. The prototype of this experiment is a situation in which the task of providing up-to-date information is solved.

Fig. 1. Interface immersion conditions

The purpose of the work is to develop a methodology for conducting experimental studies of the perception of a graphical interface depending on the conditions of immersiveness (Fig. 1). The paper considers one of the conditions of immersiveness – infographics.

The purpose of the work involves solving the following tasks:

- development of dashboards (a graphic element containing a diagram) displaying the current value of the parameter and a number of its previous values;

- development of incentive material with random arrangement of dashboards;

- creation of an experiment project in a specialized AI-tracking software and hardware complex;

- approbation of the methodology and statistical processing of the experimental data obtained.

Theoretical model

Information panels are integral elements of the graphical interface, since they are a visual part of information management systems and display quantitative and qualitative information about the state of the observed object. The expansion of the boundaries of work between the operator and the equipment is the evolution of intelligent technical systems ^[8].

A series of experiments using AI-tracking technology was carried out in the laboratory of human-computer interaction of Peter the Great Polytechnic University.

An experiment was conducted where the AI-tracking technology was used to analyze the perception of complex graphic images by subjects without a clear semantic load and showed its effectiveness ^[9].

In another experiment, the influence of stylization of graphic images on the perception of graphic information using AI-tracking technologies was studied. The dependence of the time of analysis of the graphic image by the subject on the stylization of the image of the stimulus material was revealed ^[10].

An experimental study was also conducted on the influence of the types and number of linear graphs that can be displayed on one screen at the same time on the user's perception of information. The task of the experiments was to find out how perception changes and what needs to be included in the development for effective learning at work. The subjects were shown linear graphs on the monitor panels with an increase in the number of graphs, a change in size and an increase in the total area occupied by graphs on the screen. The study showed that there is no correlation between the dynamic trend of graphs and the perception of the image in the subjects ^[11].

The article ^[12] indicates that infographics are the optimal tool for displaying the current value of the parameter of the observed object.

Infographics, as the most minimalistic and thoughtful design cluster, does not attract much attention, allowing you to work with information ^[13,14]. In conditions where a person solves an interactive task, the factor of conciseness plays a major role.

Methodology of the experiment

The developed methodology for conducting an experimental study of the perception of the graphical interface of remote control systems for dynamic objects consists of several stages:

- development of a database of dashboards reflecting the current value of the parameter and a number of previous ones;

- development of incentives with random arrangement of dashboards;

- designing an experiment in the specialized experiment center program of the SMIRED 250 AI-tracking software and hardware complex;

- training of subjects (instruction) and conducting an experiment on the hardware and software complex of ai-tracking;

- statistical processing of the experimental data obtained and interpretation of the results.

To carry out a factor analysis of the perception of graphic information, three factors were identified, the influence of which was studied in this work:

- chart type factor;

- dashboard size factor (graphic element containing a diagram);

- the factor of the number of dashboards on the screen.

Three types of diagrams were used to develop the stimulus material: linear, columnar, and figured quantitative.

Dashboards had three sizes:  4*2cm, 3*2.25cm and 2*1.5cm. 720 dashboards were developed in three sizes and with three types of diagrams.

The number of developed incentives is 54 pcs. This amount of stimuli is determined by three factors under study, which take three values. It was decided to make two stimuli with the same values of the studied factors, but with different dashboards.

The distribution of dashboards in the stimulus was carried out in a grid with five columns and five rows, which distributed the monitor into 25 zones. The size of the monitor screen was 54*29 cm. The position of the subject's head was fixed in a certain position using a special design at a distance of 74.4 cm (Fig. 2).

Fig. 2. The layout of the installation screen and the position of the subject in the experiment

A software module has been developed to create incentives with a random arrangement of dashboards. The block diagram of the algorithm of the software module is shown in (Fig. 3). For the software implementation, the open programming language processing was used. At the entrance of the software module, a database of 720 dashboards was developed as part of this work. At the output, we had 54 stimuli for the experiment (Fig. 4-6).

Fig. 3. Block diagram of the program algorithm for the preparation of stimulus material. An example for one type of diagram.

Fig. 4. Example of stimulus material with three dashboards (size – s, type – p)

.

Fig. 5. Example of stimulus material with seven dashboards (size – l, type – s)

Fig. 6. Example of a stimulus material with five dashboards (size – m, type – l)

The subjects in the experiment were tasked to find a graph with the maximum increase in the parameter value and mark the corresponding dashboard by hovering the mouse cursor and pressing the button. After pressing the left mouse button, an automatic transition to the next stimulus took place. All the subjects were instructed about the task before passing the experiment. The time to solve the task was unlimited.

Statistical processing of the results was performed by ANOVA analysis of variance ^[15].

The SMIRED 250 software and hardware complex was used as an experimental setup (Fig. 7) ^[16].

Fig. 7. SMIRED 250 hardware and software complex. Installation and on-screen interface

12 test subjects from students and young teachers of Peter the Great St. Petersburg Polytechnic University were invited to test the developed methodology.

Analysis of results

Statistical processing of the results revealed the statistical significance of the parameters of the template for viewing the stimulus material, depending on the factors:

- size, the size of the dashboard (the factor takes values according to the scale of the dashboards being demonstrated: l – the size of the dashboard is 4x2cm, m – 3x2.25cm, s – 2x1.5 cm);

- type, dashboard type (the value of the factor according to the types of diagrams: l – linear, p – column and s – figured quantitative);

- inf, the number of dashboards in the stimulus (the factor takes values 3, 5 and 7, according to the number of dashboards in the stimulus).

The experiment involved 12 subjects who were students of the Polytechnic University aged 20 to 28 years.

As a result of statistical processing of the experimental results, the statistical significance of the time of viewing stimuli by the subject was revealed depending on the inf factor (Fig.8). The influence of other factors on the parameters of the template for viewing stimulus material, according to statistical processing, was not revealed. However, the number of subjects participating in the experiment is not enough to accept the hypothesis that there is no influence of these factors.

Fig. 8. The density of the distribution of the stimulus viewing time depending on the inf and type factors for different values of the size factor

During the experiment, it was noticed that the subjects experience difficulties in solving the problem:

- more than 75% of the subjects spent more than 10 minutes on the experiment;

- when solving the problem, the subjects made a lot of mistakes (about 40% in some stimuli) (Fig. 9).

Fig. 9. A heat map of the distribution of the subjects' gaze on a stimulus with seven dashboards (a diamond marks the point of choosing the correct answer, in his opinion, the subject).

Conclusions

As a result of the work, a methodology for conducting experimental studies of the perception of the graphical interface using AI-tracking technology has been developed. The developed technique was tested on 12 subjects.

As a result of the conducted experiment on the approbation of the developed methodology, the following was revealed:

1. The factor of the number of dashboards in the stimulus has a statistically significant effect on the parameters of the template for viewing the stimulus material by the subjects.

2. Initial instruction about the proposed task during the experiment is insufficient for the subjects. The subjects have no experience with infographics, which causes them difficulties in solving the problem. Two types of difficulties were identified:

a. incorrect solution of the problem;

b. fatigue of the subjects during the experiment.

The incorrect solution of the experimental problem by the subjects did not allow to reveal the influence of factors of the size and type of dashboards on the parameters of the viewing pattern. To obtain reliable data, it is necessary to reduce the number of errors of the subjects during the experiment. To solve this issue, it is necessary to develop a methodology for preparing the subjects for the passage of this experiment.

Materials for the preparation of subjects should be developed, observing the conditions of comfortable perception of information, its consolidation and preservation. It is assumed that after completing the training, the subjects will analyze infographic information faster and better, and, accordingly, solve the problem proposed in the experiment correctly.