A comprehensive appraisal of perceptual visual complexity analysis methods in GUI design

Highlights

• Perceptual visual complexity is a measurable concept and it could be imitated by machines,

• Visual complexity analysis without human bias can be performed for graphical user interfaces,

• In the near future, machines will be able to control the design of user interfaces.

Abstract

Graphical or Visual User Interface (GUI) is recognized as one of the most important application components for safety critical and business oriented software systems. It is highly advantageous for GUI designers and application developers to analyze the visual complexity of a GUI and predict users’ perception and judgment during the design phase. Although in recent years, various methods have been developed for visual complexity analysis, these have not been widely used due to applicability, practicality and validity issues. In this respect, we have conducted a comprehensive review of studies and methods in visual complexity analysis. After identifying and analyzing 85 research studies, we grouped the visual complexity analysis methods and accordingly a taxonomy is presented. Furthermore, conceptual comparison of the methods is given and gap analysis as well as possible future directions are provided. According to the our findings, major gaps for each visual complexity analysis method may be stated as follows: 1) In metric-model based methods, there is a lack of information about the suitability of the metric-model created for analysis, since the extent to which each metric contributes to visual complexity analysis is still not known exactly. 2) In heuristic- based methods, the extracted rule set is not yet extendable enough beyond the use for specific GUIs. 3) While the visual complexity analysis could be considered as a kind of computer vision task, there exist limited studies that does so. Therefore, generalizable solutions based on machine learning techniques seem to be a promising research direction to develop efficient approaches.

Introduction

Graphical or Visual User Interface (GUI) is recognized as one of the most important application components for safety-critical and business-oriented software systems. GUI is not only seen as a kind of bridge between user and application but also as the showcase of the application for its acceptance. Almost in all application areas from web to mobile and on-platform, visual GUI design has an important role for usability. The main purposes of visual interface design are to provide necessary and related information to the user in a fast and an accurate manner, to meet user’s perceptual needs with different profiles and to help the user work effectively especially in mission critical systems.

Users may interact with many different user interfaces for various needs. These might be GUIs for desktop applications, web user interfaces (WUI) for web applications and mobile user interfaces (MUI) for mobile applications. Desktop GUIs offer a more stable visuality compared to WUIs and MUIs, on the other hand the visuality of WUIs and MUIs are more dynamic and self-explanatory. In particular, MUIs have to use smaller screens effectively and efficiently due to the device constraints. In addition, user interfaces must be adapted to meet different design criteria and user interactions in terms of the services they offer. In order to make a positive impression on the user, the user interfaces that perform different tasks within an application such as login, registration, menu, search and settings should be as clear as the main UI. As an example, the user should be able to see and examine the search results without any distraction by irrelevant UI components while on the search UI. Similarly, all options reside in the menu UI should be clearly presented and easily distinguishable from the background. For safety-critical applications such as medical imaging, C4ISR and so on, it is important to be provided with an appropriate and understandable UIs, where the attention of the operator could be focused on the desired operation. Otherwise, possible errors arising from complicated UIs in such applications that have a direct impact on human life can have catastrophic consequences. In order to appreciate the visual complexity variations according to application settings, examples of representative UIs from some popular software applications for web, mobile and desktop environments are given in Fig. 1, Fig. 2 and Fig. 3, respectively.

User interfaces, which can be adapted within a short time by the user, have an influence on the effectiveness of the application. Therefore, GUI must be perceptually in symmetry with cognitive and perceptual abilities of the user. A smooth interaction between the user and the application leads to increase user satisfaction and user expectations [15], [59].

The GUI visual complexity and aesthetics might be influenced by various factors. Most influential factors are: (1) the graphical design of the GUI, (2) the amount and variety of GUI components, and (3) the semantic relationship between GUI components [75], [16], [31], [81]. On the other hand, usability and complexity concepts may be attributed to subjective human perception. A GUI or a visualization that appears to be simple and usable for one user might appear complicated for another user. Therefore, these two concepts are difficult to be measured precisely. However, it is necessary to predict the usability and complexity to reach a certain quality criteria in terms of standards and customer requirements. The visual complexity analysis and evaluation should be conducted not only after the user interface design is complete, but also during the design process [5], [64], [41]. In this way, the design time and quality can be saved.

Although in recent years, various methods have been developed for visual complexity analysis, these have not been widely used due to applicability, practicality and validity issues. Therefore, we have conducted a comprehensive review of studies and methods in visual complexity analysis. After identifying and analyzing 85 research studies, we reviewed and grouped the visual complexity analysis methods and accordingly a taxonomy is presented. Furthermore, a comparison of the methods is also given and gap analysis as well as possible future directions are provided.

The main contributions of the paper can be listed as:

• Giving a detailed explanation about visual complexity and the identified characteristics that affect visual complexity perception,

• Presenting a taxonomy about visual complexity analysis methods, elaborating on each and discussing the advantages and disadvantages of each,

• Providing alternative approaches for future directions considering new approaches.

The rest of the paper is as follows: section II reviews the analysis of perceptual visual complexity concept. In section III, a systematic overview of GUI visual complexity analysis methods is given. An evaluation of the approaches is provided in section IV. Discussion and future directions are presented in section V. The potential threats to validity of this study is presented in section VI, followed by key findings and conclusions in section VII.