1 Introduction

Visualizing multi-dimensional data is regarded as a difficult task for novice users, especially, expressing as a cognitively optimized form. There are several reasons for that, the first reason indicates that most novice users feel it is difficult to crease a dataset for data visualization. Second, novice users could not find out or determine the effective visual effects which to be applied [1]. Above issues have been discussed several years in the field of data visualization, thus, many researchers conducted various studies to solve above issues and to help novice users their data visualization.

2 Related Work

Wongsuphasawat et al. [2] developed a kind of data visualization system, which called “Voyager”. It is supporting exploratory visual analysis method by using the Vega specification. In addition, the system recommends several options in both data field selection and visual encoding [2, 3]. Hence, users are able to visualize their data effectively with using the system, in which recommends data fields and visual encoding. In addition, the users could understand the trend of the data easily, so they can catch up insight well from the data.

However, in order to use the system provided by Wongsuphasawat et al. [2], the users who use the system are asked to have higher levels of comprehension of data visualization. Furthermore, several problems are still remaining. First, the users could not recognize how each option of the system influences on the effectiveness of data visualization. Second, the users could not perceive the potential problems of data visualization provided by the system in the aspect of cognitive engineering. Finally, the users could not know which part needs to be supplemented in the data visualization to choose the best decision making. In order to overcome above issues, we developed a system including the Interactive Guide Model that recommends how to optimize data visualization so that it helps users can choose proper decision. With this perspective, we try to stare at what kinds of knowledge users need to convey in visualizing data and how such knowledge should be communicated, and finally, we study how these systems can contribute to make the best decision on data visualization.

3 Multi-dimensional Data Visualization Guideline

In the initial stage, we conducted a work domain analysis to define components of visualization recommendation model. The results of work domain analysis described that a systematic visualization guide defined as rule-based that is required for the data visualization general principle [4]. To develop general visualization principle guidelines, we analyzed the papers published at the InfoVis conference, which is considered to be the most influential in the field of data visualization. Total 684 papers published during the period (1995–2017) have been analyzed, then 59 (8.6%) papers regarded as relevant papers to our purposes. Thus, chosen papers have been discussed in order to be used as fundamental visualization guidelines [5, 6]. These papers presented guidelines using keywords such as, consideration, design, framework, guideline, guidance, implication, lessons learned and taxonomy. The results of the study have been investigated into 9 categories, which could be divided into 3 groups according to each step in the process of data visualization to the user (Table 1).

Table 1. Visualization guidelines classified into 3 groups, 9 categories
Full size table

We have organized the contents into 3 types of templates so that users who will use this guide can understand more easily and clearly what our findings are intended to convey (Fig. 1).

Fig. 1.
figure 1

Sample guideline of data visualization general principle

Full size image

4 Interactive Recommendation Model for Data Visualization

We concentrated on “Data Operation” and “Visual Encoding” among 4 steps of the data visualization process suggested by Munzner in order to communicate the organized data visualization guidelines to users most effectively [7]. In addition, we added an interactive guide panel to the right of the existing Voyager System (Fig. 2). In that panel, the optimal tips for visualizing their data will be presented to the users. In the case of cognitive problems might be occurring, the system suggests several problems through pop-up style and the system also describes the reasons of problems and corresponding solutions to help users to know how to improve it. Through this process, although the users do not have a deep knowledge of the data visualization, they are given three detailed information to the user so that they can optimize their visualizations as intended. Firstly, the system suggests the advantage or disadvantages of choosing each option. Secondly, the system presents the preview of choosing each option. Lastly, the system provides an animated transition that indicates which components are changed from the current visualization to be an improved visualization. Particularly, Heer and Robertson insisted that animated transition could influence significantly on the users’ understanding of the difference between AS-IS and TO-BE [8].

Fig. 2.
figure 2

Sample guideline of data visualization general principle

Full size image

5 Conclusion and Future Work

In this paper, we proposed a new interaction method for optimizing Data Visualization. Although this study designed Algorithm through a combination of findings published in InfoVis, It could also further develop into user data-driven data visualization by analyzing results of user’s data visualization and creating algorithms. Finally, from a user’s perspective, we will continue to work on these studies that they will be able to quantitatively and qualitative evaluate how well they can design their desired Data visualizations to further explore the highly complex elements that make up the Data Visualization UX.