Abstraction is considered as a conceptual process, whose outcome stands for the corresponding subordinate concepts as a whole. This representative notion allows users to facilitate comprehensive understanding and better memorability of the relevant complex knowledge in a more hierarchical or structured fashion. The concept of abstraction is especially significant when the information space is large and when it comes to real-world applications. The Tube Map of London Underground is a classical map example that removes unnecessary detail but retain sufficient information for better usability. The map successfully simplified the geometry of the transportation structure, and facilitate users to effectively perform their tasks, which often includes (1) which route provide the shortest path in terms of distance or price, (2) how many stops still remain until the destination, (3) where to transfer to another line, and so on. Other than cartography, abstraction is powerful in many application domains, such as mathematics, biology, physics, and etc., because experts can focus on vital elements and think through the problem in a more abstract level.

Similarly, in data visualization, abstraction plays an important role because the outcome of the abstraction process need to reflect what an expert has in mind and the transition process should support the comprehension of the high-level concept in various aspects. However, in order to visualize essential aspects of the data, we need dedicated mechanisms that abstract the unnecessary detail to allow the viewer of a visualization to focus on the important elements. Viola and Isenberg have done an intensive investigation on this topic and confirmed that the crucial problem in this context is that it is impossible to know what is important and what is not in a general way — importance changes based on the research question, on the application domain, on the data size, on the user,on the specific situation, etc. Visualization techniques, therefore are expected to support a dynamic change of data’s visual abstraction to reflect these contextual changes.

Beyond 2D representation such as metro maps, Rauteket al. and other researchers have established an understanding of visual abstraction for spatial 3D data to classify low-level and high-level visual abstraction techniques, while the support of reasoning and insight communication for abstraction techniques is still a fun-damental challenge that remains today. Recently, Viola and Isenberg have intensively explored the concept of abstraction as it is used in visualization. They realized that researchers so far have used the concept of abstraction largely by intuition without a precise meaning, and thus initialized the pioneering discussion on theoretical foundations of abstraction in visualization. This lack of specificity left questions on the charac-teristics of abstraction, its variants, its control, or its ultimate potential for visualization and, in particular, illustrative visualization mostly unanswered. After this investigation, several open questions are still awaiting to be answered, while it requires not only visualization researchers but also domain experts in order to formulate a research agenda for the practical usage of abstraction for the visual representation and exploration of data.

With the growing of data complexity, the strong needs for abstraction techniques are demanding. Although there have been a few research studied along this line, the theory, technology, and applications have so far less integrated in terms of the stability and usability of abstraction models for visual analytics. These challenges concern approach scalability, the applicability of the models in application areas, as well as the technology of the environment in which the abstraction is performed. In summary, the theory and application of abstraction in visualization poses a research challenge not only due to the complex nature of the data, but also its dynamics and semantics. This Shonan meeting will be the pioneering seminar that initialize gathering researchers from visualization, information theory, and applied science.

Abstraction is inherently multi-disciplinary since similar ideas and concept can be applied to various such as cartography, biology, mathematics, physics, and many. In this seminar, our primary goal is to revisit and strengthen the fundamental theory and applications of abstraction in visualization techniques and summarize future challenges as a guidance for novice followers. This effort cannot be easily done through the discussion during visualization conferences, but requires the exchanges and discussion between visualization researchers and domain experts. Therefore, our second goal is to gather researchers and initialize the pioneering networking event to establish the connections between researchers sharing their abstraction ideas, particularly in biology and cartography. Thus, the outcome of seminar would be broadening the abstraction concept through discussion and move forward to initialize creative collaborations for solving real-world abstraction problems in visualization. We would ideally aim for the content to be published as a book or special journal issue.

The overarching goal of abstraction technology is to understand how and what new integrated abstraction formats, visualization approaches, and interaction technologies can be used to create a supportive environment for data exploration and analysis, theoretically and practically. A well study on the abstraction process as well as the corresponding results is targeted. These potentially require consistent domain knowledge, visualization models, and standard techniques across disciplines. Some of the main research questions are summarized as:

Theoretical Formalization of the Abstraction Concept as it Relates to Visualization: The term abstract is an antonym of concrete or tangible, resulting in an inherent difficulty to describe it. Viola and Isenberg have defined an abstraction is a transformation which preserves one or more key concepts and removes detail. A discussion with researchers from different background knowledge will formalize the developed theory.

(Semi-)Automated Visualization Design: Although automating the use of visual abstractions for the purpose of generating effective and expressive visualization designs has been the goal of visualization for already some decades, nonetheless, users need to first understand the characteristics associated with these abstractions.

Superimposed Visualization for Multi-Attribute Data: The problem with understanding data often relates to the complexity of the analyzed phenomenon and the fact that data simultaneously captures multiple attributes that need to be understood at the same time. Overlaying more information does not provide better readability due to the fact that they compete for the same screen estate. A guidance on this decision making is expected to be established.

Temporal Changes and Interpretation Ambiguity: A computational model is used in many real-world situations to simulate possible developments of a certain process. Visualization is a promising approach to ana-lyze such data, but straightforward visual playback will not uncover hidden causal relationships. An abstraction of the simulated process itself will thus be a crucial analytical instrument.

Addressing Large Scale Changes: In most real world applications, information need to be studied on vastly different scales. We hope to develop visual representations that support the scale traversal through abstractions that can cover multiple orders of magnitudes. Current visual abstraction approaches, however, only cover relatively small scale ranges.

Encoding Uncertainty: Data uncertainty in visualization is often conveyed through techniques developed in descriptive statistics. One challenge that visualization community attempts to address is how to convey the distribution or uncertainty visually for dimensions higher than two—even 3D data representation struggles with how to convey data and uncertainty simultaneously.

Perceptional Recognizability and Cognitive Scalability: Advanced visualization normally requires users’ domain knowledge for better utilization. Once new abstraction models have been developed, what paradigms are potentially enabled by these modalities? Can domain experts or the public understand the new representation?How do we evaluate them?

Generic Platforms for Supporting Analysis through Abstraction Techniques: Until recently, there is a wide range of different design frameworks, development platforms and cross-platform tools based on the application domain of the abstraction techniques. Nonetheless, a specific design or focus on the involvement of abstraction techniques should be investigated.