Interaction for visualization is the interplay between a person and a data interface involving a data-related intent, at least one action from the person and an interface reaction that is perceived as such.
Dimara and Perin’s paper deals with the idea of interaction and tries to define this broadly while catering this concept around interaction. The initial problem is that there is a lack of a unified concept of what interaction means. This is primarily influenced by the lack of the HCi - whom a large body of professionals who deal with interaction design look towards. The HCI currently defines interaction as a causal relationship between humans and data. This definition allows for confusion regarding the application of data visualization.
As technology progresses, there is a bigger need for understanding the recent plethora of data. Keefe ( 2010) points out that data visualization is “interdisciplinary” ( pp3) and therefore a suitable definition for the process is required in order to streamline the visualization process. It is argued that visualization is as important as interaction in order for the end-user to truly absorb the data. Goals for the future of this rapidly expanding need is to broaden the scope of interaction and give users a more freeform manner in which to interact with visualized data. Flexibility is a highlighted goal of creating better interaction for data. In order to understand a method to give users the freedom that is desired, a slimming down of what interaction means is necessary.
Dimara and Perin deploy a method of finding a definition for data visualization. To summarize their methodology, papers linked to interaction were collected, followed by seed papers which reference the original set and then other related papers to the second set. This process included tagging definitions of interaction in this body of research papers, as well as questioning the authors regarding their interpretations of the definitions for interaction. This process resulted in the following findings that often arose in their research that centred around the concept of interaction: definitions, critics, benefits, interaction concepts, interaction pipelines and history. The findings of the research seemed to focus particularly frequently on the idea of the user and data forming a sort of relationship.
The user is situated as an analyst who has a motivation to engage with the data. It is crucial to separate the designer and the user in order to create data that is best designed to interact with the users' specific motivations. Data is defined as intangible information and therefore requires a processing step in order for the interaction to be resultant(Büschel et. al 2018, pp7). The tools to achieve with can include representation, hardware and software. Another important aspect of processing data interaction is understanding that time as interaction occurs in real-time.
Dimara and Perin use their findings to elaborate on what processes are involved in interaction for visualization. These include external and internal entities and external and internal actions. The properties of interaction include granularity to confirm the need for levels of interaction with one visualization and the idea of continuous vs discrete interaction. Continuous and discrete interactions must be implemented in order to create immersion in the data. The final property of interaction is direct vs indirectness. This refers to the spectrum on which interaction can be situated to fit the context of the data visualization.
The findings regarding the benefits and criticism of interaction proved most interesting to me. The answers from researchers highlight the needs for interaction that can better define interaction. This includes the disdain of comparison to representation and the limited flexibility of interaction currently. This lack of flexibility has impactful consequences such as ignoring user goals and the lack of catering for human and technological modalities. This creates less successful visualization and therefore must be carefully considered when designing interaction. This poses the debate of creating task-centric interactions versus human-centric interactions. Initially, I have been aligned with user-centric design. This may be due to the corpus of content that places the user at the centre of the project yet as Dimara and Perin earlier state, there is a specific motivation from the user that initiates interaction. If the driving force of the interaction is the want to complete a specific task, would it not be better to cater towards the task in order to streamline the task completion? Medhi ( 2007, pp5) shows through their research that “users tend to focus on isolated tasks”. This may highlight a need to make the task easier to accomplish yet creating a task-orientated interaction proves to be incredibly rigid and socio-economics that affect humans have a deeper impact on the interaction. For this reason, Medhi argues that one should attempt to create both task-orientated and user-orientated interactions yet the process should begin by understanding the limitations of the user.
This supports Dimara and Perin’s observation that the current visualization view promotes a dialogue between the user and data as confining, in order to make the interaction impactful, the dialogue space must be flexible.
This means creating a form of interaction that can accommodate a number of requirements and accommodations for the user. By designing using this approach, one is creating flexible and more open-ended data visualizations. The limitations of this interaction are rooted in the HCi definitions of interaction. The HCI’s developed theory of interaction includes aspects that characterize the interaction. These include dialogue regarding cognitive interaction between data and user, the transmission of information, controls, tool usage, optimal behaviour and embodiment. There is a need to broaden the scope of both user profiling and user intent from these definitions from the HCI, as well as what good interaction means. Intent goes further than the transmission of information and therefore successful interaction can provide the user with a use for the acquired information. The use of user personas in the UX space results in dichotomies that are often reductant and therefore stagnant. This creates a flow of interaction that feels disruptive( Elmqvist et. al 2011, pp335) It must be understood that visual literacy should not be given and therefore persona research must be broadened. The definition of good interaction must be rectified to accommodate more criteria rather than merely a dialogue occurring. There is a clear difference between the HCI definition of interaction and its use for interaction.
The comparison of the lack of definition from the HCi and research has resulted in a new definition for interaction with data visualization. This includes visualization interaction that is subsequent to interaction systems and the dialogue with these systems being resultant of human modalities. The definition also includes essential components of interaction that must be considered which include people, data interfaces, actions and reactions and perceived actions. This definition allows for the development of more flexible interaction by creating more flexible interfaces to support border needs. Flexibility allows for the definition t be actualized.