This essay will explore the relationship between the ideas of representation, data visualization and data art. This essay will use the context of Kidd, Rosier and Viegas and Wattenberg as the basis for the analysis. This essay will demonstrate the need for adherence to representation and artistic theory when handling data. The key links between representation, data and artistic practice will be discussed through the lens of four mutual characteristics: Perception, Reductionism and Contextualisation. Perception will aim to illustrate the importance of semiotics, reductionism will elaborate on the process of creating representations and visualisations and contextualisation will give insight into the meaning-creation behind representations. This essay will further discuss the ethical implications of representation within data and artistic practice.
Perception
Firstly, it is important to initially define representation as representation fundamentally encompasses the act of creating visual form. According to Kidd (2015: 19), representation allows for the construction of our realities and representation is a lens through which meaning is generated. The theory of representation can be simplified to Stuart Hall’s semiotic idea of the signified and signifier. The signified is the thing that is being represented and the signifier is a symbol that is supposed to stand in place of the signified. This can refer to written words being signifiers for spoken language (signified) (Hall, 1985: 101). The act of representing allows for meaning to become imparted on the signifier and subsequently become perceived by the human, as de Saussure suggests that the human species experiences cognitive recognition through the mediation of representations. (2011: 103). Yi et.al (2007) explains that data visualisation aims to represent a larger set of data in a manner that results in rapid cognitive recognition and should encapsulate the correlation between the symbol of the data and the actual data. Hall (2020: 92) further explains that there are vast systems of representation, in which particular symbolic conventions are applied. Data visualisation and artistic practice can therefore be considered a system of representation as these fields aim to depict a set of data or emotional responses through other visual mediums. In this essay, the term representation will be treated as a parental facet of data visualisation and artistic practice as these practices aim to create a symbol from which meaning can be derived.
Kidd suggests that representation has an impact on every sphere of human existence in a manner in which production is affected by representation. As it is now understood, representation allows for meaning. Both Hall and de Saussure are categorised as constructionists and therefore highlight the importance of the act of constructing meaning. In order to understand the meaning generation of representation, we must first understand the need to initially reduce an object to result in an impactful representation.
Reductionism
The act of creating a signifier often becomes a staging process in which the signified becomes essentialized to create the signified (Evans & Hall, 1999: 6). This process is deployed in order to cause rapid analysis of broader ideas. The idea of creating a meaning that is quickly interpreted is the foundation of data visualisation as Li (2020: 30) explains that data visualisation depends on the cognitive ability for visual communication. Artistic practice too aims to achieve a sense of reduction (Cowie et.al 1993: 4). Cowie et.al suggests that artistic practice aims to package a broader experience in a lightweight visual manner. The act becomes a tool for stimulating emotional reflection, according to Roeser (2016: 127). This, therefore, depicts the intersection of reductionism in these separate spheres.
This simplification of data and experience is parallel to Kidd's interpretation of the simplification of the signified to create the signifier. Kidd explains that the visualization of data is “necessarily reductive” (2015: 31). The act of separating the signified and signifier results in the creation of new information and therefore the manner of separation must be considered. While this is true, this act of separation conversely generates meaning from the lack of it. An oversimplification of an idea can impart further connotations. The idea of connotation and denotation is vital when approaching the act of representation. Barthes suggests that there is a denotative and connotative value to everything in “the Rhetoric of the Image” (1964: 156). Connotation and denotation are explained as a literal object which is the denotative value and the anchor to the connotative value. The connotation refers to the meaning and emotional response that is generated from a representation. In terms of reducing and separating meaning and object in the aims of representation, the connotation can subsequently exist in the space created and become a signified and signifier. This space can become a danger to representation in some cases and will be further explored in this essay. This concept of reductionism must therefore be carefully considered when creating data visualisations or art. This highlights that there is the idea of reductionism in both the act of representing through visual practice and the impact of reductionism when crafting representation. Kidd explains that the core of creating connotative representation across spheres is to provide context.
Contextualisation
Contextualization of a signifier is empirical for creating a symbol that holds meaning to humans. In this sense, contextualization allows for a meaning generation. The connotative value of a representation deludes the manner in which to original object is interpreted. Contextualisation is generated through the imparting emotional value on a representation and is also naturally resultant of the environment of said representation. As discussed earlier, reductionism is a key component of both creating representation and representing. The reduction is put into action and is only useful when supported by the idea of contextualisation via correlation. Context established a correlation between signified and signifier and therefore for the representation to be understood, context must be apparent in the representation (Kerren et.al 2008: 3). Contextualisation can also encompass the inherited connotations of a representation's environment. This inherited connotation skews the interpretation of the representation and therefore must be considered. In regard to data visualization, this can refer to the quality of the data or collection of the data, which are processes that occur prior to the visualization (representation).
In the practice of data visualisation and art, this contextualization must be prevelent by constructing a correlation between the signified and the representation. Data visualisation may achieve this by titling the data set and artistic practice implements this by stimulating an abstracted experience of the signified.
The notion of data has largely been approached through the view of denotation. This is because data holds a quantitative value that has caused data to be viewed as entirely un-coded (Vande Moere & Purchase 2011: 365). Barthes argues that no representation or denotation exists in isolation and is free from the encoding of meaning. Viegas and Wattenberg (2009: 1) further support this idea of a lack of neutrality. The lack of neutral data can be argued to be freeing for a more artistic approach. While this is true, Kidd highlights the importance of understanding that data visualisation aims to generate knowledge. This conjunction of knowledge creation and the opportunity for encoding data with embodied reactions results in a more nuanced representation. In regard to data visualization, this means that the presentation of the visualization primarily impacts the viewer's relationship with the original unprocessed data. Viegas and Kidd propose that there is a greater ability to achieve reflexive connotative value through the creation of data visualisation. ,
Discussion: Interaction
The result of representation that has been constructed with adherence to the ideas of reductionism and contextualization can give way for deeper interaction and therefore more embodied meaning. Kidd states that representation can take on the role of ‘myth’ (2015: 34). In this sense, myth is the bread-crumbing of informing by leveraging the reduction and separation process of creating a representation. This lack of information and context urges the viewer to interrogate the representation further. As mentioned previously, the space between signified and signifier can be beneficial and detrimental. The detriment of this ‘myth’ is that connotations that have not been carefully constructed are applied to the representation. This can be harmful in allowing misinterpretation via underrepresentation.
This interaction of persuasion with representation is a fundament of artistic practice, according to Roeser (2016: 127) and has the ability to subvert the myth element of representation that is explored by Kidd. This means that the lack of correlation between representation and represented forces “emotional-moral reflection”. This reflection is what art aims to achieve: a subversion of meaning and interrogation of representation. The act of reflection also generates more knowledge for the user and therefore encompasses the goal of data visualization. An example of this interaction of the goals of artistic practice and data visualisation with representation allows can be seen through AI-mediated artwork. The creation of artwork that is grounded in a clear representation can allow users to subvert their connotations of AI. This connotation is the result of a lack of knowledge on this topic that causes a lack of representation of AI that is accessible.
Conclusion
This essay has demonstrated the intersection of representation, data visualisation and data art. Data visualisation and artistic practice both aim to subvert information so that knowledge can be experienced in a different form. This, therefore, highlights the need to adhere to basic principles of representation. The principles that have been discussed in this essay have been a reduction, contextualization, correlation and persuasion. Reduction allows for rapid cognitive interaction, correlation gives way for contextualization by retaining aspects of the original signified and contextualization allows for a deeper meaning to be generated by exposing the use of the artwork or visualisation and finally persuasion incites interaction and reflection of representation.