«1 Introduction and previous work By the term Interactive Digital Artwork (IDA, for short) we mean any artwork where digital technology is an ...»
A classiﬁcation framework for interactive digital artworks
Department of Mathematics
Univ. of Roma Tor Vergata, Roma, Italy
Abstract. We deﬁne Interactive Digital Artworks as Information Technology intensive systems for which spectators are involved in the production of the artistic
output. We propose a novel framework for classiﬁcation of interactive digital artworks built on the critical revision and reﬁnement of previous work. Our approach is based on the input-process-output view of Information Systems. The classiﬁcation framework is validated by applying it to the classiﬁcation of 54 interactive digital artworks realized in Italy.
Key words: Digital art, Interactive content production, Classiﬁcation framework 1 Introduction and previous work By the term "Interactive Digital Artwork" (IDA, for short) we mean any artwork where digital technology is an essential component and which is interactive (in the common sense this word is used in IT). IDAs can be physical artworks placed in a public and open space (usually called "installations") or virtual artworks enjoyed on a personal device. Digital ﬁlms/videos are usually not examples of IDA, nor is digital music, since they both lack the contribution of the user to the content production. But when the outcome of video animations or music pieces is modiﬁed according to user interaction they are examples of "interactive digital art".
In  it is suggested that any computer artwork should be considered as an information systems and six research themes are proposed. More speciﬁcally, the following three
themes from  provide a context for our research:
• Computer Art as an Information Systems: each computer artwork can be abstracted by the input-process-output reference scheme tipically used for Information Systems.
• Computer Art and The Commercial Perspective: this is about management and economic issues. How to establish the value of an IDA? What about copyright? Which are the implications for IDA preservation for museum/gallery curators?
• Computer Art as a Socio-Technical Systems: Interactivity necessarily involves people in the system and in  it is written: "We should ask whether our current understanding of participation in systems development applies equally to the creation of computer artistic works".
Our research goal is to characterize and compare IDAs. To this aim we present a framework, or a scheme, that allows to arrange the various examples of IDAs in homogenous 2 E.Nardelli classes or categories. Using this classiﬁcation framework/scheme, in the following simply classiﬁcation, it will therefore be easier to discussing and/or producing and/or buying interactive digital art.
In this respect our classiﬁcation is similar to the ones used in the standard ﬁne arts regarding, e.g., painting techniques (oil, watercolours, fresco, pastel, gouache,...), materials (paper, wood, metal, stone, canvas, silk,...), tools (brush, pencil, roller, chalk,...), which makes it easier to discuss and to teach about artworks.
We based the structure of our classiﬁcation framework on the review of relevant literature concerning this theme [8, 4, 3, 9]. Then our classiﬁcation is validated by considering (a subset of) the artworks discussed in the IDA literature and showing that they can be grouped according to our classiﬁcation in a meaningful way.
The novelty of our proposal with respect to previous work is that it is explicitly based on the standard input-process-output view used for discussing Information Systems.
Previous work addressing our research goal was published in 1999 by Sommerer and Mignonneau , in 2002 by Hannington and Reed , in 2004 by Edmonds, Turner, and Candy , and in 2008 by Trifonova, Jaccheri, and Bergaust . The main emphasis in all these classiﬁcations was on the user interaction. Hence all proposals were centered around the various kinds of interactions and did not consider the more general viewpoint of IDAs as Information Systems.
The older classiﬁcation in the literature is the one discussed in , addressing "interactive artworks". Since it is not focusing on the use of information technology it is not able to characterize its speciﬁc aspects.
Then, the classiﬁcation in  is covering "interaction in multimedia applications", hence it considers a larger and different set of works, since many multimedia applications have no artistic component.
Subsequently, the classiﬁcation in  discusses "relationship between the artwork, artist, viewer and environment", hence it does not cover those internal aspects of the artwork that are related to the processing of input from the artwork audience, which is a very important aspect of an IDA.
Finally, the classiﬁcation proposed in  addresses "interactive installation art": on the one side it hence considers a narrower set of works (just the installations and not the artworks experienced on personal devices, which are more and more important means for user interaction in the Future Internet), but on the other side it has been built by focusing just on interactivity as the main aspect of IDAs.
In the rest of the paper we ﬁrst present our classiﬁcation (Section 2), then compare it to previous ones (Section 3), and ﬁnally discuss its validation (Section 4).
2 The classiﬁcation framework An information system is conventionally seen as a system which processes a given input to produce a desired output. We consider an IDA in the same way, as a system A classiﬁcation framework for interactive digital artworks 3 which receives a certain input, called content in this context, and producing as a result the output intended by the IDA author (i.e., the artist). It is also helpful to consider the process producing the intended output as if it were a function in a mathematical sense, that is an
"device" which at each time instant transforms its inputs into its outputs according to its mathematical speciﬁcation.
The dimensions of the classiﬁcation are:
content provider: who produces the raw material processed by the IDA, processing dynamics: which kind of variability has the processing itself, processing contributors: which are the sources affecting the dynamics of processing.
For each dimension we now provide different values, that are the labels of our classiﬁcation. We use the term artist to denote the person or team who has invented and realized the IDA, audience to denote the human beings actively and consciously providing any kind of input to the IDA, and environment to denote any passive or not-conscious entity present in the environment surrounding the IDA.
Regarding the content provider dimension, the source providing the content to the IDA can be either the artist or the audience or the environment. This dimension has therefore 3 possible values, or points, and an artwork can be labeled, with respect to this dimension, with one, two or all the values.
Regarding the processing dynamics dimension, the processing function of an artwork can be static with the passing of time, or it can be dynamic, that is changing as time passes. Note that the change considered here is the intrinsic change of the processing function, not a change in its input parameters. But the input parameters may determine, partly or wholly, such a change. In the case of a dynamic processing function, we consider three values, in mutual exclusion, to be used for a better characterization of the
• predeﬁned change, where changes to the function follows the plan deﬁned by the artist;
• casual change, where changes to the function derive by random choices, even in the case the set or the domain of the possible choices have been completely pre-deﬁned by the artists;
• evolutionary change, where changes follow an unpredictable path deﬁned by the evolution (in a biological sense) of the processing function itself.
The single value for the static case plus the three above values for the dynamic one give a total of 4 values (points) for this dimension. An artwork can be labeled with exactly one of these values.
Regarding the processing contributors dimension, the elements driving the content processing can be self-contained in the IDA (hence, what the artist has put directly inside the artwork affects the processing), or these elements can arrive at the IDA through the interaction with the context the IDA is placed within (that is, the processing function has additional input parameters causing modiﬁcations to how the content is processed).
4 E.Nardelli In the latter case, the providers of values changing the behavior of the processing function can be the audience or the environment. The dimension has therefore 3 values (points) and an artwork can receive one, two or all the labels.
Note that, in strictly mathematical terms, inputs to a functions are all equals, hence the distinction between "content provider" and "processing contributors" dimensions has no compelling mathematical reason. But from the artist viewpoint this differentiation is an important one, since it distinguishes between what she has directly inserted in the artwork and what arrives from the outside of the IDA, both in terms of the raw material and its processing function.
Also, an artwork labeled both under "content provider" exclusively with artist and under "processing contributors" exclusively with artist is not an IDA, since it has no elements of interaction at all. But as long as, in at least one of these two dimensions, the artwork is labeled with at least one more label, then it is an IDA.
The overall classiﬁcation space is therefore made up by 3·4·3 values or points. Each one of them can also be thought as a "labeled cell" containing all IDAs that can be classiﬁed with the labels corresponding to the point itself. Note that an IDA can be classiﬁed at the same time under more than one cell.
To give an example, let us consider an IDA taking pictures of its audience (say, one every ﬁve minutes) and displaying them while dynamically modifying them on the basis of data provided by the environment where the work is placed, so that each displayed picture is casually altered by one of the many processing ﬁlters deﬁned
by the artist, where the parameters guiding the ﬁlter are based on values read second by second in the environment. Then the classiﬁcation label for such an IDA is:
[content provider:audience, processing dynamics:casual change, processing contributors:environment].
A real IDA similar to this one is described in : an installation which is inspired by Andy Warhol’s statement that "In the future everybody will be world famous for ﬁfteen minutes" as well as by the pop-art style of his works. The visible part of the installation consists of a digital camera and a ﬂat-panel monitor dressed up like a precious painting (see ﬁgure 1 left). A computer behind the scene runs a software that detects human faces in visitors’ images taken by the camera, graphically transforms them, and then displays them for ﬁfteen seconds. The graphical transformation actually applied is randomly selected among the ones pre-deﬁned by the artist. In such a case the classiﬁcation label would differ for the dimension processing contributors, whose value would be: artist, since the kind of processing executed on the content of the IDA depends only on what artist has directly provided within the artwork itself.
Another real IDA similar to the above two ones is described in : an installation where the self-image of the spectator is changed by randomly chosen pre-deﬁned functions whose speciﬁc input parameters are provided by the spectator itself. In such a case the classiﬁcation label is: [content provider:audience, processing dynamics:casual change, processing contributors:audience].
A classiﬁcation framework for interactive digital artworks 5
Fig. 1. The installations 15 seconds of fame (left), and Sonic Onyx (right).