In 2009 we are in the midst of an interesting era for data visualization, particularly as it becomes coupled with the social web. Increasing processing speed, bandwidth and storage capacity are making it relatively simple to render and access visual representations of data. Developers have released libraries of code so we can easily create our own visualizations; and access to all kinds of data is becoming incredibly standardized, particularly through the use of APIs. So as visualization becomes much more straightforward to integrate into online environments, it makes sense to rethink how it can best be used in this setting.
You will have possibly already come across social networks about visualizations if you’ve ever used IBM’s Many Eyes or Swivel. Now is a great time to expand on the work of these pioneers in the field, considering that there is a great need for data visualization as a way of addressing the problems of information overload and the technology to support it is now falling into place.
After some extensive research into the area, our take on visualization within a social space is that it should support a shared storytelling process around a data set. By calling it a shared storytelling process, we mean that the process of data visualization starts with an individual with an idea (or intent to tell a story) and ends with a community who share the story and adapt it. An individual who wants to tell a story needs to know how to do it. But then they need to be able to tell the community about it in a way which lets everyone understand what is going on. Communities like to be able to pass stories around to be able to entertain and educate, but without forgetting what the story was originally about.
…our take on visualization within a social space is that it should support a shared storytelling process around a data set.
This shared storytelling process can be supported by designing an appropriate web-based interface. In this article we’ll be talking about some ‘big picture’ ideas that have become design implications in our process of conceptualizing this interface. In subsequent articles we’ll be delving into these ideas in more detail and presenting excerpts from the interaction design patterns we have collected to construct the experience of using visualization within a social environment.
Rather than a medium for storytelling, visualization has traditionally been a tool for the analysis of data, and the aforementioned Many Eyes and Swivel have extended its use to the social web in different ways: Many Eyes has focused its attention on supporting the analysis of data through the visualization in a collaborative environment. The site achieves this by using an interface that supports transforming the data to change the look of the visualization, annotation that keeps to the style being used and the ability to save any view being worked on and add a comment to it.
Rather than a medium for storytelling, data visualization has traditionally been a tool for the analysis of data
Swivel takes a different tact and prioritizes the idea of using visual communication to express the meaning of the data. It achieves this by using the assigned title of the data set to query Flickr and this produces a set of images that should visually define what the data set is about. This potentially helps to create a wider understanding of what the data is, and how it might interest people within the Swivel community who are looking to contribute. But Swivel doesn’t support much community based analysis of data. Visualizations are presented in a very standard manner (using bar charts, pie charts and scatter plots) and can be commented on but not annotated or manipulated.
Support an holistic process
What we’re looking to do is extend some of the functionality of these sites to support a more holistic process that better supports social activities like collective intelligence and sensemaking as a visualisation is created and reiterated. We’ve conceptualised this as a three stage process that begins with creation, extends to interpretation and ends with capturing and reintegrating these interpretations back into the conversation.
- The creation process is an individual process;
- Interpretation is one that belongs to the community;
- Capturing saves the process for posterity and allows it to iterate.
For now we’re going to talk about the theory of object centered sociality and how it holds these three different processes together. In subsequent articles we’ll talk more about the importance of the individual processes and how interaction designers can use them when they want to get the most out visualization in a social web setting.
The basis of capturing this shared storytelling experience through visualization is creating an object-centered social network in a similar style to Flickr, where interactions occur around the photos that users share (as opposed to an ego-centric one such as Twitter where users ‘follow’ other users and receive their updates) that enables it. To do this the visualization needs to become a social object within the network. This essentially means that the interface elements used to represent a visualization will afford discussion, or contain things for people to talk about. We can create a visualization as a social object by giving it an identity and making it interactive. We’ll talk about why identity is important in a moment, and come back to interactivity after that.
The basis of capturing this shared storytelling experience through visualization is creating an object that affords discussion
Before identity comes into play, the visualisation needs to be created. This is more imperative within a social network because the creator of the visualization needs to understand their own data, and other people need to understand it as well. So we need to understand how data is best visualised. This can be difficult to understand at the best of times, but focusing on the communication goals and intended audience is often the best way to start. This prevents users from needing intricate knowledge of various visualization techniques such as the different properties of a box plot and a scattergraph. Foregrounding these aspects of the process we refer to as MAPPING in the diagram above helps people to focus on the answer they are seeking from the data rather than being bogged down in the process of presenting the visualization correctly. Once this is achieved then social processes around the visualisation can begin.
But objects can’t exist on their own within the social network and need to have identities built around them. Identity of an object can be implemented just like identity of a person within an ego-centred social network. You can see in the image above that the visualisation on Swivel has an image, a source, categories and it could have tags. These all work together to provide an sense of identity and give people within the social network some chance of being able to work out what the visualisation is about. Along with identity comes reputation and history. As different users interact with the visualisation and make contributions to it, its identity remains unchanged. Adding an avatar provides visual clues about the identity of the visualisation that may not be communicated through the text associated with it.
Back to interactivity
Now we come back to interactivity. In order for social processes within the network to have any real value, then the visualisation should be one that can be manipulated. This allows for further insights to be drawn out of the data that may have been missed when the original user chose to visualise it. So it is not enough to be able to leave comments on the visualisation, which is the standard way that content is treated within most social networks. Users need to be able to tweak a visualization, from switching axes on a line graph, to filtering or bringing in new data, or even changing the type of visualization used.
Capture the process
This process also needs to be captured. Users should be able to highlight parts of the visualization that represent interesting insights into the underlying data set. They should then be able to store these as an attachment on the original visualization that doesn’t in any way detract from what the original visualization represented. In this way the shared storytelling process we talked about earlier is at play. If the visualization is the story, then as it is passed from person to person they choose to promote different parts of the story and neglect others.
Because the visualization has an identity, its original form is recognizable through this process and the subsequent retellings of it become versions of this original that make up a bigger narrative about the issue, or in this case the original data set. This process would probably occur without the interface in place to support it – everyone sees things differently, and everyone expresses the way they see things differently. But with an interface in place to support the processes of an object-centered social network, we can capture the insight that the storytelling process brings to visualization and store it as an artifact that creates greater knowledge about the original subject.
The power of data
Considering that sites like data.gov are launching soon, and technology guru Tim O’Reilly has been saying for a number of years that data will be the ‘Intel inside’ for the web operating system the amount of data that is available for people to manipulate is going to increase exponentially. Visualization has always been a good way of gaining insight into data, and creating a storytelling experience around it feeds into human needs to ask questions and tell stories. So for interaction designers, implementing this type of experience within the data-driven websites of the future has a number of benefits. One of these is that the particular website becomes more engaging for users, and drives take-up. Another that springs to mind is that by making this an interesting and engaging experience, much insight can be drawn out of the data which can be of benefit to the community (in a similar way to Wikipedia in its collation of vast amounts of information) and to the owners of the data, who may discover things about their data they’d never thought of before. So watch out for more articles from us in the future, as we go into further detail about how to implement this experience.
In 2008 the Australasian CRC for Interaction Design (ACID) partnered with Deloitte Digital to research applications of data visualization, through the Loupe Project. Deloitte Digital was preparing its accounting firm in Australia for the introduction of XBRL (eXstensible Business Reporting Language) which would see a significant change in the way business reporting was conducted. Rather than sending multiple reports to different agencies, XBRL would produce one set of data that agencies could draw upon for their own purposes as needed. As part of this change, Deloitte has released an online accounting platform and aims to change the relationship between accountant and client to become an ongoing conversation online. This process needs visualization to make complex business data more easy to understand for the client, and an interface to make this process a better user experience. The Social Life of Visualization is one outcome of our research into this solution with Deloitte.