In 2008, a group of students from various backgrounds worked together on developing a demo of "A Tag's Life": a visualization tool, which makes it possible to track and analyze the rise and fall of tags on the online photo sharing platform Flickr. With Impakt Online, we invited them to further develop A Tag's Life, and make it into a working tool. Recently, the University of Amsterdam decided to support a follow-up of the project, a Social Networks Dashboard, which gather data from social networking sites and visualize it.

A Tag's Life is developed by George Holsheimer, Mirjam ter Linden, Daan Odijk, Putri Sadiqah, and Raoul Siepers.

Tell me a bit more about your group... who are you and what is your background?

George: We're a group of students. Mirjam and Daan are from the science department, Putri is from the arts department and Raoul and I are from media studies. I myself am officially a film student, but got more and more interested in new media along the way. Now my thesis is about the similarities between the histories of film and video games.

Raoul: My background is one of interdisciplinary media studies. As a MA student in the field of new media, the Internet provides me with the ultimate interdisciplinary playground.

Mirjam: I did the Bachelor computer science and then the Master's programme, Information Science, profile Human Centered Multimedia. I have worked for many years as a modern theatre dancer so this is my second career.

Daan: After my Bachelor in Artificial Intelligence I followed the Master's programme Information Science, profile Human Centered Multimedia. Besides my studies I've had my own company in web development and have been working as a teacher at a high school.

The first version of this work was made for an information visualization course. What was the course about and how did you start collaborating?

Daan: Actually these were three separate courses on information visualization. The student project were set up in a collaboration between two faculties of the UvA and MaHKU. We did not have shared lectured and this way we brought our own expertises to the table. We've thought up the concept of A Tag's Life together. From there the new media students polished and contextualized the concept while our designer sketched a design and the Information Science students built a prototype. In close collaboration we have progressed this into a working demo in about 2 months.

What is the concept behind A Tag's Life?

The Internet makes it possible for a massive number of individuals to represent their thoughts, their passions, and their interests. It is not far-fetched to imagine that adding up all these peculiarities on a per-object basis could measure some form of popularity. However, similar outings may have different meanings depending on medium and context. For that reason, any tool seeking to determine popularity has to be clear about its source data and its implications for the output.

User generated content combined with a tagging system has dynamic, social, and representational
properties. Subject (the user) and object (the image) connect to sketch a portrait of preference and intention, and thus enable the step from meaningless data to telling trends.

What does the tool do?

A Tag's Life is a computer application that derives popularity from user interactions in the online image database of Flickr, and tracks its development in space and time. Next, it renders the obtained data on axes of time and quantity so that trends can be spotted. Space in turn is mapped onto a geographical representation of the earth.

This project takes the Internet photo sharing community Flickr as its starting point. Its motto "Share your photos. Watch the world" aptly expresses our aim. To create a tool for watching the world as it evolves seems like an ambitious challenge, but Flickr’s database of image tags provides us with the means for a modest attempt at doing so. Concretely, our tool should be able to visualize trend development in time based on user created image tags in Flickr. A trend in this case can be defined as a significant amount of uses of a specific tag within a fixed time-span. By visualizing trends in an intelligent way, the tool allows for easy and early recognition of trends as they unfold over time.

Such a design would serve both a monitoring and an analytical function. The monitoring function refers to the identification of trends as they emerge, and the analytical function refers to the analysis of trends on a retrospective basis.

How should we read the data in A Tag's Life?

The patterns and trends that this tool reveals should be regarded as a first, exploratory step and cannot go without thorough analysis of the underlying data. The figures presented in A Tag's Life are inherently bound to the limiting factors of access and representation. However, it is important to note that this does not deprive them from meaning. Realizing that meaning and popularity do not allow for an absolute correlation opens up possibilities for multiple valuable readings of the gathered material.

Precisely this ambiguous data is what makes A Tag's Life unique. While an interesting alternative such as Google Trends offers powerful tools of comparison and derives its input from a huge database of search engine queries, it omits the connection between signifier and signified. By visualizing both, A Tag's Life accepts semantic discrepancies as meaningful social constructs, while at the same time providing an instrument to find their origins.

How could this data be analyzed?

By looking when a tag was first used, and how many times it was used after that initial use, it is possible to see trends. Let's take a popular band for example. Early discoverers, or people who are in the band themselves will use the band's name as a tag for the first time. Than as there popularity starts to spread, the
amount of photo's with their tags will grow till they are at the highpoint of their popularity. After that the number of times the tag is used will (slowly) fade. If you would put this information on a map it would give you more information. Like: where was the tag used first, is just a local phenomena or is the tag used globally?

Why Flickr?

As the largest photo sharing site on the Internet, the underlying database is sufficiently large to produce representative results. Admittedly, search engines as Google's rely on a much larger database, namely the entire Internet (at least that is what it claims). However, identifying trends as related to the combined
input of discrete individuals asks for a different approach.

At the Video Vortex conference in Amsterdam, dr. Jan Simons presented his research on tagging and pointed out that people are not exactly flawless taggers. (For instance, that tag 'new' is very popular because people also tag NYC as new york city (3 tags) instead of 'new york city' or newyorkcity, etc.) How do you deal with that?

A Tag's Life analysis how users tag photos on Flickr. For this purpose, it does not matter whether an object tagged 'sneaker' actually is a sneaker, what counts is that the user in question deemed it an appropriate description. Popularity is the keyword. This way tags being subjective does in fact support our search for trends.

Furthermore, we try to think about what tags we add to the application. A tag as 'new' will not be added as this will provide the user with ambiguous and uninterpretable information. If someone wants data on NYC, we'll add 'nyc' or 'newyorkcity'.

What fascinates you about the aspect of time, related to tagging? And what is the importance of showing the dimension of time on a platform like Flickr?

George: What we were interested in was the development of trends. Web 2.0 sites like Flickr, youtube and Digg only give you an overview of what is popular right now. We thought it would be interesting to get an insight into what was popular last month or two years ago. Also A Tag's Life is a great tool to see how something that is popular right now came to rise to fame. Internet is hardly involved with its own history so trying to see where something came from and how it developed over time can be difficult to do. What we can show you for example is if something became slowly more popular, spreading gently both in time and space, or that it was a global booming event happening all of a sudden all over the globe.

Raoul: Flickr is not a static depository of photos. Instead, its true transitional nature is revealed by the act of tagging. Over time tags alter in use and meaning. They reflect changing attitudes, language, and interpretations. In that way tagging recontextualizes the iconic relationship between picture and subject. Your picture of Times Square at a given moment is not just that; it is also what you want it to be by means of tagging.

Daan: Our application shows something about Flickr tags that Flickr itself does not show. A good example is the tag Obama. You can see it's popularity exploding and come election day in the US, the tag is used globally.

How does this application work? And could it be translated into any platform with an open API and a tagging system?

Daan: The application consists of two parts. The front-end is the application that users see and can interact with. This front-end is developed using Processing, a Java based programming language. The front-end uses the data that is collected from Flickr earlier by the back-end. Collecting a tag such as Obama takes several days and can therefore not be done by the user. The data collection is run at the Universiteit van Amsterdam on a cluster of 46 nodes owned by MultimediaN. This speeds up the data collection enormously.

Part of the work that has been done in the last couple of months is rebuilding the data collection back-end into a framework that can be used to collect other sources as well. All these sources need is an Open API and a tagging system. This data collection framework now works with Flickr and has been used for other projects involving Flickr. The work that will be done in 2009 will focus on obtaining data from social networks mainly, such as MySpace, Facebook and Hyves.

People who have tagged their pictures, might not have thought that some day their tags and pictures would show up in an online art piece. What about this public/private tension of Web 2.0 applications? Since you've all been educated in the field of new media, I'd like to hear your views on these security and privacy issues. Are mash-ups causing more of this tension?

George: I personally believe that the whole web 2.0 explosion is based on some kind of (anonymous) exhibitionism. People want other people to see and hopefully appreciate what they have made. I think people are very much aware of the fact that what they put online can be viewed or used by other people. Our project is also a kind of social study because we can see which tags people used to generate more views.

Raoul: In the context of proprietary rights, mash-ups certainly create tension. However, not all sources are the same. Compare a Walt Disney feature to a picture posted publicly on Flickr. Using the former as a mash-up source will get you into trouble, while using the latter will probably go by unnoticed. Large cooperations are more or less adequately protected against copyright infringements, but the rights of the individual are at the mercy of others. Even so, this issue also hinges on our notion of publicness. Does posting pictures on a public platform as Flickr by definition eliminate all your claims to privacy?

Can this work be read as a commentary of Web 2.0 and certain buzzwords? How should we interpret this rise and fall of tags?

George: A Tag's life shows that certain tags get larger than the thing they originally represent. When something is becoming very popular, people start using that 'buzzword' in their tags in the hope to get more views. The tag than loses its meaning. In a certain way A Tag's Life is a demonstration of how tags should not at all be general, but as specific as possible in order to remain meaningful.

Raoul: A Tag's Life's goal makes for quite some irony. While trying to capture the enormous flows of data in an enlightening visualization, it in fact establishes the opposite. Glitches, incongruences and wild trends in Flickr's tagging system are revealed, thus creating an illustration of what we don't yet understand.

Since the start of this project, some of you have graduated. Any future plans in the field of new media?

George: I will graduate next month and am looking for a job in the field of educational game design.

Daan: I've started working at the University of Amsterdam on expanding the A Tag's Life tool into a more general Social Network Dashboard as described above.

Mirjam: I am now working as a software engineer for full-service internet bureau LostBoys.