All the projects below followed the same procedure, as an exercise in computational information design:
- Retrieving metadata from a set of pictures, using a client of the Flickr API;
- Creating a network of tags from these metadata using Netbeans or a dedicated Gephi plugin; Two tags are connected if they are found to describe the same picture;
- Exploring the network visually with Gephi;
- Converting one visualization into a SigmaJS web page, uploading to a server via ftp;
- Writing a document explaining the how and why of these operations.
The short descriptions below are selected abstracts from the longer documents written by the students to describe their project:
“I am a digital illiterate. So I have to start from scratch in understanding how the process from acquiring to visualization works. I have found that I am now able to use the practical tools to create networks of tags and pictures. In later life this can be useful when developing certain projects.
But we live in a world of specialization. I can sacrifice countless hours learning to perfect this certain skill but the lack of affinity and talent will refrain me from doing so. If I want quality in the short term it is best to let others who are far more skilled apply their knowledge. So the main thing I learned is that these skill are very useful and I will use them in a practical sense by outsourcing them when needed. I do know now that I need the skills and the recognition is the main lesson learned for me.”
“What I was expecting to see in the graphical representation of the Flicker tags was different clouds of relative tags according to the different sets available on Flicker. There are 59 sets (or albums) on the Nationaal Archiefʼs profile, about different subjects and depicting different periods of time. However, some of them present overlapping subjects, like the albums “Eerste Wereldoorog” (First World War) and WO I (which is the abbreviation for the same theme).
What is seen, however, is that when looking at the network with no filters at all, there are a few terms that get some distance from the majority of the tags, which form a very confusing “hairball” effect, but most of the associated tags are mixed in a big grouping. Even though it is explained because of the same provider – the Nationaal Archief – and its public and national character, the sets (or albums) are much more related to each other than expected.”
“The end result [of this project] is an easily accessible web page that anyone can access and use. It is a good way to present data on the Internet with relatively little knowledge required for both the creator and the user. Because it doesn’t really require that much technical knowledge for the creator, it can be used by scholars that want to present their data to a wider audience. Scholars often limit themselves to academic publications that the general population generally doesn’t read. The Internet would be a great way for researchers to show their work to the public.
After all, universities are publicly funded and are therefore working for the public interest, at least in theory. Presenting results to the public should be a high priority for researchers. The Internet is one of the best ways to do this because almost everyone has access to it, as opposed to academic journals that often have paywalls. These kinds of publications will only become better as the technology improves and researches become better at producing them.”
“For the final assignment we were asked to visualize the tags used in pictures for any given subject. As a subject I chose Dordrecht and I selected the pictures from 2006. Why 2006 in particular? Once every two years Dordrecht hosts an event called Dordt in stoom. It is one of Europe’s biggest steam engine events where old steam engines of all sorts and sizes are displayed in the old city centre. Steamships steam down the river and a steam train is transporting people over a short distance to a nearby town and back.
I made some first analyses based on the relation of the colour of a node to the frequency a tag was used in the graph. The top five of most used tags were all positioned in the centre of the graph. Absolute champion was the tag Dordrecht which appeared 1148 times. On a respectable distance the tag Dordtinstoom appeared 892 times. Vehicle was number three showing up 698 times. Even though they were typos Vintag came in fourth with an appearance of 382 times and vehicle concluded the top five with 194 appearances. Two other big networks that appeared on the right side of the graph with the light freen-ish colour were about infrastructure (bottom) and city planning (urban atlas etcetera) on the top linked by a way of photography, in this case aerial photography and the province of Zuid-Holland”
“Even though I did not get to use the dataset I would have wanted, I think this one can still be useful for people who want to organize something related to the city of Rotterdam. For example, using this network, it can be concluded that anybody who visits the ‘Wereld Haven Dagen’ (World Harbour Days), a very big annual event located near the Erasmus Bridge, doesn’t care about the harbour and even less about the ships.
The main reasons people visit the event is because there is evidently quite a popular demonstration by the Corps Mariniers and a very spectacular fireworks show. This can be seen by clicking tags like ‘wereldhavendagen’ and ‘havendagen’. The tags you’ll expect to see would be related to the actual harbour or ships, but instead most related tags point to a demonstration or fireworks. The same could be done with other tags.”
“This visualisation describes my Facebook network.
Based on who knows who, mutual friends create the context within which I knew people. This is based on data downloaded from “Give Me My Data”, which is a Facebook plugin.”
“The reason why I created this network is to gain insight in the different subjects that people associate with submarines and additionally show how these subjects individually are connected to submarines and to each other. Additionally, I was curious whether the ‘big days’ of the submarine, i.e. both World Wars and the Cold War, would have an effect on the amount of tags concerning those days. So, a hypothesis could be that ‘submarine’ would closely associated with tags like ‘U-boats’, ‘Germany’, ‘Soviet-Union’, and so on. This didn’t really prove to be true, because the most common tags were ‘museum’, ‘navy’, ‘2013’ and ‘royal’. Although the tags related to the big wars of the twentieth century were present, they didn’t occur as much as was expected. A tag like navy has 1255 occurrences and ‘German’ only 98.
A few other examples and their respective amount of occurrence: ‘Soviet’ (31), ‘Coldwar’ (80), ‘Worldwar2vessel’ (48) and ‘WorldwarII’ (36). So, judging on the basis of these findings; submarines and the World Wars and the Cold War aren’t really things that people consider as very close related. In other words, people don’t seem to automatically make that connection or if they do; it doesn’t show up in the network.”