Over the course of my term in Budapest, I took a course called “The Structure and Dynamics of Complex Networks.” It ended up being a class co-taught by four different professors, and we had to complete a final project that was very open-ended. This course covered all sorts of things, from graph theory, to biology, to economics, to the internet, and so our final projects, like the course, were meant to explore other academic fields using networks. My final project ended up being a sort of research project in Computational Linguists. My task was to try to determine how users tag photos on Flickr and use this information to try to determine how “well-tagged” a photo is. Open-ended? Very. However, my good friend Margaret from my program told me about a project going on at her school called WordNet. Let me tell you, it is pretty sweet. It was started by a computational linguistics prof, and the project has branched off in several directions, including one sub-project called WordNet::Similarity which creates trees to measure the semantic similarity between words. So, armed with this great tool, I was ready to crawl Flickr to mine some information and get KNOWLEDGE!!! Optimistic? Yes. It ended up being a little tricky, and I crashed my computer a few times, but I ended up getting cool networks like this one. This is actually a list of tags from my sister’s Flickr account. Tag words are connected if they are above a certain threshhold of semantic similarity according to WordNet::Similarity’s measurements:
If you want to read more about my results, here is the final Prezi presentation I gave at the end of the term:
Overall, I am very pleased with this class, and it most definitely has sparked my interest in Computational Linguistics! I will most certainly continue experimenting with WordNet!