Cambridge – the land of squirrels and roasted pigeons

There are two squirrels that live in the garden next to my house and I see them almost daily (if studying permits me to have a look out of the window). But next to the CMS (Centre for Mathematical Sciences) there is a huge pack of them – last time I went there I saw six of them chasing each other in a tree and I believe even more of them live there. Nice to work there as a mathematician, I believe. Moreover, I have seen a kingfisher in Cambridge. It was beautiful and I have never seen one before, just read about them and how beautiful they are. So the list of animals seen in Cambridge so far got another animal species on it and now contains: a fox, squirrels, pigeons, a kingfisher, hares, cows, geese and ducks. And I believe many more, but these are the ones that you don't meet in a city, usually.

I got a green scooter for my birthday last term. It is great – I get everywhere about twice as fast compared to walking, I get some exercise and I have earned admiration of couple of small kids, judging by their loud “Mom, wow, how you seen that scooter?”.

I mentioned my project in the last blog promising I will tell you more. The title of my project is Automatic Generation of Mind Maps from Essays. To explain what my project is about, I shall first explain what mind mapping is. Mind mapping is a very powerful technique how to organise your information before writing an essay or while learning something.

Wikipedia describes mind map as “a diagram used to visually organise information. A mind map is often created around a single concept, drawn as an image in the centre of a blank landscape page, to which associated representations of ideas such as images, words and parts of words are added. Major ideas are connected directly to the central concept, and other ideas branch out from those.”

My system makes life easier by completely automating the creation of mind map from a given plain text using techniques of Natural Language Processing (i.e. by using computers to process text). The task was hence developing a system that takes plain text as an input and then does the following:

1. The text is parsed and irrelevant information is thrown away. Irrelevant information involves for instance most of the adjectives, adverbs, prepositions and conjunctions. For instance the sentence “Augustin is currently writing blog in his room” should be reduced to “Augustin (subject) writing (verb) blog (object)”.

2. Then the information obtained in step 1 should be adequately grouped together – i.e. common actors and their actions should be found. For instance, if there were two sentences (in this stage stored as a tree) “Augustin has spoon” and “Augustin has book”, they should be merged into a graph with main node “Augustin”, child node “has” with two child nodes “spoon” and “book”.

3. In the final (front-end) step it is necessary to display the generated mind map as a tree, but not in the classical layout with root on the top and its children underneath. The layout of mind maps dictates the root to be in the middle and its children are placed on a circle around it.

This went surprisingly well and now that I am ready with the implementation, I am writing it all up – we are required to write 10 to 12 thousand word dissertation about the project we have done.

I think this was rather academic, so I shall end the blog post with something more relaxing. We had Computer Science dinner with other computer scientists from my college this term. It was quite fancy and the most curious meal was the roasted pigeon meat. So now I cam say I truly understand what being a posh means. However, although the entire thing sounds very fancy and delicious, it wasn't that great. Next time I would prefer roasted pig or a duck.