I was following links yesterday and found sites that are doing exciting things with Internet data and data sharing. They are authored by Jonathon Harris and Sep Kamvar.
On Feelings
Harris and Kamvar have created a tool that mines the most common blog sites for "emotional" words. Every 10 minutes, their tool looks for entries that contain emotional keywords in a set of 5000 different words related to emotion (e.g., angry, love, lost, stupid, perpetual, futile, etc.) The full list can be found here.
Then, once it has found and plucked out the emotion-laden sentence, it mines the author's blog profile to get public demographic data about the blogger: name (pseudonym), gender, date and time of the post, location, age, etc. and mashed that data with available data such the weather at the blogger's location to create a database full of information surrounding that blog entry.
They have created a number of different pieces that use this same data -- exactly as I want to be able to do for mine. Some are:
Even better: they have created an API (Application Programming Interface) that allows the public to use the data they have collected which they've made available under the Creative Commons Attribution-NonComercials-Sharealike license.
Twistori is one award-winning site inspired by WeFeelFine. More about them and their ilk another day...
On The Use of Language
While wandering their site I found a source for one of their pieces (Word Count)-- the British National Corpus. From their site, the BNC is described as "a 100 million word collection of samples of written and spoken language from a wide range of sources, designed to represent a wide cross-section of current British English, both spoken and written."
There are a number of options for using this data. Harris and Kamvar used it to do frequency analysis of words in proximity to one another.
Buried somewhere in my art journal and in an old assignment is a piece that reads in listserve emails from listserves that I subscribe to. Their natures are completely different -- a technical listserve and (if memory serves) a bicycling mail list. The project was to read in the words used and compare their frequency of use between the two sources, representing them using side-by-side network diagrams or some other mechanism to highlight the differences and similarities in the use of language between these two populations of emailers.
I executed a proof-of-concept piece as an piece in Art 410 or 412 with one listserve.
It's exciting to find a resource where I can consider their use of language against the larger English-speaking population (UK English? American English?). I may have to resurrect that piece. I've been keeping those listserve emails so I have years worth of data to use.
No comments:
Post a Comment