When I first heard Chris Danforth speak, the first thing I did was go and follow him on Twitter. Anyone of his brilliance is sure to continue producing quality work, and I look forward to following him and his work. https://twitter.com/ChrisDanforth
I found the topic fascinating. He wants to cycle through data and be able to track how happy/sad the people are feeling. It was so interesting to see the process which they thought about solving the problem... The data they wanted to use consisted of song lyrics, Google books (books that have been scanned and read into Google) and tweets. This was fascinating to see. They then went through and had to set the dial on what were happy words, what were sad words. I found this entire process to be fascinating. They used a service called Mechanical Turk on Amazon, where they paid people to go through and rank words. This was fascinating to me. He kept using the term "set the dial" on the experiment, meaning he wanted to come up with a robust way to classify the data.
Once the data was collected, he got rid of the noise (the,and,of,it) in order to amp the signal! The signal and the noise. Great book by Nate Silver I believe.
Overall, I thought the analysis and the computational techniques were incredible. And plus Chris Danforth was such a fascinating individual. Going to this talk made me want to learn more about scraping data - and it made me want to start doing cool research and investigate things that interest me.
I found the topic fascinating. He wants to cycle through data and be able to track how happy/sad the people are feeling. It was so interesting to see the process which they thought about solving the problem... The data they wanted to use consisted of song lyrics, Google books (books that have been scanned and read into Google) and tweets. This was fascinating to see. They then went through and had to set the dial on what were happy words, what were sad words. I found this entire process to be fascinating. They used a service called Mechanical Turk on Amazon, where they paid people to go through and rank words. This was fascinating to me. He kept using the term "set the dial" on the experiment, meaning he wanted to come up with a robust way to classify the data.
Once the data was collected, he got rid of the noise (the,and,of,it) in order to amp the signal! The signal and the noise. Great book by Nate Silver I believe.
Overall, I thought the analysis and the computational techniques were incredible. And plus Chris Danforth was such a fascinating individual. Going to this talk made me want to learn more about scraping data - and it made me want to start doing cool research and investigate things that interest me.
No comments:
Post a Comment