Twitter used for tracking people's attitude towards vaccines

Conducted by a scientist, Marcel Salathe, at Penn State University, this is the first research used to measure how social media tools affect disease networks.
The study began in August 2009 and continued until 2010.
Out of all the available social media tools, Mr. Salathe chose Twitter for his experiment because usually the users subscribed in this network, want other people to read what they have to say. Also, they have to be very concise in their opinions because a "tweet" (Twitter messages posted by users) is only 140 characters long.
How did it work?
Basically, the scientist began by collecting all "tweets" related to vaccines in general. After this, he began "following" people's words and feelings towards the H1N1 vaccine against the swine flu.
Nevertheless, this is a huge amount of work, and Mr. Salathe couldn't have done it by himself.
He asked his students to rate 10% of the "tweets" selected as positive, negative or irrelevant. Afterwards, through a computer algorithm, that a computer programmer and analyst at Penn State University made, the tweets entered were rated directly into one of the three categories. This was possible using the model of the categories made by the students.
Due to the information available on Twitter about the different users, Mr. Salathe was able to identify people's attitude towards this depending on the region they came from or age. He considers this a very important result because future campaigns can be build based on regions that need more information than others.
Another very important result of this study was that people that think likewise are more likely to form a small community and "follow" each other. If a user doesn't think the way another one does, most probably he won't "follow" that user. This is how communities are build and for vaccine campaigns this can mean a lot because the companies would know whom to address in more detail.
Mr. Salathe plans on repeating this study for diseases such as heart diseases or hypertension in order to observe the people's attitudes.
This study will be published in the PLoS Computational Biology journal.