Twitter: listening to a million conversations at once

After the keynote the conference broke up into smaller sessions. I went to the Blogs, Microblogs and Twitter session which should have just been called Twitter because all three research projects presented were focused solely on twitter. Apparently the open nature of twitter and the potential of being able to eavesdrop on all the conversations around the world is very seductive to researchers. They all wanted to analyze and discover patterns in the cacophony.

The first study was Hashtag your way to Health by Sherry Pagoto, the originator of the plank-a-day hashtag. Hashtags are words or phrases that mark a social media post as being part of a larger conversation. Everyone at Med 2.0, used the hashtag #Med2 to mark each tweet as being about or from the conference. Following a hashtag allows one to see the entire conversation around a single subject. Pagoto looked at hashtags associated with health and their source (commercial vs not commercial, organic versus generated by a mobile app, e.g. RunKeeper, MyFitnessPal). She analyzed the tags for popularity, origin and the type of messages to which they were attached. Mobile apps generated tweets with little engagement, i.e. people didn’t retweet them. Non-commerical hashtags were just as powerful as commercial ones. She also found that the success of a hashtag was not dependent on the originator having a lot of followers. In order to collect the half a million tweets she used in her analysis, she prospectively created the list of targeted hashtags and then used an app called TwapperKeeper to archive them for analysis (by my search that app now seems to be gone, Dr. Pagoto elaborates further on pulling hashtags here).

It was a good talk on a creative research angle. This peer-to-peer motivation with social media seems like an important factor in generating positive feedback for healthy behaviors.

The next speaker was Nicholas Genes of Blogborygmi fame (he was part of the initial cohort of medical bloggers). It was great to meet the man behind the words. He is an emergency room doctor in NYC and he wanted to see how twitter was used in natural disasters, i.e. can twitter be used to save lives. He looked at tweets that were sent during two disasters, the Sandy Super Storm and the winter storm, Nemo. He was interested in Twitter because it is a resilient system that works well on desktops, laptops, cellphones and tablets. It is a bandwidth miser so it performs well over wifi, 4g, and older, slower, cellular networks.

He looked at the most retweeted messages from government agencies and political leaders. Major Bloomberg provided the most retweeted message of Sandy (I believe it was telling people to evacuate a specific area of the city, now)and The Weather Channel provided a fun fact that was the most retweeted of Nemo (I believe it was the number of miles in NYC that needed to be plowed).

I had the chance to discuss his research at the end of the day and Genes was frustrated with the quality of his own data, He was skeptical of the validity of some of his data. This was a recurring theme of twitter researchers, it seems that best practices for searching old tweets have not been validated and each researcher solves this problem their own way. Genes had the added difficulty of needing to search the data retrospectively while Pagoto could establish what she wanted to search prospectively and then catch the tweets as they were made, seems like a better technique. It would be great if twitter would open their data for researchers.

Nick Genes

The last presenter from this session was Urmimala Sarkar. She used twitter to see how people reacted to the change in the USPSTF guidelines for Pap smear (Abstract). They found discussion about Pap smears roughly doubled after the task force recommendations changed (2,549 to 4,673 tweets). One of her important conclusions was that to extract significant meaning from Twitter data required combining quantitative and qualitative methodologies.

One of the great things about this session was all of the presenters were great tweeters and the audience was salted with twitter-addicted people so going to this one early session allowed me meet a lot of the people who I would be reading on the twitter backchannel for the remainder of the conference.

Trackbacks

  1. […] Twitter: listening to a million conversations at once […]

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Follow

Get every new post delivered to your Inbox.

Join 4,273 other followers