A few days ago, @infovore of Twitter sent me the link to NeoFormix, an experimental Twitter processor. I was really happy about this, because I love infographics, and I love to mess around with new types of data presentation. The NeoFormixTwitter StreamGraph was built by Jeff Clark, a brilliant developer. You can follow him @JeffClark on Twitter.
My goal was to track Twitter volume, speaker quotes, and general buzz around the event of every attending Twitterer in the audience. I did not want to have a sheet of messy or relatively non-granular data like Summize or Tweetvolume provide. I wanted to show volume over time, and for this I determined that graphic output was needed.
There were two hurdles I had to bypass before being able to adequately gather uniform data. The first was that the tracking hashtag for the ISF was not stable. Tweets of attending members migrated from @summit to #ISF08 to #ISF, and thus data that coverage was spottier than necessary.
In order to solve this problem, I began to gather every Twitter user that was attending the ISF into a text document. Then I sent out Tweets welcoming them to the conference. At the end of each welcome Tweet, I signed it @summit #ISF. As I welcomed more Twitterers and gained more followers, I dropped the @summit and Tweeted as much as possible about the event with others, using simply #ISF to sign my post.
By around 10:30 Am, the #ISF hashtag was well on its way. Once this hashtag standard came about in a uniform manner, I was able to run much more accurate analytics on the data. I was then able to use the word ‘ISF‘ to track top word volumes related to most of the Tweets concerning the Internet Stragety Forum.
What It Shows
The period of time that the graph shows begins during the middle of Geoffrey Ramsey, Co-founder & CEO, eMarketer‘s speech, covers the speech of Dan Stickel (CEO, WebTrends) speech and ends a little before the end of Nancy Bhagat Vice President, Sales and Marketing Group; Director, Integrated Marketing, Intel Corp‘s speech.
The conversations bursts worked just like sine waves as audience began to engage with the material of each new speaker. As memorable quotes were released into the audience, a lot of tweeting and retweeting coverage occured, melding some of the terms into like-groups. The graph shows that people tweeted about the speaker during the middle of the speech as opposed to at the beginning or end of the speech.
The First Burst
In the first Twitter lump, Geoffery Ramsey talked about ‘FOG’, or the Fear of Google. You can track it in this graph! Fear shows up, as well as Google. During this time, attendees were using @summit more often than #ISF to track the conference, which shows.
I am trying to determine why “life” showed itself so often, but “social, network, online, trust, marketing and show” made a lot of sense.
The Second Burst
Presenter Dan Stickel was not as quotable, but the Twitter reporters recorded his name, and the fact that he was speaking. At first they used both his first and last name to ID him, and then used his last name, for the sake of brevity.
The other lumps show that there were tweets during this time, some attributed to his name, but none that were unified. I.E., many Twitterers did not quote the same parts of his speech at the same time, or in enough volume in order to show up on the graph as an actual word.
The Third Burst
Nancy Bhagat of Intel Corp’s Keynote spurred a lot of Tweets about Intel, and thus her name is associated with it.
There was also @summit, presentation, marketing, great, Nancy, and bhagat. Most of the Tweeting was done towards the first of her speech, as well as the discusion of her status as an Intel worker.
There are many graphs like this available online. Most are made by students at colleges, and a lot have to do with graphically displaying content volumes. I found this analytics visualizer to be exceptionally powerful because of its ability to track word volume over time.
The applications for this type of visual presentation of information are vast. During the ISF after party, I determined that these graphs would be an invaluable tool for examining PR statistics over time, or, as I discussed with Dan Gaul, Kent Lewis and Geoffrey Ramsey, the highlights of one’s speech. If I sat down and pulled apart the code with someone, it would be fun to develop this graphing system into an extremely granular tool for online reputation management.
After the conference, Kent Lewis of Anvil Media suggested that I demonstrate the report to Geoffrey Ramsey, because the graphics allowed a quick and easy way to show him the highlights of the speech he gave. When I showed him, he was really excited about the results, because he did not know how to gauge the success or failure of his speech.
Instead of digging through pages of Twitter data, even through Summize.com (with the search term #ISF), the method I developed allowed him to see just his speech, and exactly the topics that hit the audience the hardest.
My research depends on attending conferences because my current focus is on visualizing data with 4 main dimensions.
In this way, data becomes more like an audio file, and even closely resembles it. It is a friendlier way of viewing trends, and is more accurate (because of the added dimension of volume) than
Currently, the tool I am using is Java based. It does not yet allow the user to set periods of time, and does not have the server capabilities to store server data. It is a brilliant data analytics tool, and if it were to allow a greater amount of granularity (in terms of keywords), as well as time range, it would prove to be an invaluable tool for tracking Public Relations. Currently, it is possible to do this, it just takes a longer amount of time to do so.
My goal is to approach the tools’ inventor, Jeff Clark, about collaborating with him to create a more robust version that would incorporate a larger time frame, clickable data formats (I have a paper prototype of all of this), and a zoom feature. A sort of map of time, or an audio burst.
The interesting part about visualizing data in this way is that it shows that there is an inherent difference between what a speaker says and the audience “hears”. Hearing, in this case, is defined by how the speaker’s name, company, and words are picked up by microbloggers and re-tweeted online.
If tech conference attendees were prompted to provide their Twitter id with their conference registration, tracking processes could be preformed more easily (this was the case at Gnomedex, a conference I was invited by Chris Pirillo to attend).
This project is just one of the experiments I’m working on. As it is most easily done during conferences, I have to wait for conferences with a substantial amount of Twitter users.
I’d like to be able to show this in real life, because its more enjoyable to get really excited about the data. There are so many great potentialities with a tool like this, because being able to visualize data over time with an extra dimension of volume is really exciting.
It’s also great to be able to discuss new methodologies with people because so many more conclusions can be gleaned from discussion. I recently presented this technique to a group of Portland tech people at a sort of software demo session. There was a lot of great feedback there, and new ideas gleaned from it. It’s amazing, the value and speed of digital communities.
I’ll be applying this tool to the crowd and will be trying to trick it out, or hack it to be able to show more statistics over time. I’ll be doing the standard audio recordings of the entire conference, so that I can compare and contrast what was Tweeted vs. what was said in real life. I’ll probably be taking about 50 graph samples, so that the relative volume and interest in each speaker can be tracked. There will be a lot of write-ups about the uses of this type of visualization, and how it can be applied to PR campaigns.
Systems are optimal when the amount of time and space it takes to get pieces of relevant data from one person to another continues to decrease. Those designs/processes that exemplify this paradigm will be successful in the future economy.
Amber Case is a cyborg anthropologist, internet marketer, and speaker from Portland, Oregon. You can contact her at caseorganic at gmail.com, or on Twitter at @caseorganic