Friday, September 7, 2012

Charlotte Visualization Center Tracking Tweets Like Never Before

It’s no novel observation that communication today is vastly different than communication 20 years ago. Or 10 years ago. Even five years ago things were different: Facebook and emails ruled how we communicate with each other. Then, out of left field came a little website known as Funneling every thought down to 14 characters or less, each ‘Tweet’ is supposed to be condensed and pure, with the quick thoughts of each individual who sends them. A quick glance over at Twitter's own numbers blog reveals that the number of mobile users has increased over %182 over the past year, with an average of over 460,000 new accounts per day over the last month. Needless to say, the way things are progressing, Twitter has become incredibly important in how we as a society communicate with one another. The Visualization Center of University of North Carolina Charlotte, which focuses on visualizing massive amounts of data with computer graphics to make it far more accessible to users, has recognized this fact and is working on a project to not only measure and follow tweets being made, but understand what exactly is being said.

Collecting and researching Tweets and Facebook messages isn’t anything new. For years there has been research into actually following the fluctuation of communication online, but never as in depth as Dr. Xiaoyu Wang and Graduate Student Wenwen Dou are delving in with their project. This project consists of not only monitoring the Twitter feeds, but breaking them down so that the subject of each feed can be recorded. In this way, the two of them hope to perfect a technology that would allow the ability to figure out what exactly people are talking about and taking into account the hidden information within each message that most analysis just skips over.

Dr. Wang and Graduate Student Dou in front of the computer they demoed the Twitter Visualization for me in the Visualization Lab

By using software that they designed from scratch, the two of them are able to categorize the large groups of information. Visualizing all of these categories into colorful ribbons, the entire spectrum looks very similar to a rainbow, with over 20 different colors that fluctuate in width and height. Overall, it’s a beautiful sight. Besides the amazing pictures that it creates, this organizational method allows for extreme ease in operation of the software. By looking and scrolling over a single band of color, the words and subject that correspond to the color appear, allowing the user to see exactly what the subject is and what words are related to the subject. The words appear in a line below the visualization and each word shows up in a different sized font depending on how many different subjects the word appears in. For example, if the word “And” is within every subject, then it will be the largest word in all the subject headings. By clicking on a specific point in the color ribbon, the individual raw data appears in a pop-up graph, which allows the user to see precisely who said what about the specific subject.

A Screen shot of the colorful information ribbons

The most amazing aspect of Wang and Dou's work is the versatility and accuracy of the software. Though it’s being used for large scale visualization across Twitter and Facebook messages, it can be shrunk to be useful in smaller documents such as educational papers and website documents. It can also be used to study individuals or specific groups directly, rather than broad spectrum data gathering by analyzing the individual data and it can even use this to geographically pinpoint where the Tweets and Facebook messages are being sent from. The software is even able to decipher languages other than English, including Spanish and Italian. Dou is currently working on increasing the functionality by including Chinese.

There are a few institutions Dr. Wang, Dou and the Visualization Center are aiming this research towards. Charlotte Mecklenburg County Police Department and Bank of America are two of the main groups targeted to utilize this software. After the ‘flash mob’ incidents of last year in which groups of teenagers utilized tweets to organize public attacks on random strangers, watching social media has become far more important to the Police than ever before. As Dr. Wang describes it to me, using this technology to analyze and track suspect information would become easier than ever before, allowing police to work towards preventative measures rather than having to react to every action. Though not in the same manner, Bank of America would be able to utilize this Twitter Feed Visualization in similar means, being able to spot consumer trends and even work towards improving customer service by providing special offers to customers talking about making large purchases or complaining about current services.

As a technology enthusiast, this is all so fascinating. The world is only become smaller and better connected, and with that better tools and methods of analysis and research will be necessary, such as Wang and Dou's work. However, it's also a bit scary. Being monitored by employers on Facebook and Twitter was scary enough, but with this software, my bank can follow my every social move and complaint. Or the police can track down what I am doing with my friends by simply clicking on a colorful ribbon. Much like anything else new, it is scary and fascinating all the same.

Though there are further challenges to be met (currently, the project only utilizes 1% of the total Twitter feeds sent in 5 days), there is much promise in this technology. Dr. Wang and Dou are continuously working forward and took on student research assistants this summer as part of the university's REU research program. Stay tuned for further updates as the research progresses and stop by the Visualization Center's website.

1 comment:

  1. Information visualization Low Most of us believe it is a vey important for that life. Perhaps it will permit much more comprehension of each of our precise. And that means you More than likely that you simply permit much more powerful comprehension of this unique should you pay a visit to this fabulous website.Housing market forecast Low