RealTime IT News

IBM Crowd-Sources With Many Eyes

ORLANDO, Fla. -- IBM  launched a new social computing site today called Many Eyes, which allows users to upload very large data sets, choose different visual representations for the data sets, and engage in an online discussion of what the data reveals.

Each visualization will allow for an active discussion to take place and become a common area to share ideas, add insight and understand the visualization in a group setting.

Currently, the site presents a set of interactive visualizations that provide insight into a variety of topics -- from cereal nutrition data to the fertility rates of countries worldwide. Visitors can upload new data sets by cutting and pasting from their own Excel spreadsheets or tab/comma delimited text files and then create their own visualizations.

According to Irene Greif, director of collaborative user experiences at IBM Research, IBM launched Many Eyes in an attempt to learn whether the principles of crowd-sourcing can be applied to the analysis of visualized data, in the hopes of generating broader and deeper analysis of data.

The idea, Greif told internetnews.com, is to "start a conversation about things like data quality... The visualization lets you grok a lot more [data] at once."

The site is being launched as very much of a beta; for instance, it doesn't allow for tagging as a way of searching for particular data sets, but Greif said that this may be added in the future.

Greif said that Many Eyes is part of an ongoing effort to deliver social computing software to business users. The hope is that the site can help individuals and businesses use complex data to make smarter and more accurate decisions through visualization.

One potential use of this tool could be a government agency increasing its understanding of factors that may indicate potential recipients of governmental aid. Using data visualization, the agency could upload data sets and produce charts to determine how certain factors, such as living in an area frequented by natural disasters, income range, or possible layoffs in a certain job field, come together to provide significant predictive analysis.

But Greif noted that you always have to go back to the data because the visualization could be misleading. She said the best use of the site might well be a way of formulating hypotheses that would then have to be vetted through a thorough examination of the raw data sets.

She said IBM will also analyze user-behavior patterns to see whether people upload incorrect data and what the negative consequences of that might be.

"One of the things we'd like to answer is how you establish what's credible... Maybe that's the most important societal question," she said. "It's a fairly adventurous research project."