How Web 2.0 Can Enhance Data

Cyril Brooks raises some interesting points about business intelligence and Web 2.0 in this article on his blog, Cyril on Business Intelligence. While Cyril doubts that Web 2.0 will be at the heart of business intelligence, he does mention something that we profoundly believe in at Swivel:

For BI, the critical element... is that Web 2.0 enables collaborative commenting and assessing significance of facts - especially numbers that may or may not be important.

At Swivel, I've seen tremendous value in providing a space for people to work together with data.  Here are some examples from our Web site that back-up Cyril's point:

• There is as much information about the skyrocketing price of corn in the comments below this graph as in the numbers themselves.
• There are many great examples of the community identifying unimportant numbers at Swivel. For instance, the many people who pointed out the flaw in the analysis of the relationship between wine and crime found here.
• Finally, working together with data is more than just making comments about graphs.  It's about discovery and analysis too.  For example, user suzanucb was able to take two data sets from different areas of Swivel and combine them to highlight a relationship that has subsequently been viewed over 2,000 times.

As I know from my career prior to Swivel as a data analyst, collecting and transforming data is hard, expensive work. This is why, when those transformations are done, it's important to get as much value out of them as possible. Web 2.0 software, which allows numbers to be shared, discussed, and analyzed by as wide an audience as possible, can make a huge difference in turning data into understanding and understanding into action.

As Cyril puts it in his conclusion, "[combining] the numbers the opinions, assessments, will work well with fresh brains and lower inhibition thresholds." This has absolutely been my experience at Swivel, and, in the spirit of Web 2.0, I hope it's an experience others can share.

Comments

[Hadley]

About the most that you can see from that last graph is that both gdp and debt are increasing exponentially. A graph of dept/gdb vs time is far more revealing, but I can't see how to create it in swivel.

I drew the graph in R - but had to tweak the csv file to get it to work. Missing values don't seem to get a comma, and the year gets repeated 3 times. The output csv looks very different from the table view.

[Hadley]

To add a little more about what you can see with a better graph: a big upswing in relative dept from 1941-1945 (WWII presumably), then a fairly steady decay (with the exception of the outlier of 1953) until the start of the 80s when relative debt starts to rise again. There's a bit of a drop between 1995 and 2000, and then it rises once more.

[swiveldoubter]

How can anyone vouch for the veracity of the data used in these graphs. Firstly, the interpretation could be incorrect. The naming of the column could be incorrect. We know how wikipedia survives because all people have access to the information and it gets corroborated. But here these graphs use data points that could be incomplete, inconsistent or plain wrong.

[regülatör]

Thanks nice site mee to