Uppd8: An Engine for the Wisdom of Crowds
What people think matters. Uppd8 focuses on crowd sentiment analysis and provides tag-scored data based on different user types. Basic services will be provided for free.
To make a safer decision, most enterprise users prefer to seek advice from others. They go to bloggers and read what bloggers are telling them or to industry analysts such as Gartner and Forrester. Uppd8 found a business opportunity from it. It is a place where you could see how the entire crowd thinks about the technology, also the most useful information about it by your role.
The first enterprise topic that Uppd8 has addressed is Big Data. Its engine analyzes the sentiment of Big Data content items that were pre-processed, in order to screen irrelevant or less meaningful items.
“Enterprise users want to know what others are doing, but there was no actual technology that shows them that”, said David Barzilai, who is the CEO of Uppd8. “Today they research manually and seek the wisdom of individual industry experts. What we want is to enable users to realize what the crowd sentiment about the technology is and deep dive the information which matches their requirement.”
I will illustrate how it works with an example of SQL and NoSQL. SQL tag is used mainly with traditional databases, while NoSQL is for new non-relational databases. If I want to see what is the crowd sentiment of NoSQL versus SQL, I add SQL as the first tag and NoSQL as the second tag. The sentiment analysis result for the past 12 months is shown below.
There are three main parts. On the left side bar, we can choose time periods of analysis, user type, and data type from articles, blogs and vendors. User type is selected from developer, IT and Executive when you register to the system and can be changed by user selection, while using the system.
The line graph shows the crowd sentiment trends for my tags. They can be used as Tag tickers (similar concept to stock tickers), where the trend is affected from the sentiment of an aggregated group of articles in every time slot. You can click on every high or low point of the trend graph to move to the next page (Learn) to review the highest scored articles for your user type, in order to get high quality information that matches your user type.
The bar graph of data volume on the bottom shows you how much data was published about SQL and NoSQL for each of the time period. Why is it important? “Because sometimes it could be a lot of enthusiastic feedbacks. If it is a small amount of data, it doesn’t mean too much”, said David. An interesting fact from above figure is that different directions start to occur when the volume starts to increase.
Uppd8 provides more than a few figures. If I click on the "Learn" link on the page header, it will show all the articles for the tag scored by the quality for my user type. According to David, the idea of this is to make sure the users get the most useful information without working too hard. We can also see the sentiment of each article by the arrows on the right. The purple arrow which goes up shows that this is a positive article for SQL. The one underneath with an orange arrow going up means that it says good things about NoSQL.
Uppd8 current plans are to provide all of above at no cost. Content scoring by user type is the basis for the entire system. Uppd8 believes that unlike consumer-targeted content, where individuals are interested in personalized information, enterprise users share similar needs by user type. Furthermore, it will enable user communities around the Tag-scored data, as well as around the Industry experts that can leverage the system to communicate with Tag-interested audiences.
It provides us with an alternative access to what others think, while different from the past. “This is not newspaper. This is a crowd-driven research engine.” David said.
I am looking forward to what the future holds for Uppd8.