Blab works by listening to 60 million conversions that take place on 50,000 websites. The company uses both text and image matching to detect a websites pages content.
Blab then categorizes the conversations it collects into different general topic areas such as sports and news. Blab claims that it can take patterns from one conversation and then use them to compare new patterns as they emerge. The platform then makes an educated guess at the conversations that will occur over the next three days.
Essentially Blab is a trending tool that aims to predict trends rather than simply reporting on them.
According to the company the process is so data-intensive that it had to build its own proprietary software that is running on Amazon’s Web Services.
The program is so server intensive that it literally died when running on Hadoop, a framework built to support large sums of data.
By predicting what conversations will trend in the coming days the company could become a big help for companies looking to strategize crisis response communications.
Blab monitors many websites including Facebook, Twitter, Tumblr, YouTube, Instagram, Flickr, and others.