A new blog from our friends at DataSift is a great example of how incredible social data analysis can be, and how simple mistakes can cause a big panic.
Every year the world spins closer to streams, allowing consumers everywhere to individually curate the media they come in contact with. This is making “headline news” an increasingly nebulous term. Outside of television (which will change soon enough) news consumers create different user experiences that make catching breaking news much more difficult.
Your Reddit homepage, Twitter news stream, and Facebook timeline is different from everyone else’s. The Bank of England realized this when they began analyzing social data in Scotland as they prepared for the independence vote earlier this year. They were worried that if Scotland voted for independence, there might be a run on the Royal Bank of Scotland. Ultimately, Scotland stayed in the loving arms of the United Kingdom and the banks were fine, but for a short time, it looked like the run might happen.
Kester Ford writes that the Bank of England “did have one jumpy moment when they noticed a spike in the volume of posts related to their keyword search.” It turns out that the Bank of England was looking for the keywords “run” and “RBS,” which is what was driving the spike.
At the same time that the Scottish vote was going on, across the pond in America, the Minnesota Vikings were playing a game. The posts were not about a run on the Royal Bank of Scotland, but rather about the runs being made by the Viking “RBs” (shorthand for “running backs”). The Bank of England executed a quick fix – making their keywords case-sensitive – and thus their results returned to “normal.” Luckily, “beer me” and “these wings are bomb” are not banking terms, or the confusion could have lasted all day.
As we all learn how to better navigate this swelling ocean of big data, companies like DataSift are working hard to develop better ways to “see how language is really used on social media.”
Using computers or software that can “think” independently is one way to ensure mishaps like the running backs versus the Royal Bank of Scotland don’t happen in the future. People won’t need to perfect their use of keywords or set careful parameters, the analysis tools will determine all that autonomously.
Listening to and understanding this cascade of content in real time and with total accuracy will continue to be challenging as we diffuse our data through more and more streams. However, DataSift’s very granular and modular ‘sifting’ functions from across a wide range of social and web input feeds will lead this innovative charge. Augmenting this with Lexalytics’ sentiment analysis, storage and analytics will help to build an unrivaled service platform.