Following the topic of big data, we take a look today at the highest data producers on the web, social media networks. These networks have to deal with a lot of information and subsequently have begun to show off big data solutions which process this data in interesting ways. Handling big data on social networks serves the same higher goal advertising does: catching the interests of the users. So let’s take a closer look at some of the recent solutions which have been released:
During a Twitter Hack Week, Isaac Hepworth created a visualization which represents mutual follows from over 50,000 famous Twitter users with a color scheme. The colors represent the five major categories on Twitter: News (blue), government and politics (purple), music (red), sports (yellow) and TV (green) (If you zoom in on the map very closely the actual names will appear http://zoom.it/kxNM#full). Needless to say the colors are aggregated, but more interestingly are the boundaries between different interests. Sport seems to be the all connecting category while government and politics together with news and music are more infrequent. Unsurprisingly, TV related users are connected closely to users from the sports and music category. All-in-all, the patterns show that interests which are similar are connected together, but that complementing interests also flow into each other.
In terms of Facebook, Graph Search is one future big data solution. As Facebook always concentrates on friends and relationships, is should not be surprising that Graph Search will do too. It summarizes interests, profile information and status updates to show which friends (and even friends of friends) match search categories. This feature aims to help people to unclutter the high amount of friends people have accumulated on Facebook and merge common interests more closely. Overall, it is Facebook’s goal is to not lose users, especially to special interest sites.
While Google offers tons of dashboards and statistics, one simple but revealing big data solution is the YouTube video trend map. Simply said, it shows the most viewed or shared YouTube videos on a map. The map can then be adapted depending on gender and age. Taking a closer look at what the whole population viewed the most on YouTube, there is no major difference between the videos. For instance, almost 80% shared one video showing a whale almost eating a diver. It’s a typical viral video which spreads across a large area quickly.
However, the map changes significantly if we switch the settings to the most shared videos. The shared videos vary in different regions. A good example is a video showing a Tornado intercepting a vehicle, which was shared more often in areas that are affected by storms or tornados.
Following and interpretation these patterns reveal some important facts. It is not only necessary to catch the user’s interest on a wider global level or through an interest category, but to also be aware of connected categories of interest. Being aware of favorite user hobbies and interests is a good start, as it catches them on a local level. Ultimately, advertisers can learn a lot from these social media patterns and solutions by simply observing, lessening the need for expensive market research.