About 500 million tweets are posted every day. Add in the thousands of blog posts, forum responses and comments also posted on daily basis and then you can understand the depth of information that Kairos processes for our users. As we roll out increasingly complex updates to our product, we wanted to pull back the curtain and explain how we break down those millions of posts into the digestible, actionable data you see on Kairos today.
Just like the human brain takes in outside information constantly, so is Kairos at an increasingly speedy pace. This incoming data must be handled with speed and scalability in order to extract the complex meaning Kairos delivers to our customers every day. So, when a post enters our Complex Event Pipe (able to detect patters of activity from multiple data streams in real time), Kairos analyzes it on five different levels: machine learning, linguistics, emotions, keyword matching and audience. As the data moves through each of these five layers of classification, the brain works to detect within the post different sub-themes from a selection of 25,000 existing classifiers and marks the post as a match against each signal. This process is known as enrichment.
The enriched data file now includes the content of the original post plus all of the signals that Kairos was able to detect within it. For example, we might find a tweet that discusses an expensive grocery bill. After this post is enriched, it’s marked as being a tweet from a mother who is a healthy eater expressing frustration toward Whole Foods. What this means in practice, however, is that the millions of data points that enter Kairos as social posts multiply to billions of data points including extensive context. That’s a lot of data, and we recently improved the way we store and analyze it to benefit our users.
With easier, faster access to even more accurate data, our users can look forward to improved analytics in Kairos to achieve the best possible understanding of their brands’ key audiences, their emotions and their affinities as expressed on social media. Stay tuned for two additional blog posts on the other building blocks of our technology: taxonomy and predictive analytics.