Traditional analytics focuses in unlocking insight from the structured, or “clean” data. It relies on matching and classifying standard text fields and performing calculations of number values. But this is often under 40% of all the raw data available, leaving more than half of all data – unstructured textual information – completely excluded from analysis.
One way of viewing this is as a missed opportunity. Clearly, knowledge extraction is not being maximized. But also cases exist where key information becomes obscured, leading to what is effectively the wrong analytical conclusions. And there is no need to explain how dangerous this could be.
What is cognitive analytics?
Cognitive computing in general is the group of technology products based on the scientific disciplines of Artificial Intelligence and Signal Processing. They include machine learning, and natural language processing among others, and are a key player in increasing the usefulness of all the data businesses collect and store.
On one hand, natural language processing can enable the transformation of unstructured textual data, such as comment fields, descriptions, or notes into intelligent classifications or taxonomies that can then be used in enriching existing analytical techniques.
Machine learning, on the other hand, can make such a system learn from itself with time, becoming more and more precise in its understanding of the content of these unstructured data fields, and of its relevance to the rest of the contextual information.
What is the potential for incident management?
In handling incident tickets, a lot of information is contained in the free-form description of an issue, or the notes taken during the diagnostics or resolution phases of the incident lifecycle. This knowledge can allude traditional analytical investigation of ticket data, and in turn, misdirect any efforts for using this data to guide operational decision-making, or making improvements to the process of managing incidents or outages.
Naturally, the first benefit of enriching an analytical approach with cognitive capacities would be to increase the degree of certainty that the analysis is correct – just because more data is used to back up the conclusions. Also, it finally derives value from all the money organizations spend on collecting and storing data that was previously unused.
The second, less trivial benefit of cognitive analysis is the possibility for a new, more intuitive way of interacting with this data. Just as its name suggests, cognitive functionalities mimic the way the user thinks – meaning that smart algorithms will provide even more timely and actionable information to the user.
Cupenya makes first steps with IBM Watson
A first look of these functionalities and how they find their place in Cupenya’s process analytics solution was demonstrated on May 31 at the IBM Connect event in Utrecht, The Netherlands. The showcase included a demonstration of our first experiments with IBM’s Cognitive Computing platform Watson and its Alchemy API in using unstructured incident ticket data for analysis, pattern matching and recommendations. To see it for yourself, sign up for a personal demonstration here.