The IBM Advantage for Cognitive Discovery Cloud Architecture

Data has become the fuel of business innovation in proportion to the increase in the
amount of data available. Sensors, video, news and social media streams, and weather data
are only a few of the sources of data available to an enterprise, in addition to their private
stores. The organization that is able to tap these sources, separate out the valuable
information from the noise, see relationships and patterns in the data, and then act upon
this knowledge is best prepared to overtake their competitors.


Traditional approaches to data analytics and knowledge management typically help with
specific kinds of tasks that are related to structured data. The sheer amount of
unstructured data being produced means that human physical capacity is quickly
overwhelmed by the effort to collect and curate it. New techniques that use natural
language processing, visual recognition, and other elements of artificial intelligence can
help in identifying and organizing unstructured data. This is where cognitive computing
comes in. IBM’s cognitive services are trained by humans to augment and amplify human
cognition. The systems are not designed to replace a human’s cognitive capabilities but to
enhance them. For example, a system trained by a legal expert to sort through thousands of
files of unstructured data to identify those pertinent to legal claims can do it faster than a
person, freeing up the expert for higher value activities.


Cognitive systems can be transformative. A business can change how it operates when the
proprietary content and expert knowledge of the organization are extended into the
enterprise through applications that include natural language processing, hypothesis
generation, and evidence-based learning. Strategic and day-to-day decisions are better

informed, leading to better business outcomes. Best practices encourage the use and
embedding of cognitive decision making into existing processes and into the creation of
new processes.


The tools used to achieve these results have evolved to meet the expectations of the
enterprise. Both highly structured and unstructured data must be used. Especially in the
text-heavy, unstructured data domain, there is a natural and cumulative evolution from
basic search to cognitive search through natural language processing and machine learning,
with the goal of delivering deeper insights more accurately, faster, and at a greater scale.
IBM® Watson® Discovery is designed to make it more efficient to identify, collect, and
curate text-heavy unstructured and structured data. This can simplify human use of
information through more efficient access to large content stores or through the
integration of the service in support of larger cognitive systems.


Prior to the availability of natural language processing and contextual search applications,
keyword searches were the way users engaged with masses of information. Previous
approaches to the enterprise management of information, launched under the banner of
knowledge management, relied on the creation of complex content topologies, huge
internal indices, and the speed of the keyword search. These projects were not adopted
widely due to the level of effort required for basic results. The table below shows how
value to business increases with the adoption of more sophisticated techniques for search
and analysis.

 Data Center

Share content on email

Share