Why a Cognitive Computing Approach Supports Business Change
The advantage of a cognitive approach is that it starts with data rather than business logic.
- By Judith Hurwitz
- March 3, 2016
Cognitive computing isn't like the traditional computing solutions and traditional analytics approaches you're used to. Before I explain those distinct differences, it's best to start with a definition: Cognitive computing is a technological approach that enables humans to collaborate with machines. It enables data to be analyzed in context based on a variety of data including text, images, voice, sensors, and video.
A cognitive system is designed based on three capabilities:
- A cognitive system learns based on the data ingested. Through the learning process, the system makes inferences about the area it's analyzing.
- A cognitive system requires a model or a representation of a domain. The model has to understand the context of the data being used.
- A cognitive system must be able to generate hypotheses. To make sense of the data, a system is required to come up with an assumption or explanation of an expected outcome.
The elements of a cognitive system have a common requirement: that enough data be ingested, analyzed, and tested in order to see if assumptions can be supported. Therefore, one of the benefits of a cognitive system is that a bias must be supported by data. If a system is fed enough of the right data, it is possible to determine if a hypothesis holds up to scrutiny.
The value of a cognitive approach to computing is that the solution design is determined by the patterns within the data rather than by predetermined logic. However, in a cognitive system, it is not just data in isolation. A cognitive computing system requires the knowledge of content experts to curate the data and provide insights into the right data sources. Therefore, a cognitive system only works when there is collaboration between human experts and data that supports the way they put their data to work.
What does it mean to create a cognitive application? It is not that much different from creating any application in many ways. You have to start by understanding the objective. For example, you might want an application that will help consumers plan a trip or you want an application that helps anticipate and avoid security problems in software.
However, unlike a traditional application development process that is designed to answer specific predetermined questions, a cognitive application is intended to go deeper and explore context. In a cognitive application, the intent is to look at the relationships between data elements by creating a specific domain related to the area of focus.
By defining a narrow domain, you can then begin to determine which data sources are the most important. This data is then put into a corpus -- a machine-readable representation of the complete record of a topic -- but that is only the beginning of the process. What makes a cognitive system powerful is that the data is not static. The system learns from the interactions with both experts and users. Through an initial testing process, experts determine if this corpus provides the right context to come up with answers and solutions. If the context is wrong, it is likely that you need to work with new and different data. The power of the cognitive system comes from its ability to iterate on the data. Over time, the system learns and changes based on the interactions between machine and human.
The advantage of a cognitive approach to solution development is that it begins with data rather than assumptions about business logic. Through machine-learning approaches and the collaboration between humans and machines, the cognitive system is able to evolve as the business changes. A fast-changing world requires that organizations understand the hidden patterns and anomalies in data so that organizations are prepared for shifts in customer needs.
About the Author
Judith S. Hurwitz is president and CEO of Hurwitz & Associates, LLC, a research and consulting firm focused on emerging technology including big data, cognitive computing, cloud computing, service management, software development, and security and governance. She is a technology strategist, consultant, thought leader, and author. In 2015, Hurwitz coauthored Cognitive Computing and Big Data Analytics (Wiley, 2015). A pioneer in anticipating technology innovation and adoption, Judith has served as a trusted advisor to many industry leaders over the years. Judith has helped these companies make the transition to a new business model focused on the business value of emerging platforms. She was the founder of CycleBridge, a life sciences software consulting firm and Hurwitz Group, a research and consulting firm. She has worked in various corporations including Apollo Computer, and John Hancock. Judith has written extensively about all aspects of enterprise and distributed software. Judith is a co-author of six “For Dummies” books including Big Data for Dummies and Hybrid Cloud for Dummies. In 2011, she authored Smart or Lucky? How Technology Leaders Turn Chance into Success (Jossey Bass, 2011).
Judith holds B.S. and M.S. degrees from Boston University. She serves on several advisory boards of emerging companies. She is on the board of directors of Boston University’s Alumni Council. Judith was named a distinguished alumnus at Boston University's College of Arts & Sciences in 2005. She is also a recipient of the 2005 Massachusetts Technology Leadership Council award.