This article was original posted on LinkedIn.

This article is co-authored with Kevin Nguyen. The opinions shared in this article are our own and do not reflect those of our employer(s).

What is a cognitive enterprise? #

The short version #

Cognitive enterprise is a data-driven organisation that has a thorough understanding of the technology between different business units and why they exist, i.e., what they can do for their users and their customers. The crux of a cognitive enterprise is one that is dependent on data, the knowledge creation from that data, and application of that data in an outcome-based business context.

The long version #

Since the 1960s, there have always been two trends of data management: knowledge representation and enterprise data management. Knowledge representation has its roots in AI research with emphasis on the meaning of data and metadata. Enterprise data management started with database research with emphasis on application of data to business problems.

In practice, we see that junior to senior technologists and business users are familiar with the use of databases along with how to store domain knowledge in these databases. This is not the case with knowledge representation, however, where more rigour is needed to capture knowledge and organise it to infer new knowledge. While it’s less talked about, we do see use cases popping up from large organisations from time to time — through conferences, white papers and demonstrations, such as:

Today knowledge representation and enterprise data management are getting woven together to form a data fabric. This data fabric is the architecture that forms the basis to a cognitive enterprise.

Technology advances for a cognitive enterprise #

Weaving the data fabric and building the cognitive enterprise is a community effort, with significant integration requirements. Earlier this year, O’Reilly published The Rise of the Knowledge Graph, which at a high level, highlights three key technology advances that enable this architecture: distributed data, semantic metadata and connected data.

Distributed data refers to data shared throughout the organisation unbounded by geography. Semantic metadata tells us what the data means, its relationships to other data and helps us infer new connections. Connected data refers to how no datum ever stands on its own, where meaning is derived with other data.

For greenfield projects, the data fabric will likely be built on triple stores and graph databases with built-in reasoning, while existing data sources will require a semantic metadata layer applied on top to facilitate querying and reasoning. If we were to map the tools and standards today to these three dimensions, we’ll soon find ourselves in a buffet with too many options to choose from and with many more to come. Partly thanks to the open source movement.

Distributed data #

Semantic metadata #

Connected data #

  • Knowledge representation: interestingly, everything from the semantic metadata’s enterprise knowledge management fit very well here
  • Enterprise knowledge management: IBM DataStage, Trino, Apache Beam, etc.

Looking at this through a pure data lens is interesting but it’s important to point out that data aren’t collected and live on their own. They need to be used and meshed with the rest of the enterprise. With the increasing requirement for real-time use cases, a well-rounded automation, integration and application development tool suite based on a portable platform and skilled technologists are needed to ultimately create simple customer facing applications and drive business results.

Business & IT in the cognitive enterprise #

Today, we often see a segregation between business and IT. IT gets less money each year while still need to maintain operational and grow with new use cases, while business users get blasted with marketing materials from every vendor. To effectively innovate, business need to have an understanding of the possible and discern hype from reality.

However, saying that doesn’t mean the biz-tech world isn’t changing for the better. Just like the DevOps transformation that has been raging on in the IT world, we see more and more technology-inspired methodologies being adopted in the business world, such as Agile (SAFe, LESS), Design Thinking and domain-driven design exercises like event storming. Slowly, but surely, towards a BizDevOps culture.

References #