The next Knowledge Management revolution
Share this content

Document management systems have brought about significant improvements in the way that businesses manage certain kinds of information. They are capable of some remarkably sophisticated processing at the mere push of a scanner button.

This article describes how Natural Language Processing can solve many of the problems that have plagued traditional knowledge management systems.

Paul Haley has worked with artificial intelligence for 25 years, and is uniquely qualified to comment on the present value and future utility of knowledge.

What document management systems do not do, however, is automate the knowledge embodied in the information that they manage.

Knowledge management (KM) systems, which attempt to identify and extract meaningful content from documents, have made significant advances in this direction, but automating knowledge has been a time-consuming, costly, and often error-prone process.

Knowledge takes different forms
Documents contain a wealth of information about a company and its products, markets and customers. But knowledge recorded as information in documents is passive. It requires people to search through it, read it, and then determine how best to use what they’ve read. With luck or effort, they remember the information and where it was located, so they can quickly access it again if they need it.

While these systems make workers more efficient, their effectiveness largely depends on the ability of the users to become extremely familiar with the systems and the kinds of information they contain. Document management systems have inherent limitations that prevent companies from achieving the true power of Knowledge Management.

Knowing the rules
A different kind of business knowledge exists in the collective experience of employees who understand the way their company does business. These people understand the ‘rules of engagement’. They know the set of conditions that a bank applies to grant or deny a mortgage, or the criteria that a company uses to apply volume discounts.

These business people have direct exposure to market dynamics and therefore make the decisions as to how the business needs to adapt to changing conditions. This collective form of business knowledge is known as business rules and business processes, and these are typically automated via programming logic contained in computerized information systems and business applications.

If automation is a cornerstone of business effectiveness, a concept that has been consistently validated since the dawn of the Industrial Revolution, then the real goal of Knowledge Management as a discipline is the automation of knowledge in all its forms. Document management systems automatically manage only certain kinds of business knowledge. For a company to achieve true KM, it needs to automate its business rules and business process knowledge as well.

These rules and processes may exist as written documents and may be managed as such, but it is only through the interaction of business people, systems analysts and programmers that this form of business knowledge is transformed into automated decision support systems and other applications. These systems use knowledge actively and automatically to make business decisions.

Knowledge Management systems have been improving the efficiency of businesses since they arrived on the IT scene in the 1980s.

Full automation of knowledge has numerous business benefits:
• streamlined workflow
• improved adaptability to change
• fewer resources and less time dedicated to capturing, processing, accessing and maintaining knowledge.

All of these elements combine to yield business efficiency, which in turn sustains a company’s competitive profile.

Arcane art of programming
Unfortunately, converting the knowledge of front-line business people into active, computer-based decision and operational support systems has been a time-consuming and error-prone task. Front-line business people do not have the time for or interest in understanding the arcane art of programming and therefore cannot directly create or manage the information systems they need and use. Conversely, programmers cannot be expected to have the in-depth understanding of the myriad activities that constitute daily commerce across the enterprise.

The only way these two important parties can work together is through an intermediary – the business or systems analyst. Any business person who has ever asked to have a computer application written knows one thing for certain: by the time the business knowledge is automated (in the form of a computer application) and is made available for use, market conditions will likely have changed and the business rules and business processes will need to be altered to adapt to the new conditions. Otherwise the knowledge system will be less than optimum.

Retooling the process of automating knowledge
The challenge then is to automate the business knowledge of front-line business people effectively. Because all business rules and processes can be written as documents, it is appropriate to further qualify the task and say that the ultimate objective is really to automate managed knowledge, i.e., the knowledge that is contained or could be contained in documents managed by a traditional document management system.

There are a number of reasons why automating (programming) this form of business knowledge is difficult. Much of the difficulty has to do with the interdependence of business rules and the strict hierarchy in which they need to be coded to properly model and support the company’s business requirements. Additionally, there are the considerations of complexity in constantly maintaining codified business rules, either adding, deleting or changing them, without adversely affecting rules already correctly coded.

Easing the load
If there were some way to minimize or even eliminate the intense manual effort involved in defining, designing, coding and testing systems that automate business rules, then a new level of business efficiency would be achieved. In fact, a combination of new and maturing technologies that are closely associated with the discipline of artificial intelligence is making the automation of managed knowledge possible.

Natural language parsing, inference engines, and automated generation of case- and rules-based programming code now enable ordinary business people, those with first-hand understanding of business rules, to both manage and automate those rules –- without intervention from programmers.

Automatic code generation
Automatic generation of business rules programming code has been around for some time and today exists in several forms. Many code generators work by supplying a structured form into which technology-oriented employees can enter conditions and parameters for business rules and processes.

In the early days of automatic code generation, the generated code was then implemented into operational systems by programmers, whose expertise was required to identify precisely where the code needed to be placed within existing applications in order to work properly.

The introduction of techniques developed in the world of artificial intelligence has begun to eliminate the requirement for experts to be involved in codifying business rules or for programmers to insert generated code into existing programs.

This is possible because algorithms have been developed that allow business rules to be created and accessed independently of other existing rules. More importantly, these algorithms allow large numbers of business rules to be established and managed without degrading the performance of the information system in which they reside.

Natural Language Processing – Now we’re talking business
One of the most promising developments in the automation of managed knowledge lies in the technology of Natural Language Processing (NLP). Although still in its infancy, significant advancements in the parsing of natural language statements – their analysis and breakdown into functional units that can be converted into machine language – are making it possible to enter business rules into a knowledge management system in forms that increasingly mimic normal spoken or written sentences.

NLP then maps sentences into programming code, which can be automatically incorporated into existing programming models – databases, object models, APIs and so on using new techniques. While it is beyond the scope of this article to describe the details of how this automated incorporation works, the important point is this: NLP technology allows business people to compose and implement business rules automatically by using declarative and imperative sentences, thus eliminating the need for experienced programmers to express business rules within an application.

Given a basic understanding of the grammar, vocabulary and programming models underlying an information system, business people can use a simple set of tools to actively interface with the next generation of rule- and case-based knowledge systems to implement their own business processes.

Just as important, this business knowledge can be added incrementally, without regard to positioning relative to other pieces of knowledge. This means that adjustments and refinements to business processes can be made quickly in response to changing business conditions. There is no need to scope and define entirely new systems.

Star Trek cometh
Even more exciting is the ability to parse spoken language, generate programming code and automatically incorporate the results into operational systems. This capability exists in rudimentary form now and in very short order will explode on the scene as the new channel for man-machine interaction. While we are years away from the seamless interaction between humans and computers as portrayed in the popular ‘Star Trek’ series, we are most assuredly in the very early stages of that transition.

Clearly, allowing business people to state their business knowledge orally and have it automatically reflected within a business process significantly improves their company’s ability to keep pace in the increasingly competitive global marketplace.

Revolutionizing business
Automating knowledge and its management should be a high priority for any business because of the need to constantly revisit and revise its processes, policies and practices in order to remain viable. Any technology that promotes decreased reaction time immediately becomes a strong candidate as a necessary business tool.

Combining technologies such as automated case- and rules-based code generation, inference engines, and Natural Language Processing to supplement the capabilities of traditional document-oriented KM systems is an enticing business proposition.

Enabling business people take direct ownership and control over managing and implementing business knowledge significantly improves a company’s ability to react to business dynamics. At the same time, IT efficiency is also improved because programming staff can be reallocated to other strategic and technical IT areas.

Make it so
Natural Language Processing technology will soon have business people talking business sense to their computers. It’s only a matter of time before Jean-Luc Picard’s signature line ‘make it so’ will be the way people leverage their business knowledge and make it an active part of the way their company does business.

About the author
Paul Haley has worked with Artificial Intelligence (AI) for over 25 years. As founder and president of The Haley Enterprise, Paul is uniquely qualified to comment on the strategic visions that many companies share concerning the present value and future utility of knowledge.


Please login or register to join the discussion.

There are currently no replies, be the first to post a reply.