Knowledge management professionals help businesses improve performance by formalizing the understanding of what the business is accomplishing, how well it is doing and what is impeding the business both internally and externally. Knowledge management is a wide-ranging discipline, looking at the business in terms of requirements, data collection and improvements.

Many different approaches to knowledge management exist, and passionate professionals differ on fundamental approaches, discovery and measurement disciplines and prescriptions for improvement. However, one unifying principle is the recognition that all businesses change and that improvement requires embracing change.

This article examines two approaches to improving a business and discusses how these approaches - business process management (BPM) and semantic Web technology - may be used together to improve business agility in the face of change.

Business Process Management

A business process is a set of operations or tasks that together accomplish a particular business goal, and BPM is the systematic study and optimization of business processes.

BPM has a long history, with roots in the rise of manufacturing in the early industrial age. The focus of most recent work is the definition and optimization of processes using computer software to handle some or all parts of a BPM lifecycle consisting of process design, modeling, execution, monitoring and optimization.

An example of a process in this modern sense is opening a bank account, which requires interactions among multiple participants, the exchange and storage of data in systems, the specific activities and decisions required by the bank’s business goals and policies as well as by external requirements, such as government regulations.

An organization can treat BPM primarily as a process modeling and improvement discipline, using software to model, measure and optimize processes. Alternatively, an organization can use specialized software, such as a business process management suite (BPMS), to manage processes as objects in a system. These suites provide process control and execution and tie the process to multiple applications and associated data.

The choice of tools and focus is driven by business goals for BPM. Understanding and measuring processes may give way to monitoring and improving processes. At its most fruitful, BPM is part of business transformation, allowing a business to substantially improve efficiency, productivity and customer satisfaction.

According to Gartner, a BPMS must provide 10 areas of functionality, including a process execution engine, a design environment, a registry/repository, business rules management and a set of tools to manage process lifecycle.1

Most BPMS tools place both models and live process data in the registry/repository, and most treat the process model itself as the central design artifact. This means the data and metadata collected in a BPMS revolve around a process as an object to be controlled and optimized. Other important participants in a process include people, systems and documents, and these are modeled (and data is collected) in the BPMS as well.

The focus on the process of a BPMS often puts a prescriptive and sequential spin on data modeling. However, this is somewhat countered by the inclusion of business rules, their focus on policy as well as a descriptive or declarative approach to business or process decisions and the data needed to make them.

Businesses often use BPM as part of a services initiative to improve both business and IT agility, encapsulating processes, policies and services to improve reuse, gain flexibility and meet rapidly changing business requirements.

Semantic Web Technology

Semantics is the study of meaning in communication. Semantic data models, or ontologies, are formal models of domains of interest and have been explored for more than 25 years as part of artificial intelligence and knowledge management for computer reasoning, data classification and normalization and linguistic and text analysis.

The “semantic Web” was a term coined by Tim Berners-Lee in a paper of the same name in 2001.2 Berners-Lee envisioned autonomous agents responding to situations based on well-defined requirements, both in the physical world and on the Web. Much of his vision is unrealized, but the computer-related ideas have evolved and merged with work on ontologies into a set of standards and software used both to describe Web pages and relationships as well as describe, exchange and reason against semantic data outside of the Web.

The core standards for the semantic Web, managed as part of the World Wide Web Consortium (W3C) semantic Web activity, include the resource description framework (RDF). The framework defines abstract statements and provides an XML-based implementation language and Web ontology language (OWL), which extends RDF to define ontologies.3

OWL and RDF implement an object-relational model allowing creation of a directed graph, a network of objects and relationships describing data. This model is potentially richer and more flexible than traditional data models used to implement relational databases.

RDF represents the English language statement, “Bob Jones works for Acme Corporation” as a single statement, where “Bob Jones” is the subject, “works for” is a predicate (also called attribute or relationship) and “Acme Corporation” is the object of the statement. This subject-predicate object pattern is called a triple, and it forms the basis of both RDF and OWL. In this context, both “Bob Jones” and “Acme Corporation” might be the names of abstract objects with numerous other statements representing attributes of the objects (age, location and whatever else is relevant to the particular systems).

OWL includes capabilities to restrict and extend object definitions, using classes and subclasses to define objects and their relations. Software might use OWL’s definitions to derive new information by inference against OWL and RDF data. For example, a shipping ontology that understands location as a concept might include hierarchies of locations like street/city/state/country, and shipping software using this ontology could derive additional location information about a package (“If the package is in Boston, then it is also in Massachusetts.”) while tracking that package to its destination.

OWL includes description logic capabilities that define relations for the reasoning described above, but the reasoning itself must be done by specialized applications called reasoners or by general purpose rules engines.

OWL and RDF can be used to model any discipline at any level of detail, but this is a double-edged sword. Knowledge management practitioners may take a formalistic and intellectually rigorous approach to building ontologies that requires a huge amount of time, research and complex data representations.

A business user may want to use semantic Web concepts to model and reason against real-world data, such as forms, databases and documents, and might want a working system in three months rather than three years. Businesses are typically interested in semantic Web technology as an extension to information modeling initiatives in order to bridge data and business silos and to improve and rationalize applications.

Knowledge Models and Rules

Both BPM and semantic Web technology use models to describe problem domains and rules to implement important behaviors. But, as previously discussed, these technologies use quite different approaches to problem solving.

Semantic Web technology expands traditional data architecture’s focus on explicit data models to include additional semantics around meaning, relationships and restrictions, allowing software to automatically perform additional actions such as mapping, classification, validation and inferencing.

Rules play an important part in semantic Web technology because they represent the muscle for inferencing and reasoning. As Tim Berners-Lee said when announcing the formation of the W3C’s Rule Interchange Format (RIF) Working Group, “Rules constitute a key element of the semantic Web vision, allowing integration, derivation and transformation of data from multiple sources in a distributed, transparent and scalable manner.”4 Unfortunately, rules standards are immature, and standards bodies and practitioners have fundamental disagreements over what should be included and what standardization goals are.

BPM focuses on making process models explicit, with enough data definition to support user interfaces, integration to data sources and basic decisions. Interestingly, business rules have become a required component for an advanced BPMS. Unless BPM can include decisions and policies, it cannot address the complexity of real business. Rules engines, whether integrated or separate, allow BPMS systems to make policies and decisions explicit along with processes and work.

Synthesizing BPM and Semantic Web Technology

How might these two approaches, based on different models and different methodologies, be used together to improve business agility? Two different approaches include:

  1. Incorporate BPM methodology and models explicitly into a semantic Web-based system.
  2. Incorporate semantic Web models and methodology into a BPMS.

For the first approach, a developer might create a model that incorporates process concepts such as participants, events, cases and decisions. If the model is rich enough, a semantic reasoner might respond to incoming RDF messages by classifying events, generating case models and enriching models through inference. But such a system would have to include integration capabilities and detailed instructions in the models for how to change process state and resolve work. This approach is beyond the capabilities of most semantic Web-based software, although there are a few commercial systems that include rules engines, integration buses and other technology to construct purpose-built systems including limited process capabilities.
For the second approach, a BPMS system might use OWL and RDF as data models for case and process data in the registry/repository and incorporate a semantic reasoner as part of the BPMS. A more limited but easier approach might be to pick one or more items from a “menu” of semantic integration possibilities, including:

  • Semantic data mapping, using OWL and RDF to describe existing data sources and manage data interchange between the BPMS and other systems, would be useful when incorporating data modeling and rationalization efforts into a BPMS.
  • Semantic event classification, allowing a BPMS to respond flexibly to a range of events by inferring meaning from an event or series of events, might be useful in tracking suspicious activities, managing SLAs and responding to business changes.
  • Semantic inference might allowa reasoner (or a rules engine using translated versions of semantic models) to enrich data, drive research requests to people or systems or derive structured information from unstructured data (text fields in forms and messages or business documents) as part of process management and resolution.

BPM and semantic Web technology provide rich environments and tools to help businesses solve real problems in responding to changing business conditions.
BPMS is a maturing software sector with well-understood benefits and a good choice of software, but with just enough focus on data and information management to execute processes. Bringing semantic Web technology into a BPM practice offers rigorous and explicit data modeling and allows integrated rules engines to reason, classify and improve data quality in the BPMS.

The semantic Web offers rich and mature data standards but immature business software and less sophisticated integration with existing business goals and IT systems. As additional semantically oriented standards and capabilities (such as OWL-S for semantic services and SPARQL for semantic queries) are incorporated with rules engines into robust integrated packages, knowledge management and business users will be able to implement data-oriented processes using semantic Web tools.

Existing standards for BPM and rules are immature and do not play well with the more mature standard set for the semantic Web. Berners-Lee, in his RIF announcement referenced previously, said, “A Rule Interchange Format will, for example, help businesses find new customers, doctors validate prescriptions and banks process loan applications.” This is the promise of an advanced BPMS, and standards committees could help reduce silo standards. If enlightened vendors can deliver merged BPM and semantic Web capabilities with a strong integrated rules approach, businesses will adopt the solution as their next generation data and process architecture.

References:

  1. Janelle B. Hill, Michele Cantara, Eric Deitert and Marc Kerremans. “Magic Quadrant for Business Process Management Suites 2007.” Gartner Research, December 14, 2007.
  2. Tim Berners-Lee, James Hendler and Ora Lassila. “The Semantic Web.” Scientific American, May 17, 2001.
  3. World Wide Web Consortium. “W3C Semantic Web Activity,” www.w3.org/2001/sw, July 21, 2008.
  4. XML News Desk. Sys Con France, November 15, 2005.

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access