The Semantic Advantage

December 12, 2007

Fun with technology haters

Filed under: knowledge management,semantic technology — Phil Murray @ 8:47 pm
Tags: ,

Recently I posted an extended opinion to the OKMDS (Ontology for Knowledge Management and Decision Support) forum (okmds-convene) about paying more attention to structured support of discussion as the source of good decisions in enterprises … and less attention to computer ontologies. (The full post follows.) I made the unpardonable mistake of asserting that such discussion can and should be supported by technology.

Well, the technology-haters came out of the woodwork to complain about the pitfalls of technological approaches to knowledge management … making posts to the forum [perhaps] written in a word-processor, sent via email, organized and published via Web-supported forum technology, and viewed by more than a few people in their Web browsers and email clients (or via an RSS feed). I’ll bet some of them saved the thread on their desktops where it was indexed automatically with Google Desktop or another desktop search engine.

I was warned about the dangers of “throwing technology” at a problem, of confusing knowledge management with the tools used for it, and of being biased by my lurid obsession with technology (“If you’re a hammer, everything looks like a nail.”).

Umm, er, has everyone forgotten that people identify their business needs, then choose, deploy, use, and manage the technologies to address them? I’ve noticed that my Nissan Murano works fine most of the time, as long as I put gas in it and avoid large trees. I was aware that it didn’t get 40 mpg when I bought it, but it was big enough to haul the stuff and people I needed to transport. And it satisfied my vanity, too.

On the other hand, has everyone forgotten that designers and developers of KM technology were inspired by work problems. (Well, OK, there have been more than a few who see the world as a Matrix-like playground.) Did their resulting creations provide good solutions to those problems? Not always. Do they make bad assumptions about how their solutions will be adopted and applied? Not infrequently. But given a choice between listening to delirious rants about epistemology and listening to how a technologist addressed a real problem, I always tune in to the latter … when I’m looking for answers that work.

The irony is, perhaps, that I’m not a technologist. Never wrote or sold software. Sure, I’ve written a few small Perl scripts. Wrote SGML DTDs and XML Schemas. Served up tons of documentation on computer software. Used hundreds of different applications for creating, managing, organizing, and publishing content. Examined hundreds more software applications for KM requirements, but rejected most of them as inadequate for those purposes.

I’m not ashamed of my technology bent. Many aspects of “managing knowledge” are potentially improved, if not always solved, by technology:

  • Providing scalable solutions in general.
  • Capturing and integrating the enterprise’s “knowledge.” (I’ll admit that it’s not “knowledge” in the dictionary sense. Then we get into another endless argument about “tacit” and “explicit” knowledge. I don’t want to go there, either.)
  • Serving as a collective memory.
  • Enabling detection of patterns and relationships in communications that are effectively undetectable to individual human brains.
  • Observing and quantifying the flow of information among people in organizations. (Think Social Network Analysis.)

And that’s just the short list, of course. You can’t make judgments about the effectiveness of technologies you don’t know about … or which don’t exist yet.

Folks, ya gotta examine the needs first … from as many different perspectives as possible. Even, God forbid, a “people-oriented” perspective. But first of all from a business perspective.


As a relative newbie to formal ontologies, I apologize in advance for imprecise terminology. You did ask for input from the KM community, right?

The Ontolog forum recently featured a wonderful discussion between John Sowa, Leo Obrst and others on standards for ontologies and the utility of ontologies in enterprises. (Ontolog forum, 17-nov-2007) The two distinguished experts disagreed gently on the topic, but it appears that they agree ontologies are important for solving enterprise problems.

I have to object … also gently … and with qualifications. The distinguished members of the ontology community have demonstrated that ontologies and other technologies that have evolved from the study of logic and language can be applied successfully to data-mining, interpretation of natural language, and even situation awareness requirements in military scenarios, but the recent buzz about ontologies is — like the recent buzz about metadata in general — largely a reaction to the superabundance of unstructured information. From the perspective of what enterprises need in general, this emphasis seems to be an overreaction.

I do believe that ontologies in general and the Semantic Web in particular will play huge roles in “semantic” aspects of business activities. However, overemphasizing ontologies and metadata distracts us from the most common enterprise activities: the core processes of building an opportunity (or a domain) and successfully managing the enterprise. In the knowledge-based enterprise, those processes consist primarily of identifying and evaluating facts and conditions — statements about market and physical realities. An enterprise or domain — for example, the extended NASA community — is the sum of all the conditions and responses for that enterprise.

Stated in a somewhat different way, decisions and and implementation particulars grow out of evaluation and acceptance (or rejection) of ad hoc assertions. By ad hoc assertions I mean statements that identify an opportunity or situation — for example, “The presence of Chinese and Japanese satellites around the Moon is a threat to our pre-eminence in space exploration.” Or “Nanotubes based on composites are the best choice for building a space elevator.”

By evaluation of those assertions, I mean such statements as, “This assertion is relevant (or valid).” Or, “This assertion must be expressed more precisely.” Other evaluations may consist of identifying (and quantifying) the impact of a set of assertions A on assertion B — an analog to applying If … andIf … Then logic to conditions in programming. But I’m not talking about precise programming statements. Evaluations are (or can be) collective, quantitative, “fuzzy,” qualitative, and/or arbitrary.

No matter what our particular role in a knowledge-driven organization may be, we communicate and evaluate assertions on a daily basis — in meetings, casual encounters, emails, personal note-taking, forums, and documentation of many types. These activities of the organization are frequent, pervasive, and vital to successful decision making and execution. In spite of that, there is no mandate to apply technology or business practices to making these activities more effective.

That’s unfortunate. Assertions and the evaluations of those assertions can be represented explicitly. They can be known — expressed as structured objects. They can be supported directly by technology, management practices, and education of workers in semantic principles. Decisions can be traced back to the conditions/assertions that influence those decisions.

Objects of this type and relationships among those objects can also be visualized easily. I am aware that software applications for specifying, visualizing, and evaluating assertions do exist. But those I have seen (like the Compendium Institute’s Compendium hypermedia tool) seem fundamentally disconnected from goals of precise representation of the meaning in natural language. They lack methods of formally representing assertions as objects that can be addressed with multiple tools, or they simply don’t scale well. Others simply don’t make the distinction between assertion and evaluation of assertions at all.

Supporting these core semantic activities of the organization should drive how knowledge-development, knowledge-representation, and information-management technologies are selected and implemented in enterprises — not vice versa. Similarly, the goal of enterprise strategies and personnel-management tactics should be, first and foremost, to make these core semantic activities as effective as possible.

Re-centering our focus on assertions and evaluation of assertions provides several important advantages:

  • Participants in the enterprise can deal with such assertions directly. Average Joes and Janes have a fundamental grasp of what constitutes an assertion — although they will need help in framing and evaluating assertions in a structured way. (It’s a lot like parsing a sentence. Not everyone’s favorite activity, but a lot easier than grasping subtleties of inheritance of semantic properties in ontologies.)
  • Participants can “see” the impact of their participation. Feedback is vital to participation. And a record of decision making can be kept and analyzed.
  • Evaluations can be weighted. An approximate sum of the evaluations of the quality of assertions — evaluations contributed by multiple stakeholders in the enterprise — should point to good decisions, especially as the number of relevant assertions and evaluations grows.
  • Participation in gathering and evaluation of assertions can be a source of objective information in performance evaluations.
  • Assertions become product. If you’re a software developer, you can see specifications emerge from assertions. (Currently, implementers have to leap that great chasm between unstructured descriptions of functionality and structured modeling of processes.) If you’re a tech writer, you can see that the rationale for features and functionality — and sometimes the behavior of features themselves — is captured in the assertion/evaluation process.

I want to stress that I’m not dismissing the importance of ontologies. Among other things, ontologies should support interpretation and management of assertions and evaluations. But we need to take a step back and re-center our pursuit of effective solutions for the challenges facing knowledge-based organizations. We have to ask,

  • What matters most to the participants in an organization?
  • What explicit information or “knowledge” most directly affects understanding and successful decision making?
  • What information is most directly relevant to a broad cross-section of people in the organization?
  • What information can people react to or evaluate with minimal education and effort?
  • How, in general, can participants most effectively contribute to improvement of the information that leads to success?

If this is yet another spin on the issues, I hope it is at least a more positive spin.

Phil Murray


Leave a Comment »

No comments yet.

RSS feed for comments on this post. TrackBack URI

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

Create a free website or blog at

%d bloggers like this: