Knowledge discovery: a better search than Search
Query transformations deserve special mention. Standard queries that leverage browser interfaces, as most user systems do today, are limited by the number of characters they can pass to index servers to approximately 400 characters for simple queries and under 300 characters for more comprehensive query representations. Yet, we have found that query strings of 1500 or greater (sometimes as high as 3000 to 4000 characters) are required to provide the descriptive boundary constraints to contain the definition of search concepts that maintain results within scope of the discovery concept patterns.

This can be achieved through parallel query transformations, issuing as many as 20 to 40 parallel queries per targeted search site per user concurrently through a multi-threaded Java-based Servlet for a single document kSig. Because JavaServlets issue light-thread processes, the requirement for massive scalability is maintained.

By fractally factoring (fractoring) results into a normalized hierarchy of concept components around common vocabulary concepts (for focusing ViewPoints to the user's discovery concepts), search results can be organized to the users' needs. That way they can replace the unproductive time required to organize results for decision making, with more productive time focused on reaping the benefits of well-informed decisions with automatically prepared post-decision support materials.

ViewPoint profiling: a fundamental operation
Concept pattern formations take place through focused common concept aggregations called Profiles using a focus document as the ViewPoint perspective on document collections. This process borrows from the architecture of first-order predicate calculus prevalent in logic-based modeling systems, Artificial Intelligence, and Ontologies through the Semantic Web movement (KIF, KQML, et. al.).

Because documents, as instruments of business knowledge transactions, reflect the mapping of concept usage to business need and impact, they mirror the usage patterns of knowledge concepts and their cause-effect impact on business when viewed in combinatory concept hierarchies that are focused in an organized form against targeted discovery concepts. This provides a powerful knowledge management metaphor for leveraging the value-added reuse of business concept patterns for managing the organization's bottom line.

Pattern aggregations can be expressed in terms of standard XML relationship schemas, which can be linked to performance criteria in corporate databases and data warehouses highlighting pattern-based rules that work better than other patterns. In effect, this paves the ground for automatic recognition and expression of best practices through rule-based behavioral impact via cause-effect-impact patterns. The patterns can be guided by abductive inferencing thresholds as articulated and managed automatically in meta-rule policy system documents.

In the next article in this series, we'll discuss a hyperspatial knowledge management application architecture using MVC Model2, and will examine the components of two applications: a Knowledge Organizer, and an Ontology/Taxonomy Manager.

David Gewirtz is the author of How To Save Jobs and Where Have All The Emails Gone? For more than 20 years, he has analyzed current, historical, and emerging issues relating to technology, competitiveness, and policy. David is the Editor-in-Chief of the ZATZ magazines, is the Cyberterrorism Advisor for the International Association for Counterterrorism and Security Professionals, and is a member of the instructional faculty at the University of California, Berkeley extension. He can be reached at david@zatz.com and you can follow him at http://www.twitter.com/DavidGewirtz.