Q Engine

Every Q product, whether in use or under development, has three distinct components: (1) the Q engine, (2) a knowledge base that Q uses to read and snythesize information, and (3) a graphic user interface (GUI).

The Q engine is the processor that enables the knowledge base to be used to analyze text. While the knowledge base and the GUI change in each application, The Q engine is essentially invariant in various applications. Patterns are taken care of in an easily maintained data file using regular expressions.

The Knowledge Base

The knowledge base is another critical component of Q. It must be designed and engineered to reflect how experts think about their specific enterprise. The quality of this knowledge engineering (KE) fundamentally determines the quality of the Q application, just as the quality of human experts determines their effectiveness in problem solving. Enterprise knowledge bases clearly differ in content, size, and structure as well as in the special patterns, e.g. special symbols and formats that enable Q to read documents effectively.

The knowledge base captures and preserves enterprise knowledge and should be built deliberately and carefully. Its creation requires collaboration between enterprise subject matter experts (those who understand what is being found and why) and "knowledge engineers" (those who understand how to represent what is being found and why in the knowledge base). The difficulty and cost of the knowledge base depends on the scope and complexity of the knowledge domain that is being captured. For example, simply identifying common information protected by the privacy act, such as names, birth dates, social security numbers, etc., is relatively straightforward; however, complex information, such as the fuel capacity of an airplane inferred from the difference between empty and full weight of an airplane, is more difficult. The logical representation in both cases is similar; but the more variables and the wider the universe of language, the longer the process of constructing the knowledge base.

To assist a customer's experts in creating a knowledge base, AreteQ has developed a software program called the Knowledge Engineering Mapping Assistant (KEMA). This software will allow the user to build and maintain the knowledge base--in fact, as the knowledge base is "matured" by use and by inevitable corrections, the entire product becomes more effective. In a sense, the product "learns" through experience, just as its users better learn to represent their own knowledge in the logical network. Initially, however, in most cases, AreteQ experts need to train and to assist enterprise experts and other personnel to develop knowledge engineering (KE) capability quickly. The goal, however, is to turn the knowledge base over to those using it to maintain and create other knowledge bases as needed.

The KEMA makes the user the owner of the product because it allows the owner to add to the knowledge base, delete from it, add additional inferences, associations, etc. AreteQ will provide training, will help users set up an initial knowledge base, and will provide consultation and upgrades as needed; however, AreteQ's business plan does not include the long-term dependence of a user on AreteQ's knowledge engineering skills. The user designs, creates, and maintains its own product (with a little initial help in getting started).

Graphic User Interfaces

The interface of each program varies with the user tasks. In every instance, the GUI is designed to do a specific job. For example, the interface for classification of text documents at the Y-12 Nuclear Facility not only shows the user any offending text but also gives the reason for the offense, the category and level of the offense (classified restricted data, secret restricted data, etc.), and the concepts in the text identified as "triggers." This GUI is being extended to provide the sigma number (need-to-know) as well.

On the other hand, in the interface for e-FOIA such extensive information in most cases would only clutter the screen, if it were relevant at all. The e-FOIA screen simply shows the offending words in blue and allows the reviewer click on the word to redact (delete) the word or passage, to leave it untouched, or to select something else.

Functionality of Q >>