Compsim's Technology

Compsim's patented technology focuses on the packaging of human expertise in software applications and embedded devices. Unlike other approaches like fuzzy logic and neural nets, solutions incorporating Compsim's KEEL (Knowledge Enhanced Electronic Logic) technology can create explainable decisions. For example: This action was taken because.....

While other approaches may have been approached by taking computers and asking them to "think", Compsim's technology was created by evaluating how humans think and make decisions and then modeling a computer to perform the same functions. Compsim Management Tools (CMT) was developed to assist humans in an effort to capture their knowledge and make structured decisions in a business environment.

If you or your organizations are interested in advanced cognitive capabilities with the following features, then you might be interested in KEEL Technology.

KEEL Technology can:

allow domain experts (not mathematicians or software engineers) to create solutions to problems that they understand thus greatly reducing development costs
allow complex information fusion problems to be addressed in days (rather than weeks, months, and years), thus controlling development costs and getting solutions deployed faster than alternative approaches
be deployed in small embedded devices giving them the ability to process policies (exercise human-like judgment) that are more capable than brittle rule-based systems
allow policies to be created that describe "how to solve problems" rather than creating solutions to explicit problems, thus making systems incorporating this technology more resilient to unexpected situations
respond to real-time adaptive control systems as if human "experts" were deployed in devices (or systems or software applications)
provide a way to explicitly describe complex functional inter-relationships that can be understood by laymen (rather than mathematicians) such that the policies can be audited and updated if appropriate
allow autonomous systems to pursue goals on their own or by collaborating with other devices or humans
allow policies to be deployed on multiple platforms without re-engineering for demonstration, simulation, emulation, production and auditing
allow one to "watch" systems think (interpret information, balance alternatives, and allocate resources)
allow domain experts (policy makers) to incorporate real time information into models under development while they are being developed
explain "exactly" why a decision was made and how resources were allocated to support that decision
explain "exactly" why other decisions were not made
respond to dynamic, non-linear, inter-related, multi-dimensional problem sets with relative ease

Technology Comparison
	Neural Nets	Fuzzy Logic	Bayesian Belief Nets	AI - Forward / Reverse Chaining	KEEL
General Concept	Pattern Matching	Geometric Fuzzification / Defuzzification	Probabilities of Probability	Trial and Error	Adaptive Functional Relationships between Data Items
Source of Understanding	Patterns	Human Designer	Human Designer / Statistics	Human Designer	Human Designer
Pattern Training Required	Major Problem	No	No	No	No
Explainable Decisions	No	Difficult	Difficult	Somewhat	Fully Explainable
Small Memory Footprint	?	?	?	No	Best
Easily Extensible	Must start over with retraining	Somewhat	Statistics may have to be regenerated	Possibly	Yes
Performance	Determined by Application	Determined by Application	Determined by Application	Worst	Determined by Application
Suitable for Control	Yes	Yes	Probably Not	Probably Not	Yes
Interactive Development	No	No	Somewhat	Partial	Yes
Portable Design (device, software, web)	Probably Not	Probably Not	Probably Not	Probably Not	Yes (one design, many output formats)
Weaknesses	Pattern Training / Surprise Information	Validated Reasoning	Statistics may not be available for non-linear systems; Difficult to explain	Fragile / Brittle - Hard to maintain

More details in the Frequently Asked Questions (FAQ) area.

Compsim's technology has been embedded in a software application focused on organizations and businesses attempting to make subjective decisions (Compsim Management Tools). The same fundamental way of integrating driving and blocking information has been extended for embedding decision-making in devices that in many cases have to address much more complex inter-related problem sets (Knowledge Enhanced Electronic Logic). In this case, Compsim provides a KEEL Toolkit with the KEEL "dynamic graphical language" to assist in the design of thinking / reasoning devices.

All of Compsim's technology is focused on subjective decision-making, where the outcome is judgmental. In these cases human expertise is captured for use. These types of problems are all "data fusion" problems, where data items are integrated into packets that are interpreted to generate knowledge (or information). Information is then used to make decisions and decisions control actions. Judgment (or expertise) is demonstrated when information is interpreted to take controlled or relative actions when information needs to be balanced against competing goals. (risks vs rewards)

In organizational decision-making, the subjective rules of the organization are utilized to facilitate the decision-making activity. All information is exposed to insure that any personal biases can be removed from the process.

In embedded decision-making the expertise of the most knowledgeable human operators or decision-makers is incorporated into the products.