Expert Systems

The basic idea is that if a human expert can specify the steps of reasoning, by which a problem may be solved, so too can an expert system.

Expert systems are able to achieve a dramatically improved level of performance for tasks that require a highly specialized level of knowledge and training. Therefore, the commercial value of an expert system lies in new revenue building, rather than replacing people.

One of the most powerful attributes of an expert system is the ability to understand how it reached a decision including the factors considered when making a recommendation. Expert Systems:

  • Use a heuristic search [implicit steps] rather than an algorithmic search [explicit steps] – this speeds up the process of finding a “good enough solution” when an exhaustive search is impractical
    • Satisfactory answers are usually acceptable,
    • Some incorrect answers are tolerable
  • Perform knowledge and decision reasoning tasks vs. performing programmed step-by-step procedures
  • Employ a Knowledge-base vs. a Simple database
    • Uses stored knowledge information for conclusions vs. only providing discrete facts about a subject
  • Separate the control structure from the domain knowledge (information) so when format/structures change, processing continues without complications or interruptions
  • Encode expertise in data structures [inference rules] vs. encoding expertise in both the program and data structures so rules can be modified without rebuilding the program
  • Simplify the modifications and expand the rule sets
    • Rules are easier for (non-programmer) experts to create and modify vs. writing program code

Expert Systems are typically very complex applications used by the medical and scientific research industries. What Cognasys has succeeded in doing is developing a unique, “simplified” expert system that focuses on the processing of business transactions.