Representing Knowledge

Scientio’s products combine data mining and expert systems. This means that we can also supply an innovative, high speed and effective rules based system for a variety of applications. In many cases, while the ability to automatically generate sets of rules might be useful, potential users are primarily interested in how they can get human knowledge such as laws, codes of practice, the company rule books etc. into a machine usable form. Scientio’s systems use Fuzzy Logic which is a superset of conventional Boolean logic. This allows users, amongst other things, to reason about numeric values as well as categorical ones. Scientio's expert system and data mining tool XmlMiner also supports Fuzzy Arithmetic, which means that interval type calculations are performed automatically. Real world data values often have tolerances and certainty values associated with them. You can supply these directly to XMLMiner which will calculate with them in a single pass. Studies have shown that this fuzzy arithmetic based approach is very much more efficient and useful than comparable Monte Carlo methods.

• Simulation - rule based engines can be powerful simulators, and the addition of fuzzy arithmetic mean that there is a rich range of data types to use. Examples of simulation are risk analysis, engineering system analysis, war gaming and many others.
• Building of business rules into ecommerce systems and intranets. XmlMiner can be integrated into SharePoint and other web sites to perform rule based processing. A great example of this is our web service billing system that assigns each customer an individual rule set which defines how transactions are priced.
• Encapsulation of law, best practice, rule books etc. into 'advice engines'. Our new Questionnaire interface makes it simple for a user to navigate through complex sets of rules and obtain advice stored in our rule sets.

In addition, a working system needs a few extra facilities to be really useful:

Interactive editing of rule sets.

Our Metarule guided editor makes it simple for untrained users to create and edit rule sets. The ability to treat rule sets as building blocks and create structures of rule sets feeding into each other. XmlMiner support processing maps which can combine multiple rule sets to make a single functional block.

Discovery of incomplete rule sets.

When humans write rule sets they often leave out key cases. Our product Lacuna automatically identifies these using Genetic search techniques.