Rule Declaration Language

The rule declaration language supports two types of rules: Decision rule or a Score rule.

Score rule:

• A Score rule is composed of one or many rule sets.

• Each rule-set computes a sub-score and is applied a weight.

• The total score then would be the sum of all the individual scores of all the rule sets belonging to a rule.

• A rule set is composed of one or many rule rows.

• You can ‘roughly’ think of each Rule Row as a Conditional evaluation of the facts (a.k.a antecedent) & a score based on these conditions (a.k.a consequent).

Decision rule:

• A Decision rule is always composed of only one rule set.

• A rule set is composed of one or many rule rows.

• You can ‘roughly’ think of each Rule Row as a Conditional evaluation of the facts (a.k.a antecedent) & a decision based on these conditions (a.k.a consequent).

• A decision rule always arrives at a single decision at the end of parsing.

• The decision can be anything (a numeric, a string such as YES/NO or even a JSON)

• Once a Rule Row evaluates to True, the corresponding decision is returned immediately.

Antecedent and Consequent

• An antecedent at the core is an evaluator. It evaluates one or many facts through an operator.

• For evaluating numeric facts, a numeric operator is used. It can be one of (<=, <, >, >=, ==, <>, between, is_none)

• For evaluating string facts, a string operator is used. It can be one of (in_list, contains, is_none, equals)

• You can mix evaluation of more than one fact & combine the result with an “and” or “or” condition.

• You can perform complex evaluations involving multiple facts combining AND and OR conditions recursively in the antecedent. See Examples.
• A rule can be part of another rule. See Examples.