DocuMine Documentation

The rule engine is responsible for storing, processing, and evaluating data relevant for executing the defined business rules. It matches the facts stored in its working memory to the conditions of the rules stored in its production memory to decide which actions to take.

The following diagram illustrates the basic components of the Drools rule engine:

Rules engine components

The rule engine matches the facts in the working memory against the conditions of the rules stored in the production memory (a process known as pattern matching). When the conditions are met, the rules are registered and prioritized on the agenda for execution.

The Java objects and the facts in the rules engine are separate. This implies that when rule-related information in the rule engine is updated by either you or an automated process, truth maintenance is required to insert or update the facts in the working memory. For each fact, there is a fact handle, which associates the fact in the working memory with the Java object.

After the initial firing of all rules, the rule engine repeatedly cycles through two phases:

After executing the rules, the engine switches back to the “when” evaluation phase to re-evaluate the rules, as new rule data may have entered the working memory, resulting in additional rules becoming fully matched and eligible for execution. If no matching rules are identified, the cycle ends.

Rule execution flow

If several rules match, the agenda follows the order of the rules in the rule editor. In this case, the removal of one rule may lead to the removal of another rule from the agenda. (This applies at least to the current Drools version, where the order is unspecified but not randomized, although Drools does not guarantee it in general.) To avoid conflicts like the one mentioned above, you can prioritize rules by specifying execution flow parameters (rule salience and/or agenda group) where necessary. Please have a look at the Drools documentation if you are interested in further approaches.

When you create rules that modify Java objects, you will have to update the corresponding facts in the rules engine’s working memory to ensure synchronization between Java and the working memory. In DocuMine, you can accomplish this by invoking the EntityCreationService class.Invoking the EntityCreationService class: "then" part

rule "DOC.7.0: Performing Laboratory (Name)"
    when
       $section: Section(containsString("PERFORMING LABORATORY:"))
    then
        entityCreationService.lineAfterString("PERFORMING LABORATORY:", "laboratory_name", EntityType.ENTITY, $section).findFirst().ifPresent(entity -> {
            entity.apply("DOC.7.0", "Performing Laboratory found", "n-a");
        });
    end

Drools actually supports various keywords. However, we only recommend the following and only to advanced users, as these keywords can result in endless loops. For instance, using “insert” is not necessary in DocuMine as it is handled by the entity creation service.

If it turns out during rule firing that multiple rules apply to the same entity, the matched entities are applied in the following order:

For further information, please see Rule naming conventions.

The execution flow parameters (rule salience and agenda group) have no impact on the priority of the matched entities. These parameters affect the rule execution order, not the matched entities priority.

Example:

Rule order in the rule file:

The following matched entities application order applies: