A Research On Automatic Generation Of Rules And Its Application In Medicare Audit

Nowadays,more and more software systems use the rule-engine technology to implement business functionality,so that the business rules can be separated from the business logic and become configurable dynamic script.But the business rules of most software must be translated by the developer into the corresponding rule description language before they can be understood and executed by the rules engine.Therefore,the new rules must be understood by the developer first,then take effect after coded and test.And the business rules also lost confidentiality since developers are acquainted with them.So business users prefer to create and manage their business rules through easy-to-use UI components without the developers.Based on Drools rules engine,this thesis designs a rule generation framework named Ginkgo for business users in common domains as middle-ware for user interface and the rule engine.The Ginkgo framework consists of a dedicated rule description language(GRL)and its parser,as well as a set of rules serving APIs for improving Drools' ease of use.GRL is designed based on abstract syntax tree and JSON data exchange format.It supports the following six data types: text,number,boolean,date,time and timestamp,and contains constants,variables,functions,rules of action,rules and rules package as syntax elements.GRL is responsible for the expression and storage of business rules,but still need to be resolved to Drools native rules description language(DRL)at runtime in order to compiled and executed by the engine,this work is done by the GRL parser.The parser works with the help of JSON tool.First,extract all the rule elements according to the GRL syntax,then combine them into rule scripts according to the DRL syntax and supplement some necessary items.The rules service API encapsulates the functions of Drools to simplify the loading,compiling and execution of business rules.The Ginkgo framework is actually applied to a medicare audit project.It completes the implementations of all meta-rules and compound rules and the functionality and usability are verified by actual tests.A meta-rule is a rule that contains only one single logic in the condition,while the condition of a compound rule contains combinational logic.There are also some procedural rules in this project requirements,among which linear rule flows can works in the Ginkgo framework,but the ones with complexbranches can not be supported yet,which is the main direction of the follow-up study.At the same time,the support of the Ginkgo framework for Drools is still not comprehensive enough,which needs to be further improved.