规则自动生成技术研究及其在医保审核中的应用

发布时间：2018-03-03 13:43

本文选题：规则引擎　切入点：规则描述语言　出处：《电子科技大学》2017年硕士论文　论文类型：学位论文

【摘要】：目前,越来越多的软件系统使用规则引擎技术来实现业务功能,将业务规则从业务逻辑中分离出来,作为软件系统的动态可配置脚本。但是多数软件的业务规则必须由开发人员翻译为相应规则描述语言之后,才能被规则引擎理解和执行,因此新规则仍要通过开发人员学习理解,并编码、测试通过之后才能生效,同时业务规则也失去了保密性,于是业务用户更加希望实现规则的自动生成——即无需开发人员直接参与,通过简单易用的人机界面独立完成业务规则的生成和维护。本文基于Drools规则引擎设计了一种服务于业务人员的Ginkgo规则生成框架,作为业务系统中规则生成界面与规则引擎的中间件。Ginkgo框架由一种专用的规则描述语言GRL和解释器,以及一套用于提高Drools易用性的规则服务API组成。GRL基于抽象语法树和JSON数据交换格式设计实现,支持文本、数值、布尔、日期、时间和时间戳等6种数据格式,定义了常量、变量、函数、规则动作、规则和规则包等语法元素。GRL负责业务规则的表达和存储,但在运行时仍需解析为Drools原生规则描述语言DRL才能由引擎编译和执行,这一解析工作由GRL解释器完成。解释器基于JSON工具软件实现,首先按照GRL语法提取出全部规则元素,再根据DRL语法将它们组合为规则脚本,并补充必要的DRL语言成份。规则服务API对Drools的相关功能进行了封装,以简化业务规则的加载、编译和执行等操作。Ginkgo框架被实际应用于某医保审核项目,完成了全部元规则和复合规则的实现,经过实际测试,其功能逻辑和可用性得到了验证。其中,元规则是指条件中只含有单一逻辑的规则,复合规则的条件则由任意嵌套的组合逻辑构成。审核业务中还存在流程性的规则,Ginkgo框架能够支持线性的规则流,但尚无法实现包含复杂分支的规则流,这是后续研究待突破的主要方向。同时Ginkgo框架对Drools功能的支持仍不够全面,也有待进一步提高。
[Abstract]:At present, more and more software systems use rule engine technology to achieve business functions, separating business rules from business logic. As dynamic configurable scripts for software systems, however, most business rules of software must be translated by developers into the corresponding rule description language before they can be understood and executed by the rules engine. So the new rules still have to be learned by developers, coded, tested before they come into effect, and business rules lose confidentiality. So business users want to automate the generation of rules-that is, without the direct involvement of developers, The generation and maintenance of business rules are accomplished independently by a simple and easy-to-use man-machine interface. Based on the Drools rule engine, a framework of Ginkgo rule generation for business people is designed in this paper. As a rule generation interface and rule engine in business system, the middleware. Ginkgo framework consists of a special rule description language (GRL) and an interpreter. And a set of rule service API components used to improve the usability of Drools. GRL based on abstract syntax tree and JSON data exchange format design implementation, support text, numeric value, Boolean, date, time and time stamp, 6 data formats, defined constant, The syntax elements, such as variables, functions, rule actions, rules and rules packages, are responsible for the representation and storage of business rules, but at runtime they still need to be parsed into the Drools native rule description language DRL to be compiled and executed by the engine. This parsing is done by the GRL interpreter, which is based on the JSON tool software. Firstly, all the rule elements are extracted according to the GRL syntax, and then combined into regular scripts according to the DRL syntax. The rule service API encapsulates the related functions of Drools to simplify the operation of loading, compiling and executing business rules. The framework of Ginkgo is applied to a medical insurance audit project. The implementation of all meta-rules and composite rules has been completed, and the functional logic and usability of the meta-rules have been verified by practical tests. Among them, meta-rules refer to rules that contain only a single logic in a condition. The condition of composite rules is composed of arbitrary nested combinational logic. The Ginkgo framework can support linear rule flow, but it can not implement rule flow with complex branches. At the same time, the support of Ginkgo framework for Drools function is still not comprehensive enough, and need to be further improved.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F842.684;R197.1;TP311.52

【相似文献】