Key Business Rules-based Expert System Technology Research

Improving the business agility of the software systems is a very hot research area in the academia. In the traditional software applications, the business rule logic which could be very complicated and keeps changing frequently is normally implemented in program code and embedded in the application logic. This makes the rule logic extremely hard to be maintained and managed. Business rule-based expert systems (BRES) can separate business rule logic from the application logic and manage the rule logic in the similar way as managing data. BRES is an effective solution to achieve business agility.This dissertation investigated and analyzed the key techniques of the BRESs systematically and in depth, including the way to categorize rules, the way to structure rules, the way to represent rules, the way to store and organize rules, the way to execute rules and the process models can be applied to develop BRESs. The research is significance and has immense application value. The main innovations are stated as follows:1. The dissertation presented a new method to classify and formalize rules. A fresh new way to classify rules from both business perspective and software systems perspective and the corresponding rule classifications are presented respectively. A set of system-oriented rule templates is given. It solved the issue of the formalized representation of rules and established the base for rules creation and edition in BRESs.2. A rule repository model based on the system-oriented rules and the essential functionality to manage the repository are given. The XML is proposed to be used to store rulesets. It solved the issues of rules organization, rules management and rules storage.3. Based on the analysis and the research on the pattern-matching problem and the techniques improving pattern-matching efficiency of the system, the runtime environment including the rule engine and application API are presented and the corresponding rule pattern-matching algorithm is given. It provided the environment to execute business rules in BRESs.4. A suitable process model used to develop BRESs is described. It solved the problem not splitting the business rules in each development stage in traditional software development model and provided a kind of methodology for BRESs development.5. A J2EE-based BRES platform on the basis of the research in this dissertation is designed and built. A business rule-based insurance underwriting expert system prototype is implemented. The result shows that a BRES can be implemented rapidly based on the method in this dissertation and the rules can be updated and maintained easily to adapt the frequent changes of business rules. The agility of the rule logic is improved tremendously.