Abstract: Normally there are complex semantic relationships between two objects which in turn are inverse in the real world entities. The current databases, no matter Oracle which extends object-relation model, or ODMG (object data management group) which first proposes the inverse relationships, are both insufficient in description and manipulation of the complex semantic relationships and its inverse relationships. Given this fact, this paper proposes a new method which can succinctly represent complex semantic relationships. Firstly, this paper describes the schema and instance language and the advantages of this new method in representing and manipulating the complex semantic relationships. Then, this paper analyzes the mechanism of the new method that automatically deals with the inverse relationships and generates target classes. Thirdly, this paper introduces how this method keeps database consistent. Finally, this paper compares the performance of new DBMS (data base management system) based on the new method and Oracle, and proposes several key issues in the new DBMS.

Key words: complex semantic relationship, inverse relationship, consistency

摘要： 现实世界的实体间往往存在着复杂的语义关系，而且实体间的关系往往是相互的。现有数据库无论是扩展了对象关系模型的Oracle，还是首次提出逆向关系定义的ODMG（object data management group），它们对现实世界中实体之间复杂语义关系及其逆向关系的描述和处理依然存在着很大的不足。为了更好地表述和处理现实世界中实体间复杂的语义关系，提出了一种能够简易描述实体间复杂语义的新方法。介绍了该方法如何针对复杂语义关系建立模型和插入对象，指出了该方法相对于传统模型的优越性，设计实现了实体间逆向关系和目标对象类的自动创建和生成机制，分析了数据库一致性维护问题。比较了基于新方法实现的DBMS（data base management system）和Oracle的性能，指出了基于新方法的DBMS存在的问题。

关键词: 复杂语义关系, 逆向关系, 一致性