The Research And Implementation Of Fact Database Persistence In Prolog

It is important for an information process system to be capable of dealing with data in vast quantity. Data persistence is one of the essential ways to approach this problem.Prolog is one of the most important computer languages used in the field of Artificial Intelligence. Besides its important role in AI, Prolog is also being more and more widely used in many other information process systems, as a result of the fact that ever more intelligent processing systems are required in our society.In recent years, much development has been made in data persistence in Prolog. However, most persistence products, including Prolog's, are based on third party relation database management system today. It may increase the cost of applying and maintaining it as well as its complexity, so, decrease the efficiency of the system, etc.By means of the theory and method of relation database, especially the technology of B~+ tree, a built-in persistence subsystem, called FTDB(abbreviation for "Fact Database"), in Prolog is designed and implemented. It is independent of any third party software products, through which Prolog fact database persistence system has been realized. Therefore the internal database is expanded from memory to secondary storage. And it would reduce the complexity and cost of usage, promote the access efficiency, compared with other persistence solutions.On the basis of the technology of simplified prefix B~+tree, more nodes at top levels of a B~+ tree in this system could reside in memory.The concept of variable-order B~+ tree was brought forth. It could avoid either waste or overflow of a block space for the different length of index keywords.In addition, a new concept of correlative synchronization has been defined and as a result, a control mechanism has been established in order to cope with synchronization. This special synchronization involves several resources shared by multiple threads that while some of those resources are being accessed, any access to rest of the resources must be prohibited. So those resources are called correlative.The traditional method of binary search has been reformed to satisfy the special need for inserting and deleting a fact.The physical organization and management of the fact database on secondary storage has been completely implemented as well as other parts of the system, like buffer management, log management, transaction management and synchronization control.The low level functions of the persistence sub-system and some high level functions have been realized, such as inserting , deleting and querying fact.This persistence system is still under research and development. Although basic functions have been realized, many functions remain to be realized and there are some problems to be resolved.