GRAPH REDUCTION MODELS OF LOGIC PROGRAMMING CONTROL

不确定性程序控制的图约简模型初步实验表明，这是一种可行的、自然的方法。和和或组合子的适当约简规则与新类型的标记节点（alt节点）一起，使声明性语言语句能够直接编译成图形部分，通过简单的重写规则进行的约简将影响回溯。逻辑语言的其他图约简方法将这种负担放在元语言上。操作图约简模型与声明性语言B的指称语义相匹配，为扩展到完整的逻辑编程和将来合并更高级的控制结构提供了一个优雅的框架。尽管当前的模型通过显式的状态参数抽象了状态的复杂性，但在扩展中可以采用适当的替代策略。其他研究人员目前正在制作包含逻辑变量适当行为的图约简模型，这将在我们的扩展中考虑。主要目标是开发当前模型，以便正确有效地为声明性语言的复杂控制建模。对如何将Prolog的切割叠加在一起的研究已经在进行中。该模型将寻求将必要的控制信息合并到图中的方法。简单alt节点在非确定性建模中的成功是一个令人鼓舞的开始。

Initial experiments with a graph reduction model for nondeterministic program control indicate that this is a viable and natural approach. Appropriate reduction rules for and and or combinators together with a new type of tagged node (alt nodes) enable the straightforward compilation of declarative language statements to graph sections whose reduction by simple rewrite rules will effect backtracking. Other graph reduction approaches to logic languages place this burden on the metalanguage. The matching of an operational graph reduction model with a denotational semantics for the declarative language B provides an elegant framework for extension to full logic programming and the future incorporation of more advanced control structures. Although the current model abstracts the complexities of state by making explicit a state argument, there are appropriate alternative strategies that can be employed in the extension. Other investigators are currently producing graph reduction models incorporating appropriate behavior for logic variables, which will be considered in our extension. The main objective is to develop the current model to correctly and efficiently model sophisticated control for declarative languages. Investigation of methods to superimpose Prolog's cut is already underway. The model will seek ways to incorporate the necessary control information into the graph. The success of simple alt nodes in modeling nondeterminism is an encouraging start.