Integration of Logic and Functional Languages Based on Dataflow

逻辑和函数式语言的技术优点已经得到了广泛的认可，这些语言的集成问题已经出现，需要从形式语义和实际实现技术的角度进行讨论（例如，参见De Groot和Lindstrom， 1986）。已经开发了几种语言，重点是将等式与逻辑语言相结合。这些语言的主要计算机制是基于解析的，但术语重写规则、窄化等应该与解析统一起来。如何统一它们甚至在逻辑编程中也引起了一些有趣的问题。本研究项目一直遵循这种集成的方向，但避免了统一不同计算的方式。考虑到FP程序的可表达性和逻辑程序解析演绎中涉及的代数操作，我们得到了一种具有FP函数等式的逻辑语言fplog，它的语义是用数据来表示的；与LEAF定义语义的方式不同（giovvannetti et al., 1991）。本报告中提出的语义反映并建议了逻辑语言和由FP程序组成的方程的数据流计算。为了实现FPLOG的语义化，我们首先使用数据流构造来处理逻辑程序的语义。语义域是基于Herbrand基或由幂等替换的等价类组成的商集的序列域。作为相关主题，我们关注不确定性数据流和具有通信通道的逻辑程序。有一个制定的网络通信逻辑程序。网络可以表达序列逻辑程序的数据流计算。然后给出了从逻辑程序到功能程序的转换技术，特别是FP程序的转换技术。转换过程不像Debray和Warren（1989）那样采用谓词中涉及的输入-输出映射，而是将每个确定子句将替代集或序列转换为替代集或序列的解释，这种解释是由推理引起的，因此替代集或序列是原子集的虚拟表示手段。翻译可以用一个FP程序来表示，它也可以看作是从逻辑程序中提取出来的函数。最后是FPLOG，它由包含FP函数项的等式的定子句集组成。由于相等性可以通过数据流计算来实现和评估，因此建议对FPLOG进行数据流计算，以便语义可以反映它。少

The technical merits of logic and functional languages have been widely acknowledged, and the integration problems of these languages have arisen to be discussed from formal semantics points as well as practical implementation techniques(See, for example, De Groot and Lindstrom, 1986). Several languages have been developed in focus on integration of equality with a logic language. The primary computing mechanism for such languages is resolution-based, but term rewriting rules, narrowing and so on should be unified with resolutions. How to unify them has caused interesting aspects even in logic programming. This research project has been following the direction of such integration, but avoids the way of unifying different computings. Taking the expressiveness of FP programs and algebraic manipulationinvolved in resolution deductions of logic programs into account, we have a logic language with FP functions equalityFPLOG, for which semantics is formulated by means of data ; flow, differe … More nt from the way of defining semantics for LEAF(Giovannetti et al., 1991). The proposed semantics in this report reflects and suggests dataflow computing for both logic languages and equations consisting of FP programs.For the goal to semantics for FPLOG, we firstly deal with semantics for logic program using dataflow construction. The semantic domain is a sequence domain based on the Herbrand base or the quotient set consisting of equivalence classes of idempotent substitutions. As related topics, we are concerned with nondeterministic dataflow and logic programs with communication channels. There is a formulation of the network of communicationg logic programs. The network might express the dataflow computing for a sequential logic program.Then the transformation technique from a logic program into a functional program, especially an FP program, is provided. The transforamtion process does not take the input-output mapping involved in predicates as Debray and Warren(1989), but the interpretation that each definite clause translates substitution sets or sequences to a substitution set or sequence, caused by inferences such that the substitution set or sequence is means of virtual representation of atom sets. The translation might be expressed by an FP program, which is also regarded as an extracted functional from a logic program.Finally we have FPLOG, which consists of definite clause sets involving equalities for terms containing FP functions. Becauses the equality might be realized and evaluated by dataflow computing, dataflow computing for FPLOG is suggested such that semantics might reflect it. Less