Abstract

Automated rule checking aims to automate design checking via computer-processible rules and has been extensively studied for years. Existing methods for rule interpretation are often limited in their ability to represent implicit complex computational logic, and the integration between complex logic representation methods and rule interpretation outcomes remains insufficient. In this paper, we propose a graph-based method for representing knowledge rules for automated design checking. The graph representing procedure consists of two main steps. Firstly, semantic mapping is employed, whereby the rules are represented as a semantic combination of three corresponding graph branches. Secondly, atomic function mapping is utilized, which adds atomic function nodes to the graph from the level of computable logic to form a complete graph representation, thereby forming a complete graph representation. We then provide a method for automatically generating programming code from the graphs, which, combined with the development of atomic functions, can be used for checking. This approach demonstrates considerable potential in terms of its interpretability and comprehensibility, and it provides novel ideas for the development of fully automated rule checking systems.

Download paper here

Download preprint here

The authors are grateful for the financial support received from the National Key Research and Development Program of China (No. 2023YFC3804600) and National Natural Science Foundation of China (No. 52378306) .

Financial Sources: