建筑工程质量管理规范建模与检索的本体化方法研究

The relatively well-established system of construction quality management regulations has been developed in China. However, the regulation documents are mostly paper-based, and not computer-processable. The provisions of regulations are organized hierarchically and statically. The relevant constraint provisions about the same construction component may be distributed in the different regulation documents, which even conflicts each other. Therefore, it is difficult for users to find the applicable provisions or use these regulation knowledge in the intelligent application systems. In the background mentioned above, how to make regulations be computer-processable, organized dynamically according to different classification systems and retrieved semantically, becomes the starting point of this research project. In this research project, the regulation document is regarded as the semantic network of all kinds of constraints, in which the net nodes are connected according the semantic relations between the constraint objects or the classification systems by which the provisions are organized. This research project focus on the study of the ontology-based semantic modeling and retrieval of the construction quality management regulations. A semantic framework for the regulation knowledge modeling will be proposed. In order to automate the process of regulation knowledge extraction, this research will treat the regulation knowledge extraction as the process of concept instance and the relation instance population of reguation knowledge reference ontology, and try to develop a method of ontology-driven regulation knowledge extraction, basing on the nature language processing technique and semantic analysis technique. Moreover, an algorithm for evaluating the semantic similarity between two provisions is designed to implement the regulation semantic retrieval. Finally, the approaches and algorithm proposed above are applied in the typical construction quality management regulation document, and the regulation knowledge are extracted and formalized into the semantic knowledge base. The research outcome of this project will improve the effect of the regulation retrieval, and enhance the computer-processable level of regulation knowledge. Also, the research outcome will lay the solid theory foundation for the semantic applications of regulations.

我国已形成相对完善的建筑工程质量管理规范体系。但规范条文的组织编排以静态的单一分类结构进行。对同一对象的相关约束分布在不同规范条文中，甚至相互冲突，查找应用不便。本项目以上述问题为背景，以规范条文的计算机可处理、多维动态组织与语义关联检索需求为出发点，将规范视为一个由规范约束项依据其约束对象在领域中的语义关系以及约束条文编排/管理分类关系所构成的语义网，基于本体论方法，建立规范的语义描述框架，构建能够表达规范完整知识的层次语义模型；将规范约束知识的抽取视为本体概念与关系的实例化过程，研究本体驱动的规范知识自动抽取方法；开发规范条文语义关联度度量模型，实现语义检索。最后，依据所提框架、方法和模型，以建筑工程质量验收规范为对象，开发规范知识库原型系统，实现规范的多维动态组织和语义检索。研究为规范条文的电子化和语义应用提供理论、方法和技术支持。

我国已形成相对完善的建筑工程质量管理规范体系。然而，规范的组织编排以标准体系分类结构进行。规范以纸质或电子文本形式存在，计算机可处理程度低。对同一对象的相关约束规定分布在不同规范条文中，条文间相互参照、彼此嵌套，有时条文甚至相互冲突。这些都给规范知识的检索和智能应用带来了困难。.基于此，项目组在过去三年里，以建筑工程质量管理规范的计算机可处理、多维动态组织与语义关联检索等需求为出发点，采用理论研究、方法实现与原型系统开发验证相结合的方法开展研究。.主要研究内容包括：.（1）研究基于本体的建筑工程质量管理规范知识模型，增强知识模型的语义表达力，提升规范知识的计算机处理和语义理解程度。.（2）在规范知识语义模型基础上研究开发本体驱动的建筑工程质量管理规范知识(半)自动抽取方法。.（3）研究规范条文的语义关联度，开发规范条文语义检索原型系统。.根据上述研究内容，截止2016年12月31号，共发表期刊论文6篇(SCI论文5篇，EI论文1篇)。.主要成果包括：.（1）建立建筑工程规范的语义描述框架，构建能够表达规范知识的语义模型，提升规范知识的计算机语义理解和推理程度；.（2）将规范约束知识的抽取视为本体概念与关系的实例化过程，开发本体驱动的规范约束知识自动抽取和语义标注方法，提高规范知识建模效率；.（3）设计规范条文语义关联度度量算法，实现规范条文的语义关联检索。.（4）最后，依据所提框架和方法，以建筑工程质量验收规范为对象，开发规范知识库原型系统，实现规范条文的多维动态组织和语义检索等功能。.研究为建筑工程规范条文的电子化和智能应用提供理论、方法和技术支持。