【作者】 周扬；

【导师】 王振国；

【作者基本信息】 山东中医药大学 ， 中医医史文献， 2010， 博士

【摘要】 目的：切实立足中医药传统理论,从医药一体的角度全面表征药性,实现中药药性内涵从自然语言描述向现代科学语言描述的转换。本文利用自然语言处理(NLP)的理论和技术方法,将本体(ontology)引入到中药理论研究领域中,基于中药历史文献数据构建中药本体。将与中药药性相关的概念术语重新进行知识架构和关联映射,利用公认的中药领域知识及现代研究成果,对中药核心概念进行全面诠释,自动构建基于中药药性表征体系的中药本体。以期通过概念及概念间的关系全面描述中药药性的本质,为中药药性理论以及其他相关学科的研究提供一组公共的、可共享的中药概念体系,帮助领域专家分析中药药性知识,实现中药信息的知识化重组,为中药领域的数据挖掘和知识发现提供知识层面的本体库。方法：充分利用已有的公认的中医药领域知识,通过重构概念关系、重构知识描述体系、重构知识描述方式建立起多维中药本体。首先采取自顶向下法,从顶层概念着手,定义大部分显而易见的概念,而后再通过添加子类,分别将这些概念进行恰当地归纳和细化；同时,采用自底向上法,以历代重要本草著作为信息源,由最底层、最细小的概念的定义开始,从古代本草文献中抽取概念,然后将这些细化的概念组织在更加综合的概念之下,并逐一确立概念属性,以此建立本体知识描述体系的概念模型原型。并利用多维中药本体构建平台,形式化表示这一概念模型。结果：提出构建中药药性表征体系的思路与方法,构建了药性概念的立体化表征模型,实现了药性概念从自然语言描述到现代科学语言描述的转变,实现药性概念的准确性、系统性描述,更全面地揭示药性内涵,同时修订了部分原有公认的中医药领域知识结构。结论：面向中药药性理论的中药本体,以传统概念体系为初始框架,源于大量原始文献语料,经过三级知识重组,重新构建而成。在描述方式、揭示深度、共享等方面均与传统概念不同。基于实际文献数据构建的多维网状的概念关系体系,具有可扩展性,并同时具有自然语言和主题关联词两种描述方式等特点。更多还原

【Abstract】 Object:We lead the ontology into the field of Traditional Chinese Medicine theory research with Natural Language Processing theory and technique. We build the Traditional Chinese Medicine ontology Based on Traditional Chinese medicine historical document data.We carry on the knowledge construction and shine upon the connection to concept terminology related with 973 plan projects "The Chinese herbal property theory related foundation question research". We carry on the comprehensive annotation to the Traditional Chinese Medicine core concept and construct the ontology-based representation of Chinese herbal property theory automatically. We hope to descript the Traditional Chinese Medicine property comprehensively by the time through the concept and the concept essence, provide the Traditional Chinese Medicine concepts system which for the Traditional Chinese Medicine property theory as well as other related discipline’s research a group public, help the domain expert to analyze the Traditional Chinese Medicine property knowledge, realize the Traditional Chinese Medicine information knowledge-based reorganization, provide the ontology storehouse of the knowledge stratification plane for the Traditional Chinese Medicine domain’s data mining and the knowledge discovery.Methods:We make full use of existing recognized knowledge in the field of medicine, the concept of Chinese colleges and universities nationwide unified teaching materials seventh edition, and the terms from Tang materia medica, Compendium of Materia Medica, Shen Nong’s Herbal Classic variorum, Chinese Materia Medica, etc., as the concept of ontology in the alternative, establish a multi-dimensional reconstruction of knowledge and describe the methods of Chinese Medicine ontology through the reconstruction of the concept of relations, reconstruct knowledge description systems.First of all to take a top-down method, starting from the top-level concepts, we definition most obvious concept, and then appropriately summarized and refined by adding sub-categories, appropriately summarized and refined those concepts. At the same time, using the bottom-up method, to an important herbal medicine in ancient writings as a source of information, from the bottom, the smallest of the definition of the concept, taking literature from the ancient concept of Materia Medica, we refine the concept of these organizations in a more comprehensive concept, and one by one establish the concept of property, in order to establish a prototype system of ontology knowledge description.Results:The paper presents the thoughts and methods of representting Chinese herbal property, and constructs the representation model of it. We have acquired multi-dimensional mesh system of Traditional Chinese Medicine herbs, and partially revised the original structure of domain knowledge recognized.Conclusions:Ontology-based system of theoretical concepts of Chinese Medicine herbs, the traditional concept of system as the initial framework,derived from a large number of the original literature corpus, after a three knowledge restructuring, has been rebuited. In describing the way, revealing the depth, sharing, etc.It is all different from the traditional concept. It has the characteristics of the literature data based, scalable, being network concept of relationship, being natural language and being the methods of key words description and so on.更多还原