土壤动物知识图谱构建理论、方法与技术——以浙江天目山土壤螨类为例

土壤动物学面临以全新知识体系为科学研究框架的变革时期,其核心内容是以数据驱动为主要特征的人工智能技术方法。目前广泛应用的基于数据库的数据处理分析方法,面临着数据多源异构、快速增长和处理能力不足之间的矛盾。基于快速发展的大数据科学和人工智能技术的数据挖掘方法在解决前述矛盾中有突出优势,但需要依赖一个强大的领域知识库,然而土壤动物领域知识图谱的研究十分匮乏。土壤动物知识图谱是一个具有有向图结构的知识库,其中图的节点代表与土壤动物相关的实体或概念,图的边代表实体或概念之间的各种语义关系。提出了土壤动物知识图谱的定义、内涵、理论模型和构建方法,以浙江天目山土壤螨类多样性为例,分析了构建山地土壤动物知识图谱的技术方法;以土壤动物多样性研究关注的物种分布、物种共存、环境条件对物种的影响作用为例,探讨了基于山地土壤动物知识图谱可以解决的相关科学问题。研究表明,土壤动物知识图谱在解决生物多样性重要科学问题方面具有独特的潜力和优势,有力推动了土壤动物学、信息科学和数据科学交叉的土壤动物信息学的发展。

Theory, method and technique of soil animal knowledge graph construction: a case study of soil mites in Tianmu Mountain, Zhejiang Province

Soil zoology is undergoing a period of development with a knowledge system at its core and data-driven. The current widely used data processing and analysis methods based on databases are facing the contradiction between the rapid growth of data and the lack of data processing capacity. Data mining methods based on rapidly developing big data science and artificial intelligence techniques have the outstanding advantages in solving the aforementioned contradictions, but they need to rely on a strong domain knowledge base. Yet there is a paucity of research on knowledge graph in the soil animal domain. The soil animal knowledge graph is a knowledge base with the directed graph structure, where the nodes of the graph represent entities or concepts related to soil animals, and the edges of the graph represent various semantic relationships between the entities or concepts. This paper introduced the concept, connotation and construction method of the soil animal knowledge graph. The soil animal knowledge graph is composed of triples (entity, relation, entity) to describe the relationships among entities. Take the soil mite diversity across an altitudinal gradient in Tianmu Mountain, Zhejiang Province as an example, this paper provided the method and process of constructing a mountain soil animal knowledge graph. In the process of constructing a soil animal knowledge graph, the data source was identified firstly, and the ontology construction objective and process of the soil animal knowledge graph were designed. The process of ontology construction included designing the soil animal knowledge graph model, dividing the core classes and class hierarchies, defining ontology attributes, and evaluating the ontologies. Finally, a preliminary graph of the mountain soil animal knowledge graph was displayed with Neo4j platform. To show the process of a data mining for a soil animal knowledge graph, this paper answers three important scientific questions based on the constructed mountain soil animal knowledge graph:that is where are the soil animals distributed, what species live together, and how environmental conditions affect the soil animal distribution. Finally, this paper pointed out that the soil animal knowledge graph has friendly portability and excellent scalability and indicated that some important scientific issues would be solved quantitatively for soil zoology in the future. In total, this study shows that the soil animal knowledge graph has unique potential and advantages in addressing importantly scientific questions about biodiversity, and has promoted the development of soil animal informatics at the intersection of soil zoology, information science and data science.