Construction Of Knowledge Graph For Rare Species In China

Since the concept of knowledge graph was proposed by Google in 2012,great progress has been made in the research of knowledge map in various fields.In education,science,finance,criminal investigation and other fields,professional knowledge has been constructed to serve various industries,and has achieved certain results.However,as the focus of the national government,the construction and research of knowledge graph in ecological is rare.With the rapid increase of population and the increasing demand for natural resources in economic development,deforestation,industrialization and urbanization will destroy the living environment of wildlife,and lead to the loss of biodiversity.Among the 640 world endangered species listed in the Convention on international trade in endangered wildlife species,156 species are in China,That’s about 24%of the total.Due to the interrelationship and restriction between species,if one species is extinct,there will be 10 to 30 other species attached to the plant disappear.Therefore,it is very beneficial to construct a knowledge graph of the current situation of rare species in ecological field.This paper mainly aims at the current situation of rare species in China to construct knowledge graph.The whole article studies from the following four aspects:first,data crawling and processing.The data in this paper are from Baidu Encyclopedia and other search websites and various official websites related to rare species and species protection.Thus obtained the text information related to the construction of knowledge map,including the species,subjects,habitats and survival threats of rare species in China,and de-noised the data;second,named entity identification.The main methods used are conditional random field(CRF)and bi-directional long-term memory network(BiLSTM-CRF)model.And the unstructured text data is combined with knowledge by Word2vec model and semantic model.The entity information such as species name,distribution area,subject and endangered level are identified;thirdly,the relationship extraction based on the classification method.Combined with convolutional neural network model,six relationships between entities are extracted considering the position and distance information of entities.Fourth,data storage and visualization.Choose the Neo4j graph database to store data,and tens of thousands of entities and relationships are displayed in the graph,obtain the relationship between the rare species in China and the distribution area,the endangered level and the subject.Thus,the knowledge map of rare species in China is constructed completely.It provides more perfect species information for relevant departments,staff of ecological protection and rare species lovers,helps government departments provide more reasonable and effective species protection measures,and promotes the sustainable development of ecological environment.