Construction and analysis of knowledge graphs for multi-source heterogeneous data of soil pollution

The decommissioning and overall relocation of urban industrial enterprises have generated a large number of decommissioned contaminated sites, and the leftover soil pollution is gradually becoming a major problem that restricts urban green development and damages human health. Deep mining and efficient management of site soil pollution information through digitization and informatization are needed to solve these problems more accurately and efficiently. Knowledge mapping for visual analysis of relevant pathways is a forward-looking approach in soil contamination management that does not require complex testing instruments, thus saving research manpower, time and cost. Data associated with contaminated sites come from a wide range of sources and have different structures. Through the natural language processing technology of computer, suitable methods such as entity recognition, relationship recognition and knowledge fusion are selected to extract various types of information from contaminated sites and establish semantic networks for fast targeting of soil contamination sources, thus providing a more convenient solution. In this paper, we propose a knowledge graph construction method for multi-source heterogeneous data of contaminated sites, find sulphide contamination sources through visual analysis of knowledge graph and explore the application prospects of natural language processing techniques such as knowledge graph in contaminated site management.