Fine-grained deep mining of factors influencing carbon emissions in China based on graph adversarial learning

Extensive mining and deep understanding of carbon emission influencing factors are of great significance for the global carbon emission reduction process and even the realization of global sustainable development. In this paper, we propose a fine-grained mining framework for carbon emission influencing factors based on graph adversarial learning, which is based on the self-attention mechanism for information extraction to construct a heterogeneous graph of carbon emission influencing factors, and complete the graph through generative adversarial learning for deep mining of implicit or indirect information to realize the fine-grained characterization of carbon emission influencing factors. Experimental results show that the method proposed can effectively extract relevant information and deeply mine the fine-grained influence factors. The framework explores the significance and logical relationship of factors from a relatively micro industrial perspective, and further present the knowledge structure and network relationships of carbon emission influencing factors through the knowledge graph, which can visually and comprehensively present complex and effective information in a large number of texts. Studies case in this paper show that the framework can be distinguished from the traditional bibliometric and statistical perspectives, deeply exploring the key carbon emission influencing factors, and comprehensively analyzing these factors from different regions and industries.