Causal Discovery and Knowledge Linkage in Scientific Literature: A Case Study in Biomedicine

Scientific literature is the main carrier to express innovation thinking, and the discovery of laws of knowledge from the literature is the necessary basis for scientific research to achieve innovation development, but current knowledge mining methods still have deficiencies in logic and reasoning. And causality is a higher-order cognitive relationship with logical reasoning ability, so it is necessary how to mine causality from the literature and establish knowledge linkage based on causality. [Methods] This paper proposes to find causality from the scientific literature and make a knowledge linkage based on the causal events and take full-text data in the biomedical field as an example. Firstly, we design a causal event extraction method that sythetically employs rules and deep learning. Secondly, the causal events are connected globally to build a causal knowledge network. Then, based on the graph embedding, a feature representation of the causal knowledge network is performed. Finally, we analyze knowledge community differences and identified potential causal events. [Results] The results show that causal networks can realize medical knowledge logical association more comprehensively, and can correlate local information from single literature into global knowledge elements. Moreover, our study can discover the knowledge of potential medical causality, which provides an important reference for disease diagnosis and treatment and academic innovation.