MENG Xi, SHANG Shuo, GUO Yajun
LIBRARY TRIBUNE. 2025, 45(12): 108-119.
Identifying the key factors that influence algorithmic risk and analyzing their interactions are significant for creating a scientific and reasonable governance system for algorithmic risk. This paper establishes an integrated TOE-H model by incorporating the human (H) factor into the technology-organization-environment (TOE) framework. Next,an indicator system for algorithmic risk factors is created based on a literature analysis and the Delphi method. Then,the relationships among these factors are quantitatively analyzed using the Decision Making Trial and Evaluation Laboratory (DEMATEL) and Interpretive Structural Modeling (ISM) methods to explore the factors' influence and criticality. Thus,a hierarchical structure model of the influencing factors is developed. The results show that,among the 20 factors,the most influencial ones are capital-driven decision logic,legal regulatory measures,algorithmic ethical norms,algorithmic evaluation mechanisms,overreliance on algorithmic intelligence,developer cognitive biases,and algorithm literacy levels. In the hierarchical structure model,legal regulatory measures,algorithmic ethical norms,algorithmic evaluation mechanisms,developer cognitive biases,and algorithm literacy levels are root factors,while capital-driven decision logic and overreliance on algorithmic intelligence are surface factors. Based on these findings,corresponding governance strategies are proposed.