Beyond the Black Box: An Interactional Transparency Audit Framework for AI-driven Recommendation Systems via Actor-Network Theory

AI-driven recommendation (AIR) systems have become the foundational infrastructure of the digital economy, profoundly reshaping how information is disseminated and consumed.However, their pervasive application has created unprecedented governance challenges. Due to their black box nature and dynamic characteristics,these systems often operate opacity that evades traditional oversight, rendering static transparency solutions ineffective. To address this auditability challenge, we reconceptualizes transparency not as static information disclosure, but as a dynamic interactional process within a socio-technical network through introducing Actor Network Theory. Grounded in this theoretical lens, we propose an executable transparency audit framework. Wethen pioneer a computational synthesis approach, utilizing Natural Language Processing techniques to distill core transparency requirements from a multi-source corpus of regulatory mandates and expert interviews to construct a verifiable assessment checklist.The framework is empirically validated through a multi-case study of Chinese social media platforms. The results reveal the pervasive phenomenon of "paper compliance", a strategy where platforms adhere to regulative mandates through formalized compliance while neglecting the intrinsic intent. Further analysis delineates two distinct patterns of this symbolic compliance: "formal regulation-oriented" and "selectively interactive-oriented". In-depth experiment identifies insufficient interaction among platforms, users, and regulators triggering the lack of substantive transparency. This study enhances the concept of transparency from an interaction based perspective and developing an operational transparency audit framework. Moreover, the paper uncovers the manifestations and formation mechanisms of "paper compliance", offering a critical foundation for substantive governance.