Mapping collective knowledge flows in the digital public sphere: a multi-layer network approach to chatbot discourse

Abstract

Purpose – This study aims to examine the generation, diffusion and aggregation of collective knowledge on chatbots in the digital public sphere, focusing on multidimensional interactions that shape the structure and evolution of knowledge communities. Design/methodology/approach – A clustered multilayer User–Opinion–Topic–Keyword network model is developed to map and analyze knowledge flows related to chatbot on a social media platform. Unsupervised clustering, network metrics and Moran’s I spatial autocorrelation are used to assess patterns of knowledge creation, diffusion and aggregation. Findings – Results reveal tightly connected knowledge communities with strong internal cohesion and limited intercommunity knowledge exchange. User and topic layers are modular and efficient, while keyword and opinion layers provide semantic depth but slower long-distance flow. Moran’s I reveals significant positive spatial autocorrelation in knowledge popularity, reflecting reinforcement within localized clusters rather than broad integration across the network. Originality/value – This research advances knowledge management theory by presenting a scalable, multilayer framework for analyzing online knowledge exchange. Integrating structural and semantic perspectives, it provides novel insights into knowledge fragmentation and integration dynamics in the digital public sphere, offering practical implications for managing knowledge flows and fostering collaborative knowledge environments.