Layer-Sensitive Selective Merge In Multi-Source Bibliographic Graphs

This study presents a layer-sensitive selective merge mechanism applied to multi-source bibliographic graphs. The proposed method prevents improper collapses among heterogeneous en-tities—such as authors, papers, and venues—during the integration of data retrieved from multiple academic engines (e.g., Semantic Scholar, Open-Alex). By combining heuristic validation, canonical keys, and structural auditing, the model ensures a controlled, reproducible, and transparent fusion process. It was implemented within the GrafoCitas Abstract Data Type (TDA) and validated across several graph-merging scenarios, yielding a reduction in cross-layer entity collapses of approximately 50% and improving overall structural consistency across merged citation networks