Abstract: To address the issues of scattered data in power grid marketing and distribution (MD) operations and inefficient topology management this study focuses on the design and implementation of application service modules for the "MD One Map" system constructing an integrated service framework based on multi-source data fusion technology the module design establishes a three-layer system architecture-data layer algorithm layer and application layer-supported by a data middle platform and real-time measurement data: The data layer integrates multi-source data from marketing systems PMS (Power Management System) and power consumption collection systems the algorithm layer deploys core models such as medium-low voltage topology identification and automatic file verification the application layer develops three major functions: complete topology identification automatic file maintenance and intelligent line loss diagnosis enabling visual representation of "substation-line-transformer-box-meter-household" topological relationships automatic verification of file anomalies and multi-dimensional line loss analysis the system employs the Python + Spark framework for data computing Java for developing application service interfaces and the VUE framework for front-end visual display pilot validation in industrial parks demonstrates that the module achieves a topology identification accuracy rate of ≥90% and improves line loss anomaly location efficiency by 50% providing robust support for MD business coordination and digital management in power grids.