TSOKG: A Methodology for Constructing a Telecommunications Service Operation Knowledge Graph

The operation of telecommunications carriers involves managing a vast range of information across multiple domains, including network resources, equipment alerts, network performance, critical configurations, and operational logs. Current marketing and customer service efforts demand more efficient multi-domain knowledge processing. Extracting business logic from active systems is critical for restructuring resources and developing tailored operational solutions. This paper presents a method for constructing a Telecommunication Service Operation Knowledge Graph (TSOKG). By using a multi-round iterative process, the method integrates multi-layered data to build an ontology model from the top down, while utilizing a deep learning-based text entity relationship extraction approach to gather knowledge from the bottom up. The TSOKG has successfully connected over 200 interfaces across 27 information systems, consolidating approximately 2.595 million data points. Compared to mainstream algorithms, our model demonstrates superior accuracy, validating its effectiveness in mining and integrating structured data across sources and domains that share the same data structure.