en.Wedoany.com Reported - The Global System for Mobile Communications Association (GSMA), in collaboration with French infrastructure organization Pleias, has developed the "Telco Common Corpus," a compilation of telecom-related data and technical resources containing over 10 billion tokens, aimed at addressing the shortcomings of existing AI models in telecommunications capabilities.
The GSMA noted that current benchmark tests and platform work for AI have revealed deficiencies in multiple areas, including network management. To address these gaps, the association has partnered with Pleias to advance this corpus project.
The corpus will be freely available, providing technical literature, patents, open data, and network projects. The GSMA explained that such information is difficult to obtain directly through traditional research methods, and attempts to use existing frontier models typically cover only non-telecom domains.
The GSMA committed that the corpus will provide the open training data needed for the industry to build high-performance models and open data infrastructure. All materials have been verified, and the project is considered a first attempt to integrate a "public telecommunications knowledge base." The GSMA emphasized that general-purpose and frontier models remain weak in handling tasks related to real-world network operations, such as troubleshooting or reasoning using Third Generation Partnership Project (3GPP) processes. The corpus aims to address these issues by providing training data for specialized telecom models and generating specific data based on telecom heuristic rules rather than random data scraping.
Additionally, the GSMA stated that the corpus will evolve with the release of new materials, helping to meet growing requirements for source traceability, thereby benefiting operators, equipment vendors, research institutions, and regulatory bodies.
This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com
