LARGE LANGUAGE MODELS FOR SERVICE-ORIENTED COMPUTING: A CASE STUDY IN TELECOM CUSTOMER SUPPORT SYSTEMS
Merita KASA HALILI, Donat REÇICA, Festim HALILI, Gazmend XHAFERI
Abstract
Increasingly complex telecommunications systems demand the application of intelligent and efficient methods to ensure smooth operations, sound decision-making and high-level services. With the increasing paradigm of Service-Oriented Computing (SOC) - where systems are composed of distributed services, the management of such services is becoming an even greater challenge in the customer support domain. One of the key obstacles in customer support is obtaining real time diagnostics of the problem and creating a detailed report that confirms the occurrence of the failure. This paper explores the functions and potential provided by large language models in the context of telecom customer support. The paper particularly explores the opportunities provided for automatic summarization of interactions, artificial intelligence in knowledge management and enhancement of the service diagnostics functionalities. The research is carried out through a case study, analyzing existing solutions applied in practical scenarios such as diagnostic platforms, CRM systems and AI assistants. The study explored human-in-the-loop approaches of agent feedback, which trains the AI to improve its output for better accuracy, adaptability and overall reliability. The findings revealed that Large Language Models (LLMs) boosted agent productivity, reduced human work and ensured consistency in communication. However, the output of the LLMs also presented some issues including inaccuracies and data privacy. Based on the study’s results, Large Language Models (LLMs) are actually used as a complement to human agents, enabling building more intelligent, adaptable and efficient service-oriented telecom systems.
Pages: 11 - 21