Search academic texts

Information × Registration Number 2124U004685, Article popup.category Препринт Title popup.author Karaim Olena popup.publication 01-01-2024 popup.source_user Український католицький університет popup.source https://hdl.handle.net/20.500.14570/4855 popup.publisher Description Digital transformation is a crucial strategic plan element for various industries in to- day’s fast-paced and dynamic market, including the supply chain. Providing excel- lent customer service is an essential aspect of logistics management; thus, integrat- ing artificial intelligence into customer service is an effective solution. To enhance external communication with customers, AI chatbots that use Large Language Mod- els can lead to satisfactory user experience, although this subject has not yet been thoroughly explored. This study investigates the application of LLMs in develop- ing chatbot solutions for customer service within the logistics industry. The research aimed to employ an effective fine-tuning approach and compare leading LLMs, con- sidering customer satisfaction crucial for business growth. Various pre-trained mod- els, including Llama, Vicuna, and Mistral, were evaluated using parameter-efficient techniques like QLoRA for domain-specific fine-tuning. A dataset with logistics- related queries was employed to evaluate the models, utilizing multiple evaluation methods such as human reviews, cosine similarity, ROUGE, SacreBLEU, and Per- plexity measurements, and the innovative LLM-as-a-Judge approach. This study may assist logistics companies in making informed decisions regarding integrating LLMs into customer service. popup.nrat_date 2025-05-09 Close