Akademik Veri Yönetim Sistemi
4th International Conference on Intelligent Systems and Pattern Recognition, ISPR 2024, İstanbul, Türkiye, 26 - 28 Haziran 2024, cilt.2303 CCIS, ss.51-64, (Tam Metin Bildiri)
In an effort to deliver exceptional customer service, organizations are increasingly recognising the critical role that advanced language models (LLMs) play in the integration of CRM systems. This article explores the forefront of chatbot technology, with a particular focus on the essential function they play in contemporary consumer relationship management. The ability of chatbots to provide personalized and contextually pertinent responses to customer inquiries has been significantly enhanced by the rapid development of LLMs, specifically in the areas of sentiment, intent, and context comprehension. Organizations have the opportunity to enhance customer contentment and optimize operations by leveraging the capabilities of these advanced AI systems. In addition, offline capabilities guarantee continuous support, thereby enhancing customer confidence and loyalty in a time when connectivity fluctuations continue to be a challenge. The present study introduces an innovative offline chatbot system that aims to overcome the limitations of traditional cloud-based counterparts. By efficiently integrating data from various sources, such as project documentation and social media, this chatbot is capable of operating independently and delivering assistance to users, even in locations with sporadic internet connectivity. The experimental outcomes illustrate that our newly developed chatbot model outperforms established benchmarks, as evidenced by its 4.1-second inference time, 90.2-point BLEU score, and 9.7-point WER score. These metrics underscore the model’s effectiveness, precision, and timeliness in handling user inquiries and producing responses of exceptional quality. The experiments we conduct are designed to validate the effectiveness of our model in improving customer service experiences within CRM systems through comprehensive performance testing.