Türkçe Finansal RAG Uygulamalari için Gömme Modellerinin Karşilaştirilmasi

ÇETİN U., Colakgil I. E.

33rd IEEE Conference on Signal Processing and Communications Applications, SIU 2025, İstanbul, Türkiye, 25 - 28 Haziran 2025, (Tam Metin Bildiri) identifier

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Doi Numarası: 10.1109/siu66497.2025.11111893
  • Basıldığı Şehir: İstanbul
  • Basıldığı Ülke: Türkiye
  • Anahtar Kelimeler: Anahtar Kelimeler - NLP, Embedding, Finance, RAG
  • Galatasaray Üniversitesi Adresli: Evet

Özet

In this study, embedding models used in the retrieval phase, which play a critical role in Retrieval-Augmented Generation (RAG) systems, are compared on financial texts. The performances of open and closed source models are analyzed and their advantages, disadvantages, and usage scenarios are evaluated. In addition, important factors such as security and copyright risks of open source models and access restrictions of closed source models are discussed. Our study presents findings to improve the accuracy, reliability, and efficiency of RAG systems in the financial domain. In the financial sector, where fast access to accurate information is critical, the effects of embedding model selection are examined in detail and conclusions are drawn to guide future research and industrial applications.