[2604.04997] Evaluation of Embedding-Based and Generative Methods for LLM-Driven Document Classification: Opportunities and Challenges

Computer Science > Information Retrieval arXiv:2604.04997 (cs) [Submitted on 5 Apr 2026] Title:Evaluation of Embedding-Based and Generative Methods for LLM-Driven Document Classification: Opportunities and Challenges Authors:Rong Lu, Hao Liu, Song Hou View a PDF of the paper titled Evaluation of Embedding-Based and Generative Methods for LLM-Driven Document Classification: Opportunities and Challenges, by Rong Lu and 2 other authors View PDF HTML (experimental) Abstract:This work presents a comparative analysis of embedding-based and generative models for classifying geoscience technical documents. Using a multi-disciplinary benchmark dataset, we evaluated the trade-offs between model accuracy, stability, and computational cost. We find that generative Vision-Language Models (VLMs) like Qwen2.5-VL, enhanced with Chain-of-Thought (CoT) prompting, achieve superior zero-shot accuracy (82%) compared to state-of-the-art multimodal embedding models like QQMM (63%). We also demonstrate that while supervised fine-tuning (SFT) can improve VLM performance, it is sensitive to training data imbalance. Comments: Subjects: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG) Cite as: arXiv:2604.04997 [cs.IR]   (or arXiv:2604.04997v1 [cs.IR] for this version)   https://doi.org/10.48550/arXiv.2604.04997 Focus to learn more arXiv-issued DOI via DataCite (pending registrati...