[Purpose/Significance] This study proposes a multimedia resource cataloging solution based on multimodal large models to address the efficiency bottleneck in the cataloging of CD audio and video publications. An empirical study was conducted using CD audio and video publications as an example to improve cataloging efficiency and data processing quality. [Method/Process] First,the solution integrates large vision models and large language models(LLMs) to recognize and extract information from the physical appearance of optical disc publications. It then employs video toolkits(e.g., FFmpeg) to generate standardized cataloging metadata. Finally, the M3-Embedding model is applied to compare the generated data with manual data. [Result/Conclusion] Experimental results demonstrate that the similarity between the generated core metadata and manual cataloging data reaches 74.3%, which can effectively assist catalogers in pre-cataloging and improve work efficiency. Based on this design, a multimodal large model-powered multimedia resource cataloging assistant software was developed, aiming to enhance cataloging efficiency and accuracy through platform implementation.
Qian Guofu
,
Zhang Zhidong
,
Li Gui
. The Exploration of Multimedia Resource Cataloging System Based on Multimodal Large Models[J]. Library and Information Service, 2026
, 70(9)
: 105
-113
.
DOI: 10.13266/j.issn.0252-3116.2026.09.009
[1] 杨新涯,戴立伟,钱国富. DeepSeek在图书馆的应用场景构架研究[J].图书馆论坛, 2025, 45(10):90-98.(YANG X Y, DAI L W, QIAN G F. Research on the application scenario architecture of DeepSeek in libraries[J]. Library tribune, 2025,45(10):90-98.)
[2] 张新新,游恒飞.从ChatGPT到DeepSeek:多模态大模型与出版业未来--兼论走向"十五五"的出版智能化转型[EB/OL].[2025-09-15]. https://www.cnki.com.cn/Article/CJFDTotal-BJZG20250409001.htm.(ZHANG X X, YOU H F.From ChatGPT to DeepSeek:multimodal large models and the future of publishing industry-with a discussion on the intelligent transformation of publishing towards the"15th Five-year Plan"[EB/OL].[2025-09-15]. https://www.cnki.com.cn/Article/CJFDTotal-BJZG20250409001.htm)
[3] 赵美花.人工智能在图书采分编领域的实践与展望--以广东省立中山图书馆"采编图灵"为例[J].图书馆论坛, 2025,45(8):126-132.(ZHAO M H. Practice and prospect of artificial intelligence in the field of acduisition and cataloging-taking "library turing" as an example[J]. Library tribune, 2025, 45(8):126-132.)
[4] PaddlePaddle. PaddleOCR documents[EB/OL].[2025-09-15]. https://paddlepaddle.github.io/PaddleOCR/latest/.
[5] LAMM B, KEUPER J. Can visual language models replace OCR-based visual question answering pipelines in production?a case study in retail[EB/OL].[2025-09-15]. https://arxiv.org/abs/2408.15626.
[6] ZHANG J, HUANG J, JIN S, et al. Vision-language models for vision tasks:a survey[J]. IEEE TPAMI, 2024, 46(8), 5625-5644.
[7] XU D, CHEN W, PENG W, et al. Large language models for generative information extraction:a survey[J]. Frontiers of computer science, 2024, 18(6):186357.
[8] BAI S, CHEN K, LIU X, et al. Qwen2.5-VL Technical Report[EB/OL].[2025-09-15]. https://ArXiv.org/abs/2502.13923.
[9] WEI H, LIU C, CHEN J, et al. General OCR theory:towards OCR-2.0 via a unified end-to-end model[EB/OL].[2025-09-15]. https://ArXiv.org/abs/2409.01704.
[10] SEBASTIAN RASCHKA. Understanding reasoning LLMs[EB/OL].[2025-09-15]. https://magazine.sebastianraschka.com/p/understanding-reasoning-llms.
[11] DeepSeek-AI. DeepSeek-R1:Incentivizing reasoning capability in LLMs via reinforcement learning[EB/OL].[2025-09-15]. https://doi.org/10.48550/arXiv.2501.12948.
[12] QWEN Team. QwQ-32B[EB/OL].[2025-09-15]. https://github.com/QwenLM/QwQ.
[13] Kimi Team. Kimi k1.5:scaling reinforcement learning with LLMs[EB/OL].[2025-09-15]. https://arxiv.org/pdf/2501.12599v2.
[14] 刘贝玲,张雨卉.大语言模型在图书馆采编业务中的应用展望[J].信息与管理研究, 2024, 9(5):14-23.(LIU B L, ZHANG Y H. Prospects of the application of large language models in library acquisition and cataloging business[J]. Journal of information and management, 2024, 9(5):14-23.)
[15] CHOW E H C, KAO T J, LI X L. An experiment with the use of ChatGPT for LCSH subject assignment on electronic theses and dissertations[J]. Cataloging&classification quarterly, 2024,62(5):574-588.
[16] OpenAI. API Reference-OpenAI API[EB/OL].[2025-09-15]. https://platform.openai.com.
[17] Flutter team. Flutter:build apps for any screen[EB/OL].[2025-09-15]. https://flutter.dev/.
[18] XU X, LI M, TAO C, et al. A survey on knowledge distillation of large language models[EB/OL].[2025-09-15]. https://ArXiv.org/abs/2402.13116.
[19] PROTTASHA N J, CHOWDHURY U R, MOHANTO S, et al. PEFT A2Z:parameter-efficient fine-tuning survey for large language and vision models[EB/OL].[2025-09-15]. https://arxiv.org/abs/2504.14117.
[20] GUPTA S, RANJAN R, SINGH S N. A comprehensive survey of retrieval-augmented generation (RAG):evolution, current landscape and future directions[EB/OL].[2025-09-15]. https://arXiv.org/abs/2410.12837.
