中文    English

Journal of library and information science in agriculture

   

Large AI Model Utilization Optimization in Libraries Based on Multimodal Resource Profiling

QIN Miao1, WANG Qingfei2   

  1. 1. Henan University of Animal Husbandry and Economy Library, Zhengzhou 450046
    2. School of Information Management, Zhengzhou University, Zhengzhou 450001
  • Received:2025-05-15 Online:2025-09-17

RichHTML

1

PDF (PC)

2

Abstract:

[Purpose/Significance] With the rapid advancement of artificial intelligence (AI) technologies, libraries are transforming their service models and content offerings. Large AI models have opened up broader development opportunities for smart libraries. However, the rational adoption and application of these models has posed a significant challenge to libraries. This study employs multimodal resource profiling to conduct research on the optimization of large AI model utilization in libraries, revealing the intrinsic relationships among various types of library resource data. Based on these insights, the optimization methods and related strategies are extracted to enhance the efficiency of library resource utilization and improve user experience. [Method/Process] Multimodal resource profiling is a comprehensive representation that captures the intrinsic characteristics of library resources through tag extraction, aggregation analysis, and visualization of diverse data generated within the libraries. By utilizing a novel clustering algorithm, it overcomes the high sensitivity to input parameters characteristic of traditional algorithms and achieves natural clustering across resources with varying densities, thereby enabling the generation of accurate multimodal resource profiles. The resource profiling model provides a theoretical foundation for optimizing the deployment and utilization of large AI models in libraries, while also delivering rich data support for subsequent AI model applications. The adoption strategy proposed in this study is divided into two aspects: model selection and model utilization. Model selection focuses on compatibility and accuracy to achieve an optimal match between the model and both library resources and user needs. Model utilization emphasizes the effectiveness and usability of the output, thereby enhancing operational efficiency and user experience. Based on this framework, the overall operational mechanism of the adoption optimization strategy is designed around continuous model monitoring, real-time collection of user feedback, iterative model updates, and dynamic adjustment of multimodal resource profiles. [Results/Conclusions] This study takes a public digital library on "Telegram" as a case study to generate multimodal resource profiles, which meticulously categorize user groups, interests, and emotional intensities. By integrating the large AI model adoption optimization strategy with the outcomes of multimodal resource profiling, the model autonomously identifies the most task-relevant features, reducing the need for manual intervention. Not only does it achieve high prediction accuracy, but the explanatory feature weights it outputs also provide a quantifiable basis for service optimization. Through comparative experiments with commonly used structural modules, the proposed method demonstrates significant advantages over traditional recommendation systems in terms of both resource utilization efficiency and user engagement. This study lays the foundation for the future development of library technology and opens up new possibilities for the application of multimodal resource profiling.

Key words: multimodal resource profiling, libraries, AI large models, resource optimization, user experience, information behavior

CLC Number: 

  • G258

Fig.1

The process of constructing an large AI model and the path to empowerment for libraries"

Table 1

Library resource profiling construction objectives"

序号 分类 二级分类 目标释义 数据来源
目标1 用户资源分析 基本信息 用户的个人信息，如性别、年龄及地域 综合管理系统
行为习惯 用户行为记录，如入馆时间、借阅记录 文献流通系统
交互活动 用户与图书馆的互动数据，如咨询、反馈 社交网络平台
目标2 馆藏资源分析 文本文献 纸质馆藏中的文本资源，如图书、期刊 综合管理系统
图片文献 纸质馆藏中的图片资源，如画稿、照片 综合管理系统
音频视频 纸质馆藏中的音视频资源，如碟片、磁带 综合管理系统
目标3 服务资源分析 服务内容 图书馆读者服务的种类及内容 官方门户网站
资源配置 开展服务所需配置，如人员、设备 文件通知公告
标准制度 服务开展过程中遵循的流程及制度 文件通知公告
目标4 数字资源适配 商业化资源 图书馆需要付费使用的数字资源 数字资源平台
非商业化资源 图书馆自主建立或免费开放的资源 数字资源平台

Fig.2

Resource-associated user label scale processing diagram"

Table 2

Unified standards for resource-associated user label scales"

分类 子标签 标准依据 标签标度1 对应标准1 标签标度2 对应标准2 标签标度3 对应标准3
基本信息 性别标签 直接输出 男性标签 直接输出 女性标签 直接输出 Null Null
年龄标签 年龄层次 年轻用户 <24周岁 青年用户 25~45岁 中老年人 >46周岁
地域标签 行政划分 东部地区 东部省市 南部地区 南部省市 北部地区 北部省市
行为习惯 登录频率 登陆次数 偶尔登录 <5天/次 定期登录 2~4天/次 频繁登录 >1天/次
访问时长 持续时间 短暂访问 <2min/次 中等访问 3~15min/次 深度访问 >15min/次
使用服务 服务种类 单一服务 1种类型 多项服务 多种类型 Null Null
内容偏好 收藏记录 收藏类型 文本偏好 文本类型 视频偏好 视频类型 音频偏好 音频类型
检索记录 检索领域 窄向检索 单一领域 广泛检索 多项领域 Null Null
阅读记录 阅读篇幅 轻度偏好 碎片阅读 中度偏好 全文阅读 强烈偏好 大量阅读
UGC数据 互动频率 互动次数 低频互动 <1次/天 中度互动 2~5次/天 高频互动 >6次/天
内容类型 内容结构 单一类型 单一结构 多种类型 多元结构 Null Null
情感特征 情感倾向 积极情感 正向内容 消极情感 负向内容 理性情感 陈述内容

Fig.3

Resource-associated quality label scale processing diagram"

Table 3

Unified standards for resource-associated feature label scales"

子标签 标准依据 标签标度1 对应标准1 标签标度2 对应标准2 标签标度3 对应标准3
主题标签 词汇聚类 主题词1 S(R)<0.6 主题词2 S(R)<0.4 主题词3 S(R)<0.2
形式标签 文档类型 文本资源 文本类型 图像资源 音、视频 音视频类l null
主题关联 主题相似 高相似度 >80% 中等相似 30%~79% 低相似度 <29%
词项关联 词项重叠 高重叠度 >60% 中重叠度 20%~59% 低重叠度 <19%
引用关系 资料引用 双向引用 相互引用 单向引用 单一被引 复合引用 null
时间关联 时间划分 近期相关 <1年 中期相关 1~5年 长期相关 >5年

Fig.4

Resource-associated service label scale processing diagram"

Fig.5

AI large model optimization hierarchical structure diagram"

Fig.6

Overall operational mechanism diagram of library large AI model"

Table 4

Preliminary statistical data of public digital libraries in “Telegram”"

初步统计 统计类型 预期统计1 预期结果1 预期统计2 预期结果2 预期统计3 预期结果3
行为习惯 使用频率 偶尔使用 F(t)＞5d 定期使用 F(t)≤5d 经常使用 F(t)＜1d
访问时长 快速访问 F(t)＜1min 适度访问 F(t)≥15min 持续访问 F(t)＞15min
服务类型 临时服务 咨询类型 常用服务 借阅类型 多项服务 多种类型
内容偏好 收藏记录 音频类型 音频偏好 文本类型 文件偏好 视频类型 视频偏好
搜索记录 临时搜索 需求事件 历史搜索 回顾内容 趋势搜索 热门话题
阅读记录 快速阅读 F(t)＜1min 连续阅读 F(t)≥10min 深度阅读 F(t)＞10min
UGC数据 互动频率 低频互动 F(t)＜t 中频互动 F(t)≥5t 高频互动 F(t)＞6t
内容类型 主题结构 集中研究 层次结构 理解导航 叙事结构 沉浸参与
情感特征 情感强度 情感表达 情感变化 反映情绪 情感共鸣 引发情感

Fig.7

Generation results of multimodal resource profiling in "Telegram" public digital library"

Fig.8

Optimization scheme intervention results under AI large model"

Table 5

Performance comparison experimental results of corresponding methods for structural modules"

方法类型 结构类型 对比方法1 对比方法2 对比方法3 降低率/% 增长率/% 优势率/%
统计方法 描述推断 Skewness Bar Chart Mean -3.98 18.73 7.98
时间方差 STL分解 ADF检验 X-12-ARIMA -5.55 22.82 12.75
聚类因子 K-Means DBSCAN HDBSCAN -2.11 6.01 1.41
画像方法 标签行为 LabelMe Deep Lab Transformer -4.30 12.11 10.97
情感模态 SWN-LIWC Emo Dataset open-faced -2.88 7.36 4.23
动态关系 ARIMA Optical Flow ARIMA模型 -2.01 19.75 21.44
取用方法 检索推荐 TF-IDF F1-score CARS感知 -9.67 7.34 6.64
语义推送 Word2Vec Siamese Sentence -7.37 13.97 12.28
定制处理 NLG语言 Z-score标准 Random Search -6.58 11.57 9.56
[1]
智慧图书馆技术应用联盟. 《智慧图书馆大模型创新与应用白皮书》（征求意见版）发布[EB/OL]. [2024-12-05].
[2]
李兵. 基于查询意图识别的自适应图书分面检索研究[J]. 图书馆学研究, 2017(15): 57-64.
LI B. Self-adaptive book information faceted search based on query intent identification[J]. Research on library science, 2017(15): 57-64.
[3]
张铭洁, 赵瑞雪. ChatGPT驱动的智慧图书馆情感感知与服务优化[J]. 农业图书情报学报, 2024, 36(12): 74-88.
ZHANG M J, ZHAO R X. Emotion perception and service optimization in ChatGPT-Driven smart libraries[J]. Journal of library and information science in agriculture, 2024, 36(12): 74-88.
[4]
杨帆. 画像分析为基础的图书馆大数据实践: 以国家图书馆大数据项目为例[J]. 图书馆论坛, 2019, 39(2): 58-64.
YANG F. Library big data practice based on user profile analysis: A case study of national library of China big data project[J]. Library tribune, 2019, 39(2): 58-64.
[5]
徐海玲, 张海涛, 魏明珠, 等. 社交媒体用户画像的构建及资源聚合模型研究[J]. 图书情报工作, 2019, 63(9): 109-115.
XU H L, ZHANG H T, WEI M Z, et al. Research on the construction of social media user portrait and resource aggregation model[J]. Library and information service, 2019, 63(9): 109-115.
[6]
刘芳, 朱沙. 基于读者画像的高校图书馆精准服务研究[J]. 大学图书情报学刊, 2020, 38(1): 73-75.
LIU F, ZHU S. Research on accurate service based on reader portrait of university library[J]. Journal of academic library and information science, 2020, 38(1): 73-75.
[7]
刘海鸥, 李凯, 姜波. 移动图书馆推荐系统中的用户画像与资源画像情境化融合研究[J]. 图书馆, 2021(6): 66-71, 93.
LIU H O, LI K, JIANG B. Research on contextualization integration of user portrait and resource portrait in mobile library recommendation system[J]. Library, 2021(6): 66-71, 93.
[8]
张壮, 冯小年, 钱铁云. 基于多模态融合技术的用户画像方法[J]. 北京大学学报(自然科学版), 2020, 56(1): 105-111.
ZHANG Z, FENG X N, QIAN T Y. User profiling based on multimodal fusion technology[J]. Acta scientiarum naturalium universitatis pekinensis, 2020, 56(1): 105-111.
[9]
曹树金, 岳文玉. 面向精准服务的图书馆用户画像研究[J]. 农业图书情报学报, 2021, 33(10): 4-19.
CAO S J, YUE W Y. Research on library user profiles for precision services[J]. Journal of library and information science in agriculture, 2021, 33(10): 4-19.
[10]
张俊, 徐箭, 许沛东, 等. 人工智能大模型在电力系统运行控制中的应用综述及展望[J]. 武汉大学学报(工学版), 2023, 56(11): 1368-1379.
ZHANG J, XU J, XU P D, et al. Overview and prospect of application of artificial intelligence large model in power system operation control[J]. Engineering journal of Wuhan university, 2023, 56(11): 1368-1379.
[11]
王鸑飞, 卢垚, 袁雪, 等. 西部地区科技文献资源保障策略研究[J]. 农业图书情报学报, 2023, 35(12): 94-103.
WANG Y F, LU Y, YUAN X, et al. Security strategy of scientific and technological literature resources in western regions of China[J]. Journal of library and information science in agriculture, 2023, 35(12): 94-103.
[12]
涂佳琪, 杨新涯, 沈敏. 需求与决策驱动的图书智能采访系统研究与实践: 以重庆大学图书馆为例[J]. 图书情报工作, 2020, 64(11): 28-34.
TU J Q, YANG X Y, SHEN M. Research and practice of decision-driven book intelligent interview system: Taking Chongqing University as an example[J]. Library and information service, 2020, 64(11): 28-34.
[13]
柳益君, 罗烨, 蔡秋茹, 等. 基于机器学习的高校图书馆个性化智能推荐服务方案[J]. 图书馆研究与工作, 2020(3): 28-34.
LIU Y J, LUO Y, CAI Q R, et al. Personalized intelligent recommendation service scheme of university library based on machine learning[J]. Library science research & work, 2020(3): 28-34.
[14]
寿建琪. 走向“已知之未知”：GPT大语言模型助力实现以人为本的信息检索[J]. 农业图书情报学报, 2023, 35(5): 16-26.
SHOU J Q. Towards known unknowns: GPT large language models empower human-centered information retrieval[J]. Journal of library and information science in agriculture, 2023, 35(5): 16-26.
[15]
沈奎林, 邵波, 陈力军, 等. 基于超高频RFID的图书盘点机器人的设计和实现[J]. 图书馆学研究, 2016(7): 24-28.
SHEN K L, SHAO B, CHEN L J, et al. The design and implementation of book inventory robot based on ultra high frequency RFID[J]. Research on library science, 2016(7): 24-28.
[16]
辛沅霞, 华道阳, 张犁. 基于智能规划的多智能体强化学习算法[J]. 计算机科学, 2024, 51(5): 179-192.
XIN Y X, HUA D Y, ZHANG L. Multi-agent reinforcement learning algorithm based on AI planning[J]. Computer science, 2024, 51(5): 179-192.
[17]
郑人荣, 王钰, 梁承姬. 时变需求下汽车零部件整包配送库存路径优化[J]. 计算机集成制造系统, 2025, 31(1): 367-383.
ZHENG R R, WANG Y, LIANG C J. Inventory routing problem of packed auto parts considering time-varying demands[J]. Computer integrated manufacturing systems, 2025, 31(1): 367-383.
[18]
刘晓鹏, 陈炫锐, 席少辉, 等. 基于分解协调机制的多层单元流水式车间布局优化方法[J]. 计算机集成制造系统, 2024, 30(5): 1823-1833.
LIU X P, CHEN X R, XI S H, et al. Decomposition-coordination method for multi-floor flow shop facility layout problem[J]. Computer integrated manufacturing systems, 2024, 30(5): 1823-1833.
[19]
蔡毅, 唐振鹏, 吴俊传, 等. 基于灰狼优化的混频支持向量机在股指预测与投资决策中的应用研究[J]. 中国管理科学, 2024, 32(5): 73-80.
CAI Y, TANG Z P, WU J C, et al. Research on the application of GWO-SVR algorithm in the prediction of reverse mixed data in stock market and investment strategy[J]. Chinese journal of management science, 2024, 32(5): 73-80.
[20]
李煜, 梁晓, 刘景森, 等. 基于改进平衡优化器算法求解工程优化问题[J]. 计算机集成制造系统, 2025, 31(8): 2964-2982.
LI Y, LIANG X, LIU J S, et al. Solving engineering optimization problem based on modified equilibrium optimizer algorithm[J]. Computer integrated manufacturing systems, 2025, 31(8): 2964-2982.
[21]
梁玉芳, 刘凡儒. 人工智能时代的图书馆: 技术、问题及应用[J]. 情报资料工作, 2018, 39(5): 107-112.
LIANG Y F, LIU F R. Libraries in the era of artificial intelligence: Technology, problems and applications[J]. Information and documentation services, 2018, 39(5): 107-112.
[22]
张若雅, 储开稳, 徐旭光, 等. 高校图书馆用户画像构建与精准服务模式研究[J]. 图书馆学刊, 2023, 45(9): 57-62.
ZHANG R Y, CHU K W, XU X G, et al. Research on user portrait construction and accurate service mode in university library[J]. Journal of library science, 2023, 45(9): 57-62.
[23]
郭宇, 孙振兴, 刘文晴, 等. 基于数据驱动的移动图书馆UGC用户画像研究[J]. 情报理论与实践, 2022, 45(1): 30-37.
GUO Y, SUN Z X, LIU W Q, et al. Research on UGC user portrait of mobile library based on data driven[J]. Information studies: Theory & application, 2022, 45(1): 30-37.
[24]
丁锋, 郑嘉芸, 张霄, 等. 基于数据滤波的随机梯度辨识方法[J]. 控制与决策, 2024, 39(7): 2259-2266.
DING F, ZHENG J Y, ZHANG X, et al. Filtering-based stochastic gradient identification methods[J]. Control and decision, 2024, 39(7): 2259-2266.
[25]
周林兴, 王帅. 危机情境下公共文化数据开放平台多重耦合资源画像研究[J]. 情报科学, 2023, 41(8): 71-80, 88.
ZHOU L X, WANG S. Multi coupling resource portrait of public cultural data open platform in crisis situation[J]. Information science, 2023, 41(8): 71-80, 88.
[26]
孙林, 刘梦含, 徐久成. 基于优化初始聚类中心和轮廓系数的K-means聚类算法[J]. 模糊系统与数学, 2022, 36(1): 47-65.
SUN L, LIU M H, XU J C. K-means clustering algorithm using optimal initial clustering center and contour coefficient[J]. Fuzzy systems and mathematics, 2022, 36(1): 47-65.
[1] ZHAO Yajing. A Study of the Factors Influencing Participation Behavior among Users with Depression on User-Generated Content (UGC) Platforms [J]. Journal of library and information science in agriculture, 2025, 37(6): 70-86.
[2] XIAO Keyi, LI Yunfan, CHEN Yingying, PENG Xi. User Experience of Public Library Intelligent Service from the Perspective of Actor Network Theory [J]. Journal of library and information science in agriculture, 2025, 37(4): 51-65.
[3] SANG Yuanyuan. Multimodal Learning Technology Aimed at Exploring the Innovative Path of Library Intelligence Service [J]. Journal of library and information science in agriculture, 2025, 37(3): 42-52.
[4] GOU Ruike, LUO Wei. Influencing Factors of Continuous Use Intention of "Generation Z" Users of an AIGC Platform [J]. Journal of library and information science in agriculture, 2025, 37(3): 66-80.
[5] LI Xiao, QU Jiansheng. Influencing Factors of User Participation Intention of Crowdsourcing in Evidence Synthesis [J]. Journal of library and information science in agriculture, 2025, 37(3): 92-105.
[6] QIAO Jinhua, MA Xueyun. Risks and Regulations for Application of the LLaMA Model in University Future Learning Centers [J]. Journal of library and information science in agriculture, 2025, 37(2): 37-48.
[7] SHI Qin, XIE Jing, WU Shang. Influencing Factors and Correlations of User Satisfaction with Mobile Health Applications [J]. Journal of library and information science in agriculture, 2025, 37(1): 33-46.
[8] LUO Guofeng, YI Tong, YAN Zhouzhou. Critical Information Literacy Education in University Libraries from the Perspective of AIGC Application [J]. Journal of library and information science in agriculture, 2025, 37(1): 47-58.
[9] YOU Ge, LI Jielin, ZHANG Fangshun. Generating Mechanism of Online Public Opinion Heat in Public Emergencies from the Perspective of Information Ecology: Fuzzy Set Qualitative Comparative Analysis Based on 50 Cases [J]. Journal of library and information science in agriculture, 2025, 37(1): 86-99.
[10] Jiaxin HUANG, Xiaofang ZHANG. Application Models and Innovative Approaches of Smart Libraries from the Perspective of MR Technology [J]. Journal of library and information science in agriculture, 2024, 36(9): 78-88.
[11] Keyi XIAO, Yingying CHEN. Scientific Data Management Based on a Data Life Cycle Perspective: Using the Institutional Repositories Base of 24 Universities in the United States as an Example [J]. Journal of library and information science in agriculture, 2024, 36(7): 88-99.
[12] Guowei GAO, Shanshan ZHANG, Jialan YU. A Review of Health Information Behaviors of Older People from the Perspective of Topic Differentiation [J]. Journal of library and information science in agriculture, 2024, 36(7): 34-49.
[13] Xiaolin ZHANG. Beyond Resources, Beyond Technologies, Beyond One's Institution: Developing New Productive Forces for Knowledge Services through Reform and Innovation of Traditional Knowledge Service Mechanisms [J]. Journal of library and information science in agriculture, 2024, 36(6): 4-15.
[14] Liqin YAO, Hai ZHANG. Model Construction and Empirical Research on the Influencing Factors of AIGC User Dropout Behavior [J]. Journal of library and information science in agriculture, 2024, 36(5): 79-92.
[15] Chunling GAO, Liyuan JIANG. Elderly People's Online Health Information Seeking Behavior Based on Evolutionary Dynamics [J]. Journal of library and information science in agriculture, 2024, 36(5): 65-78.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
Editorial Board of Journal of library and information science in agriculture
Tel：(010)82109667　Fax：(010)82109667　E-mail：xuekan＠caas.cn
Powered by Beijing Magtech Co., Ltd.
知识共享许可协议