陈浩楠,张乐,许央科,梁高山

• 1:北京信息科技大学计算机学院

摘要(Abstract)：

针对传统旅游信息检索系统难以深度挖掘与理解用户口碑信息的局限，提出一种景区口碑知识图谱构建方法。利用大语言模型与精细化提示工程，从海量非结构化评论文本中自动化抽取关键实体、识别实体间的语义关系，并同步完成方面级细粒度情感分析。提出RBHA(RoBERTa-BiLSTM-hybrid-attention)模型，通过融合RoBERTa(a robustly optimized BERT pretraining approach)的动态编码、双向长短期记忆（bidirectional long short-term memory, BiLSTM）网络的上下文特征提取以及混合注意力（hybrid-attention）机制，实现对评论文本的整体性粗粒度情感分析。通过聚合与量化分析多维抽取与情感分析结果，为各景区实体构建多维度、可量化的口碑特征画像。采用混合策略融合多源异构数据，将结构化口碑知识存储于Neo4j图数据库以实现高效检索。研究结果表明，该方案能够有效提升旅游信息服务的智能化与个性化推荐水平，具有较高的应用前景与实用价值。

关键词(KeyWords)： 景区口碑知识图谱;大语言模型;情感分析;实体关系抽取

Abstract:

Keywords:

基金项目(Foundation): 北京市教委科研计划科技一般项目（KM202311232001）;; 省部共建藏语智能信息处理及应用国家重点实验室、藏文信息处理教育部重点实验室（青海师范大学）开放课题（2024-Z-005）

作者(Author): 陈浩楠,张乐,许央科,梁高山

DOI: 10.16508/j.cnki.11-5866/n.2025.06.011

参考文献(References)：

• [1]李东，邓秋菊.基于大语言模型的虚假评论检测研究综述[J].现代信息科技，2025,9(21):59-64.LI D,DENG Q J. A review of fake review detection based on large language model[J]. Modern Information Technology,2025,9(21):59-64.(in Chinese)
• [2]彭纪扬，郑昂.基于Neo4j的湘西地区旅游知识图谱构建研究[J].科技资讯，2025,23(8):62-66.PENG J Y,ZHENG A. Research on the construction of a tourism knowledge graph in the Xiangxi region based on Neo4j[J].Science&Technology Information,2025,23(8):62-66.(in Chinese)
• [3]张云中，李茜.沪上名人故居知识图谱构建与应用研究[J].情报科学，2023,41(10):1-11.ZHANG Y Z,LI X. The construction and application of knowledge map of the former residences of celebrities in Shanghai[J]. Information Science,2023,41(10):1-11.(in Chinese)
• [4]黄瑞潇.典型固体废物指纹特征提取与分类溯源模型研究[D].北京：中国环境科学研究院，2025.HUANG R X. Typical solid waste fingerprint feature extraction and classification research on traceability model[D]. Beijing:Chinese Research Academy of Environmental Sciences,2025.(in Chinese)
• [5]朱亚萍.面向水利工程抢险文本的实体关系联合抽取方法研究[D].郑州：华北水利水电大学，2024.ZHU Y P. Research on the joint extraction method of rescue entities relationships in water conservancy project text[D].Zhengzhou:North China University of Water Resources and Electric Power,2024.(in Chinese)
• [6]薛亚东，侯颖材，曲立清，等.基于本体空间的海底隧道施工风险动态管控机制[J].工程地质学报，2025,33(3):1121-1130.XUE Y D,HOU Y C,QU L Q,et al. Dynamic risk management framework for subsea tunnel construction based on ontologydriven approaches[J]. Journal of Engineering Geology,2025,33(3):1121-1130.(in Chinese)
• [7]王世文，刘劲.基于本体的重大突发事件网络舆情案例数据库数据模型研究[J].情报理论与实践，2023,46(10):163-173.WANG S W,LIU J. Research on ontology-based data model of network public opinion case database for major emergencies[J].Information Studies:Theory&Application,2023,46(10):163-173.(in Chinese)
• [8]成贵卿，孙晓龙，郗宵，等.基于SQL server的中枢神经系统损伤康复数据库构建与应用[J/OL].空军军医大学学报，2025:1-11.(2025-06-23)[2025-07-18]. https：//link. cnki.net/urlid/61. 1526. R. 20250621. 1430. 002.CHENG G Q,SUN X L,XI X,et al. Construction and application of central nervous system injury rehabilitation database based on SQL server[J/OL]. Journal of Air Force Medical University,2025:1-11.(2025-06-23)[2025-07-18]. https：//link. cnki.net/urlid/61. 1526. R. 20250621. 1430. 002.(in Chinese)
• [9]黄敬茹，刘智宇，刘瑞瑞，等.红色文化遗产的文化基因挖掘及其数据库构建：以湘赣边区红军标语为例[J].长沙大学学报，2025,39(4):32-38.HUANG J R,LIU Z Y,LIU R R,et al. An exploration on the cultural genes of red cultural heritage and the construction of its database:taking the red army slogans in Hunan-Jiangxi border areas as an example[J]. Journal of Changsha University,2025,39(4):32-38.(in Chinese)
• [10]华林，冯安仪，谭雨琦.关联数据环境下我国海洋历史文献资源知识组织研究[J].数字图书馆论坛，2023,19(8):15-25.HUA L,FENG A Y,TAN Y Q. Knowledge organization of marine historical literature resources in China under the context of linked data[J]. Digital Library Forum,2023,19(8):15-25.(in Chinese)
• [11]佟瑞娟.基于关联数据的图书馆馆藏数字资源语义化组织研究[J].河南图书馆学刊，2023,43(8):132-135.TONG R J. Research on semantic organization of library digital collections based on linked data[J]. The Library Journal of Henan,2023,43(8):132-135.(in Chinese)
• [12]周贞云，黄昭昭，邱均平.基于LLM与可视化的图数据库专利知识图谱构建及分析[J].情报杂志，2025,44(10):155-162.ZHOU Z Y,HUANG Z Z,QIU J P. Construction and analysis of patent knowledge graph in graph database field based on LLM and visualization[J]. Journal of Intelligence,2025,44(10):155-162.(in Chinese)
• [13]KOMMINENI V K,K??NIG-RIES B,SAMUEL S. From human experts to machines:an LLM supported approach to ontology and knowledge graph construction[EB/OL].(2024-03-13)[2025-07-18]. https：//arxiv. org/abs/2403. 08345.
• [14]赵嘉琪.古诗词知识图谱构建及在预训练模型中的应用研究[D].北京：北京邮电大学，2023.ZHAO J Q. Construction of knowledge graph of classical Chinese poetry and its application in pre-training model[D].Beijing:Beijing University of Posts and Telecommunications,2023.(in Chinese)
• [15]CHOWDHERY A,NARANG S,DEVLIN J,et al. PaLM:scaling language modeling with pathways[J]. Journal of Machine Learning Research,2023,24(1):11324-11436.
• [16]SCHAEFFER R,MIRANDA B,KOYEJO S. Are emergent abilities of large language models a mirage?[J]. Advances in Neural Information Processing Systems,2024,36:5565-5581.
• [17]LIU S Y,LI Y,LI J,et al. Unleashing the power of large language models in zero-shot relation extraction via selfprompting[EB/OL].(2024-12-21)[2025-07-18]. https：//doi.org/10. 48550/arXiv. 2410. 01154.
• [18]易晓婕，钟志峰，严诗玛，等.基于层权重融合与预测优化的中文短文本分类模型[J].湖北大学学报（自然科学版），2025,47(2):230-238.YI X J,ZHONG Z F,YAN S M,et al. A short Chinese text classification model based on layer weight fusion and prediction optimization[J]. Journal of Hubei University(Natural Science Edition),2025,47(2):230-238.(in Chinese)
• [19]周志华.机器学习[M].北京：清华大学出版社，2016.ZHOU Z H. Machine learning[M]. Beijing:Tsinghua University Press,2016.(in Chinese)