基于ChatGPT和ROST方法的敦煌旅游地形象感知分析与对比  

Analysis of the image perception of Dunhuang as tourist destination:A comparison between ChatGPT and ROST method

作　　者：代浩宇 马国强[1] 汪胜兰[1,2] 黄银洲 DAI Haoyu;MA Guoqiang;WANG Shenglan;HUANG Yinzhou(School of Earth and Environment,Lanzhou University,Lanzhou 730000;School of Resources,Environment and Tourism,Hubei University of Arts and Science,Xiangyang 441053,China)

机构地区：[1]兰州大学资源环境学院,兰州730000 [2]湖北文理学院资源环境与旅游学院,襄阳441053

出　　处：《干旱区资源与环境》2024年第12期151-159,共9页Journal of Arid Land Resources and Environment

基　　金：国家自然科学基金面上项目(42077425)资助。

摘　　要：ChatGPT作为大语言模型的代表,具备出色的文本分析能力。文中以敦煌为例,基于从携程网和去哪儿网收集的6750条鸣沙山月牙泉和莫高窟的游客评论,使用“认知-情感-整体“模型,分别通过ChatGPT和ROST软件进行分析,对比两者在旅游地形象感知上的差异,并评估ChatGPT的精细化和深入分析能力及其应用潜力。研究发现:1)ChatGPT比ROST能够捕捉更多元的旅游要素和更细致的情感反馈,ROST则只展现常规的认知形象和总体的情感趋势。2)这些差异源于技术基础的不同,ChatGPT依靠深度学习和大规模多样化的文本预训练,能够处理复杂的情感逻辑和多元要素,而ROST受限于固定词库和共词分析,难以处理复杂信息。3)ChatGPT一类的大语言模型在处理复杂文本和海量数据时展现出强大的工具价值,具有广泛的应用前景,在旅游领域展现出较高的实践价值。As a representative of large language models,ChatGPT demonstrates excellent text analysis capabilities.Taking Dunhuang as an example of tourist destination,reviews of the 6,750 tourists visited Mingsha Mountain Crescent Spring and the Mogao Grottoes collected from Ctrip and Qunar were analyzed using the "Cognition-Affect-Overall" model,and analysis was conducted through both ChatGPT and ROST software,their differences in tourism destination image perception were compared,and ChatGPT' s refined and in-depth analysis capabilities as well as its potential applications were evaluated.The study found that:1)ChatGPT is able to capture more diverse tourism elements and more detailed emotional feedback than ROST,which only reflects conventional cultural and natural landscape images.2)These differences stem from the different technical foundations of the two methods.ChatGPT relies on deep learning and large-scale pre-training on diverse text data,allowing it to process complex emotional logic and multi-dimensional elements,while ROST is constrained by fixed lexicons and co-occurrence analysis,making it difficult to handle complex information.3)Large language models such as ChatGPT demonstrate powerful value in handling complex texts and large datasets,with broad application prospects and significant practical value in the tourism sector.

关 键 词：ChatGPT ROST软件 文本分析 对比研究 旅游目的地形象感知 

分 类 号：F590[经济管理—旅游管理]