基于方面级情感分析的博物馆数字化服务用户体验研究

Exploring Factors Influencing User Experience of Museum Digital Services via Aspect-Level Sentiment Analysis

【目的/意义】分析用户对博物馆数字化服务的体验满意度及影响因素,为提升博物馆数字化、智慧化建设提供科学参考。【方法/过程】首先采集博物馆用户评论,构建数字化服务方面词词库,抽取方面级语句,然后使用BERT-wwm-ext模型(F1=95.47%,Acc=95.36%)对方面级语句进行情感分类,最后根据分类结果对用户体验影响因素进行分析。【结果/结论】将博物馆数字化服务用户体验划分为5个方面(预约、验证、导览、讲解、展示),分析发现用户对数字展示服务和网络预约服务满意度最高,对数字讲解服务满意度最低,并归纳了影响用户体验满意度的12项主要因素。【创新/局限】将方面级情感分析应用到了公共文化服务领域。提出了基于内容抽取和深度学习的方面级情感分析模型,通过迁移学习有效降低了模型对方面级标注数据的依赖性,同时对用户评论中的隐式情感也具有较好的识别效果。

【Purpose/significance】Analyze users’experience satisfaction and influencing factors of Museum digital service,so as to provide scientific reference for improving Museum digital construction.【Method/process】We crawled online reviews of historical museums,built an aspect-level lexicon of digital service,extracted aspect-level sentences,and use the BERT-wwm-ext model(F1=95.47%,ACC=95.36%)to classify aspect-level sentences.According to the classification results,the specific factors affecting users’experience are analyzed.【Result/conclusion】The museum’s digital service user experience is divided into 5 aspects(online booking,offline verification,tour guidance,interpretation,and exhibition).The analysis found that users have the highest satisfaction with exhibition services and online booking services,and the lowest satisfaction with digital interpretation services.12 main factors affecting user satisfaction are summarized.【Innovation/limitation】The aspect-level sentiment analysis is applied to the field of public cultural services.An aspect-level model based on content extraction and deep learning is proposed,which reduces the high dependence on aspect-level annotation data through transfer learning,and has a better recognition effect for implicit emotion expression in reviews.