Abstract: 【Purpose/significance】Analyze users' experience satisfaction and influencing factors of Museum digital service,so as to pro⁃vide scientific reference for improving Museum digital construction.【Method/process】We crawled online reviews of historical muse⁃ums,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 verifi⁃cation,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 an⁃notation data through transfer learning,and has a better recognition effect for implicit emotion expression in reviews.