運用數位人文工具探討檔案時事議論主題及其情感分類之實作

Using Digital Humanities Tools to Explore the Implementation of Topic and Sentiment Analysis in Archival Affairs Commentary

本文以中國大陸網路時事論壇中討論與檔案相關內容為例，分析2019－2020 年有關檔案貼文呈現的主題概念及其內容情感傾向，運用NLPIR 大數據語義智能分析平臺進行斷詞、詞性標註及詞頻統計等文本預處理，進而採用中文情感詞彙本體庫進行文本情感分析。研究結果顯示，民眾最關注的檔案議題歸為兩類，一是特定人物與事件議題；二是以「檔案工作實務」和「檔案公開」為代表的社會議題。在情感方面，檔案時事論壇在針對特定人物與事件的討論時，呈現負面情緒較為明顯；而在談及檔案工作實務時，則展現出對於檔案開放應用的正面期待。

This paper collected forum discussions related to archives in mainland China and analyzed the topics and sentiment trends presented in archives posts in 2019- 2020. This paper used the NLPIR platform to perform text preprocessing such as word segmentation, part-of-speech tagging, and word frequency statistics. Chinese sentiment vocabulary ontology is then introduced to precede the sentiment analysis. The results showed that the topics that people are most concerned about are generally divided into two types. One is the specific figures and events, the other is the social issues represented by“archival work practice”and“archives access”. In terms of sentiment analysis, the archival affairs forum showed obvious negative emotions when discussing specific people and events; and when it came to the practice of archival work, it showed positive expectations for the archival access.