文本分析的新工具:以主题模型的可视化为例

A new tool for textual analysis based on the visualization of the topic model

主题模型是帮助我们理解文献集的重要手段。但LDA的原版算法需要用统计语言学的知识进行解读,不合适人文学者使用。借助py LDAvis、TOM等软件包进行可视化处理,为主题模型的结果解释提供了便利。从本质上看,任何可视化的方法都需要结合研究项目的具体需求,而不能为了绚丽的可视化结果而毫无目的地使用可视化的方法,否则就有本末倒置的嫌疑。

The topic model is an important tool to help us understand the corpus.However,the original algorithm of LDA requires statistical knowledge to interpret the results from humanities scholars.We used several Python packages that included pyLDAvis and TOM for visualization,which facilitated the interpretation of the results of the topic model.Essentially,any visualization method needs to be integrated with the specific needs of the research project,not to aimlessly use the visualization methods only for gorgeous visualization charts,or it may have the suspicion of having the order reversed.