主题：Textual analysis with python inaccounting research

报告人：Liu Zheng（刘正） 博士（Smith School of business, Queen’s University）

时间：2019年9月19日（星期四）15：00-17：00

内容摘要：The process of Textual analysis is derivinghigh-quality information from text. It is a setof linguistic, statistical, and machine learning techniquesto model and structure the information content of textual sourcesfor business intelligence, exploratory data analysis, research,or investigation. Converting text into numbers, and then using regularanalytics techniques(e.g., python). Main Practice of textual analysis inaccounting research including readability, targeted phrases, sentimentanalysis, topic modeling and document similarity. Readability is whether thereceiver of information can accurately reconstruct the intended message.Targeted phrases and Word list is focusing on a few unambiguous words orphrases. The typical way we do sentiment analysis in accounting research is tocompile word lists that share common sentiments (e.g., positive, negative,uncertain). In measuring the tone or sentiment of a financial document,researchers typically count the number of words associated with a particularsentiment word list scaled by the total number of words in the document. Topicmodeling is finding the main topics or themes in a set of documents. Thepurpose of document similarity is calculating the similarity between thefollowing two sentences.