토픽 모델링을 이용한 유튜브 설교 동영상 리뷰 분석

With the development of computer technology to process various large amounts of data, the method of analyzing text data has made great progress in recent topic modeling in machine learning and natural language processing. Topic modeling is a model for discovering abstract "topics" in a document and is one of the text mining techniques used to discover the hidden semantics of text bodies. This study analyzed the response to the sermon by topic modeling based on the reviews of the YouTube sermon video. The results were classified into five topics. Three of the five topics, topics 1, 2, and 4 can be viewed as key attributes of this sermon video, like online customer reviews, while the other two topics 3 and 5 contain reviews that are far from the sermon video. This is not a sermon video but a review of the preacher's previous words or deeds. Therefore, from the church's point of view, systematic and objective evaluation and management of reviews is required. There is a need for a rearrangement method of reviews in which reviews recorded by users are classified by topic to highlight positive reviews. We can also consider ways to hashtag using topics that include positive reviews to find ways to spread the number of visitors to sermon videos.