| Summary: | 碩士 === 國立臺北大學 === 資訊管理研究所 === 107 === Researchers need to refer to many pieces of relevant literature in the process of setting research directions. However, when there is a large amount of digitalized online data, researchers often face the dilemma of not knowing where to start. To mitigate this issue, we propose a method that may help researcher select suitable topics from voluminous relevant literatures.
In past researches, scholars proposed to analyze the relationship between research topics and related fields in the form of supervised learning, and utilized regression analysis to estimate literature production of research topics in coming years. However, these studies have their limitations. Therefore, we propose a method using topic modeling to analyze research topics trends to overcome research constraints.
We utilize the scholarly data provided by Semantic Scholar online research database. The dataset includes the bibliographic data and abstract of forty millions of academic articles. The top 10 research topics along with their associated keywords are extracted from the articles’ abstract in the dataset. The number of papers in each research topics in four periods ranging from 2001 to 2018 affords us the capability of trending the popularity of these research topics. The resulted research topics and their trend were partially validated by relevant review studies and research trend exhibited by an online scholarly database.
|
|---|
