| Summary: | 碩士 === 東吳大學 === 資訊管理學系 === 106 === Abstract
Text mining is an application that can tap or explore important information in textual data. With the rapid development of the Internet, the amount of information also becomes larger and larger. Large amounts of text make reading a time-consuming and strenuous task. The content often contains many trivial, unimportant sentences. Thus, increasingly importance has been attached to text mining. Most of the related text exploration papers only analyze various data by a single model. The use of thematic model analysis can only display the key words covered by each topic. Moreover, no research uses the visual relevance of the links between concepts to do the text of the exploration for the theme of the text do. Therefore, this research takes travel magazine, Lonely Planet, based in North America as the paper data, adopts the topic model to present the structure of the theme from the data, utilizes the keywords of all the topics as attributes, and then extracts words, belonging to each topics, from each article as an object. Then, we establish the association between objects and properties as the database, and import the information into the concept explore program to show the correlation diagram. The aim of this study is to use the model to find the theme of articles, and to make a clear conceptual framework. This research combines topic model with the concept of regularization, so that we can quickly understand the article structure and the potential information of different articles. The experimental results show that the topic model can effectively classify the theme of the text. By formal concept analysis, the relation between each concept can be precisely linked to help readers quickly search the information they want to know.
Keywords: text mining, topic model, formal concept analysis, text architecture.
|
|---|
