Using queried keywords or full-text extracted keywords in blog mining?

In recent years, increasingly worthy content and reliable information have been disseminated on blogs. This means that readers are accustomed to searching information from blogs. However, the number of blogs has increased tremendously, so it is difficult to identify related information in the enormous number of existing articles. In general, the keyword is used as the basic method for readers to filter unimportant information via search engines. Keywords can be seen as reflecting the essence of an article. That is, a full-length article can be referenced through several keywords. As a result, if the keywords chosen for an article are accurate, the users' intentions will be satisfied. Therefore, the methods for selecting accurate keywords are critical. One general approach generates keywords by applying full-text keywords retrieval process, but it is a time-consuming process for handling the enormous number of articles. In this paper, to save time in generating keywords via full-text keywords retrieval process, a system called Blog Connect is proposed to embed the tracing code for selecting queried keywords as keyword candidates. Experiments provided positive data to confirm the effectiveness of the proposed method.