According to research on Koean news usage in 2016, Internet news users account for about 70 percent
and 76 percent people of them read news articles on portal sites. Also, the news service rate via mobile
Internet recorded 70.6 percent. Even though there are various ways to view the news, users are provided
the information which is only headline, press and first one or two sentences of an article. For this, this
paper choose keywords to provide to users. Because it is judged that it is convenient for users to provide
search terms in the search ranking in the recent search portal. And this paper extracts the issue-keywords
and keywords of articles using by cosine similarity and TF-IDF methods of Text-mining. The subjects of
this study are articles ranked first in each categories(political, social and economic pages) from portal site.
and related articles with high cosine similarity to them. And this research is able to be used as a study to extract sentences the relevant issues.