News Filter By Using Scraping Web with Crawling RegEx and TF-IDF Filter

Abstract

The most important information in Education is in the form of announcement, as in the results of a poll site in Blitar Education Office which reached 69.8% on 10 November 2016. In fact, the information can be derived from other Education Offices, provinces, or CSR. Based on this problem, the researchers propose two ways to get relevant news from various sites. First, create a virtual robot for crawling. Second, do filtering and arranging a rank in HTML documents crawling results. Crawling is operated by scrapping method, which uses Regular Expression. Regular Expression has a function to cut HTML screen based on a predetermined pattern. URL that has been obtained is used for crawling the content details through the location of the original details one. Documents then are stored and ranked using TF-IDF. Next, the rank results sent to dewiberita.com using a post method agent request. On the first experiment, it is resulted that 100% RegEx parsing through the entire of URL address running successfully. Next on the second experiment which is using Recall and Precision, it produce a similar level until 71.4% if the news filtered visualization as maximize as a half of the total document.