The development of digital technology has a positive impact on information media such as news. Online news sites have become one of the most trusted source of information for conducting news searches. Finding news information online enables users to search directly in the search column of a news site. Online news has a structure or fields such as title, content, and additional tags. However, this will cause problems if many users searches are following the structure other than the title and content of the news which should have a more appropriate portion. Indonesian news information retrieval system is needed to solve this problem. Using methods that can solve the problem of structured documents. BM25F method is a ranking method that can consider the structure or field in a document. Tests carried out on BM25F free parameters get the best results at boost title = 5, content = 1, bc = 0,75, and k1 = 1,2. And the BM25F ranking test using precision @ k and r-precision in 600 news documents for 12 queries obtained the best average precision @ k = 0,95 when k = 5 and the best r-precision value = 1.
Copyrights © 2020