Peringkasan Teks Otomatis Berita Menggunakan Metode Maximum Marginal Relevance

Abstract

Development of Internet technology affects the increasing number of Indonesian language news website and creates an explosion of information. It requires all the information that can be accessed quickly and does’n require a lot of time in reading a news headline. Automatic text summary technology offers a solution to help search us news content in the form of a brief description (summary). The study begins with a five-stage preprocessing text: solving sentence, case folding, tokenizing, filtering, and stemming. The next process is computer tf-idf weighting, weighting query relevance and similarity weights. Summary results from the extraction using the maximum sentence of marginal relevance. Marginal relevance maximum extraction method is the method used to reduce redundancy in multi ranking sentence on the document.Keywords : Summary, text preprocessing, tf-idf, query relevance, similarity, maximum marginal relevance