PENGARUH STEMMING TERHADAP EKSTRAKSI TOPIK MENGGUNAKAN METODE TF*IDF*DF PADA APLIKASI PDS

Abstract

Personal Digital Secretary (PDS) is a system that was developed to be a "personal secretary" who work alongside users digitally. PDS convey information to users in the form of email, social media and news. In order to know the information and news from the outside, it must be done by extracting user topics through email and social media, with the result that news information will have corresponding relationships with users. User topic extraction through email and social media in PDS is using modified weighting method in TF*IDF algorithm named TF*IDF*DF. In the further development, added stemming process in hopes of obtaining an appropriate topic. From the research that has been done, there are differences in terms obtained from the topic extraction without addition stemming process and with addition of stemming process. News information obtained by the addition of stemming process has more focused results than the news information obtained from the topics extraction without additional stemming process. With the addition of stemming process on the TF*IDF*DF algorithm indicates that the word (terms) results obtained from the extraction process has become the basic words because of stemming process. These Basic words are the basic form that an indication of a topic

Keywords: User topic, topic extraction, TF*IDF, topic model, fiture selection.