A document recommendation system of stemming and stopword removal impact: A web-based application

Abstract

Stemming and stopword removal is process that requires a lot of resources in the text pre-processing. The resources used in stemming and stopword removal are directly proportional to the amount of stopword, text, and document. Elimination of stemming and stopwords is one of many options which can reduce process in document-based recommendation system. However, the elimination of stemming and stopword removal have an impact on the recommendation system accuracy. This study determines the impact of stemming and stopword removal in document-based recommendation system. The system used in this study is recommendation system that recommend final project supervisor based on similarity student preliminary research proposal (UPP) document and lecturers’ scientific publications. Student preliminary proposal and lecturers’ scientific publications are document in Bahasa Indonesia. This study begins with analysis to map out the components to be used in each recommendation systems testing. Then proceed with the rearrangement of recommended system components based on testing focus. The study result is precision, recall, and f-mesure values comparison between each recommendation system component elimination. Elimination of stemming and stopwords show that the elimination generates precision and f-measure value worse than the system with stemming and stopword removal. However, system with elimination give a better result at recall value. In the future study, recommendation system needs some development to improve the precision, recall, and f-mesure value with modification in stemming method using Sarawi project and increase the amount of lecturer’s publication.