<![CDATA[In the text mining there is a process for information retrieval. Problems related to information retrieval are found in universities, especially in the Faculty of Computer Science, University of Brawijaya (FILKOM UB). The problem is the selection of the thesis supervisor for the FILKOM UB Informatics Engineering S1 study program in the interest of Smart Computing is still done manually. Determination of supervisors only relies on personal knowledge related to the specialization of lecturers needed to guide during the execution of the thesis. These problems can be solved through a recommendation system based on information retrieval using the BM25 method. The process carried out is document preprocessing, calculation of BM25 score in each document, and taking the highest BM25 scoring result as much as k. In this study three tests were carried out. Each test uses the same testing data of 20 documents. The average results of each test obtained the best recommendation results, namely at the value k=3, with a value of precision @k of 0.87. The higher the value of k used can affect the recommendation results to be less optimal because more and more irrelevant documents are counted.]]>
Copyrights © 2019
