This study aims to determine how much the accuracy of the use of Fuzzy Logic Algorithm in the Final Task classification. The method used is Fuzzy K-Nearest Neighbor (KNN). The first step in the process is to do Text Preprocessing to find out what terms will be carried out for the next process, then weighting each term using the TF.IDF algorithm, after weighting, the terms of testing with the training term using Cosine Similarity are calculated, the next step the similarity value will be sorted and then some data will be taken from a number of neighbor values determined to calculate the KNN and see which categories appear more, the last is calculated the accuracy value seen from the comparison of training data and testing data. The final result of accuracy is 76.67%, thus Fuzzy KNN can be used to help the process of grouping research groups.

Keywords - Cosine Similarity ; KNN ; TF.IDF