Institusion
Institut Teknologi Perusahaan Listrik Negara
Author
NOVITA, SHINTIA
Agtriadi, Herman Bedi
Palupiningsih, Priatasari
Subject
Teknik Informatika
Datestamp
2023-05-31 07:13:25
Abstract :
In this era, the development of technology is developing so rapidly. One of the impacts of
this progress is progress in the field of archiving student final assignment documentation
at tertiary institutions. Various kinds of student final assignment documentation such as
theses, final reports, journals and so on have been stored in digital form via the
university's repository website. However, with these developments, it was not
accompanied by the provision of information. Because of this, it can cause data buildup
in the memory where the document or data is stored. One solution to overcome this
problem is to apply the Text Mining method as its preprocessing. Therefore it was chosen
to use the K-NN algorithm for student documentation such as theses, final reports,
journals, etc. which are stored in digital form through the archive website repository of
the PLN Institute of Technology. The test results for cosine similarity that were carried
out produced an average precision of 0.86%, 1 thesis title was used to calculate the
precision value. For the classification of thesis titles, 52 comparative data were used and
from the classified thesis titles only 43 book data were classified according to the system.
For testing the software system, it can get a score of 88% in the very good category. From
the results of the exact system test made with the K-nearest neighbor algorithm, the value
of k = 5 is used in the model that corresponds to 75%. There is bad data that affects the
classification results and affects the accuracy of the K.-Nearest Neighbor algorithm. The
results of the K-nearest neighbor accuracy test on the cosine matrix reach 60% which is
quite good.