<![CDATA[Data shows very large numbers for Internet use in Indonesia. In the field of education, online libraries are an effort to facilitate researchers to search for references to research documents. Based on observations, UIN Jakarta already has a good repository of research documents, but the online research document repository does not fulfill the Knowledge Acquisition feature. This capability allows users to obtain knowledge information that is not easily accessible to users. This research build a machine learning system using the Support Vector Machine algorithm so that the system built can categorize documents based on the informatics research fields. This research also builds a system services API (Application Program Interface) so that data output from machine learning systems can be used by a variety of platforms and different operating system environments. The accuracy of the machine learning system in this study resulted in a percentage of classification accuracy of 73.2% with a parameter value of 0.9. At the preprocessing stage the selection of unigram-bigram is the best in this study. Preprocessing affects the level of classification of machine learning systems. Preprocessing using stemming improves the results of ability accuracy. The amount of data affects the accuracy of the machine learning classification ability, it can be seen when the data is increased to 488 accuracy increases to 74.49. When the experiment was done again so that the data increased to 492 data, the accuracy increased again to 77.78%.]]>
Copyrights © 2020
