Open AccessComparison of classification model and annotation method for Undiksha’s official documents
Author(s) -
A.A. Gede Yudhi Paramartha,
Núria Martí,
Kadek Yota Ernanda Aryanto
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1516/1/012026
Subject(s) - computer science , annotation , levenshtein distance , information retrieval , categorization , upload , naive bayes classifier , tf–idf , similarity (geometry) , decision tree , artificial intelligence , natural language processing , world wide web , support vector machine , image (mathematics) , term (time) , physics , quantum mechanics
Shakuntala is a system that manages official documents and letters at UniversitasPendidikanGanesha. The system stores various documents in PDF format which are categorized by type of document. But Shakuntala can only receive scanned documents, and document categorization were done manually by the operator. Documents uploaded to Shakuntalaalso generally contain information about people who were manually tagged by the operator. This causes inefficiencies that should be carried out automatically by machine. This study aimed at finding the best classification model for determining document categories. In addition, this research also intent to figure out the best method for tagging the people listed on the document. The results of the study showed that the Decision Tree classification model was the best model with an accuracy of 83.06% compared to KNN and Naive Bayes. As for the annotation of the person’s name, the Levenshtein distance method with a similarity threshold of 95% obtained an accuracy of 68.20%.