Evaluating the Quality of the Indonesian Scientific Journal References using ParsCit, CERMINE and GROBID

Authors

Ariani Indrawati, Center for Scientific Data and Documentation, Indonesian Institute of Sciences, Jakarta, IndonesiaFollow
Ambar Yoganingrum, Research Center for Informatics, Indonesian Institute of Sciences, Cibinong, IndonesiaFollow
Pradipta Yuwono, Center for Scientific Data and Documentation, Indonesian Institute of Sciences, Jakarta, IndonesiaFollow

Date of this Version

9-10-2019

Abstract

There are several open-source tools available to extract the bibliographic references of the Pdf. Those tools based on the various approaches including rule-based approach, knowledge-based approach, machine learning-based approach, and the combination. To improve the services of the Indonesian Scientific Journal Database (ISJD), Center for Scientific Data and Documentation – Indonesian Institute of Sciences (PDDI-LIPI) intends to have an automatic bibliographic references extraction tool. The paper aims to analyze the quality of the reference metadata of the local journals with the three open-source tools, namely ParsCit, CERMINE and GROBID. The accuracy test of the three tools are poor. Those are 0.555, 0.633, and 0.605 for ParsCit, CERMINE, and GROBID respectively. It caused by many authors do not use a reference manager when they write the bibliography section. On the such condition this paper proposed to build an application to identify and correct errors in the bibliographic references of paper in ISJD. This application become a liaison between ISJD and open source tool for the bibliographic reference extraction. This paper proposed the combination of building software and using an open source.