Automatic Subject Classification of Korean Journals

Abstract

Subject classification of journals is important because it can be utilized for the improvement of scholarly information services and analysis by research area. The classification by experts in a subject area wastes a lot of time and expense. On the other hand, the simple classification with basic information, such as the journal title has limitations. To solve this problem, this paper suggests the automatic classification of Korean journals using the SCI journals information cited by Korean journals, and an analysis of the classification result. In particular, this study adopted the WoS subject categories for classification to support the base for comparison between the Korean citation database and the global citation database (KSCI vs. SCI).

1. INTRODUCTION

Subject classification of journals is necessary because it can be utilized for the improvement of scholarly information services and analysis of intellectual structure by research area. The classification by experts in a subject area wastes a lot of time and expenses. On the other hand, the simple classification with basic information, such as the journal title, has limitations. To solve this problem, this paper suggests the automatic classification of Korean journals using the Science Citation Index (SCI) journals information cited by Korean journals, and analysis the classification result. For this, this study adopts the Web of Science (WoS) subject categories for classification to support the base for comparison between the Korean citation database and the global citation database (KSCI vs. SCI). To show the result effectively, we maps WoS subject categories to 14 subjects by Leydedorff [5].

 

2. RELATED RESEARCH

Thomson Reuter’s (before ISI) WoS is the first citation index database for citation analysis and the most well-known in the world. Thomson Reuter’s WoS covers a variety of disciplines: Science, Social science and Arts & Humanities, and supports citation indices such as SCIE, SSCI and A&HCI (classified by subject respectively). Specifically, all of the journals (8,100) of WoS are classified by subject categories of WoS and this is one of reasons for much research related citation analysis. Dewey Decimal Classification (DDC) is used to classify journals widely in libraries [4]. Mahdi and Joorabchi suggested the method automatic subject classification of scientific literature using citation metadata based on DDC [6]. We adopted the subject categories of WoS for subject classification of Korean journals instead of DDC in this study.

 

3. AUTOMATIC SUBJECT CLASSIFICATION OF KOREAN JOURNALS

Fig. 1.KSCD, KSCI and KJCR

We utilized KSCD (Korea Science Citation Database), which supports the base for KSCI of KISTI (Korea Institute of Science and Technology Information). KSCI covers all of the important Korean Journals and supports the same level of service as WoS and JCR (Journal Citation Reports). KSCD covers the period from 2002 to 2013, and there are 760 journal titles, but it has changed with the criteria of journal selection annually. Figure 1 shows the relation among KSCD, KSCI, KJCR (Korea Journal citation Reports).

The previous studies for citation analysis of Korean researchers, using KSCD, revealed some significant findings, as described below. First, journals are the most cited document type among scholarly works (e.g books, research reports, thesis, proceeding etc.). Second, the citation rate of global journals to domestic journals is 8:2. Third, Korean researchers prefer journal papers in the global citation index service ,such as SCI or SCOPUS, over non-indexed journal papers for citing references [1]-[3]. Subject classification information for both SCI and SCOPUS journals can be obtained very easily from their own service sites.

The result of related research provided the motivation and background for this study. In particular, this paper utilized SCI journal, which is the most widely used in the world. WoS subject categories for SCI journals consist of 180 categories, and 1.6 subject categories are allocated on average per journal in JCR 2010.

Fig. 2.WoS Subject Categories

Fig. 3 shows the method and the process for automatic subject classification.

Fig. 3.Process of Automatic Subject Classification

The detail on Fig. 3 is as below.

KSCI’s 743 journals were classified by subject categories once they were submitted, as figure 3 shows, and 3 journals, which do not cite any SCI(E) journals were excepted. Verification for the automatic classification of subject categories was accomplished using 90 Korean SCI(E) journals. The matching percentage was 87.8% through comparing the answer (already classified subject categories by WoS) and the test (automatic subject classification). Specifically, the percentage when checking the answer among subjects (the average number of subjects categories of Korean SCI(E) journals is 1.37) was included in two test subjects. Through the verification test, we confirmed the precision of the research result. Table 1 shows the result about 20 journals among the classified KSCI 743 journals.

Table 1.Classified KSCI Journals (20 among 743)

 

4. COMPARISON OF SCIENTIFIC OUTPUT BY SUBJECTS (KSCI VS. SCIE)

The result of automatic subject classification for Korean journals using WoS subject categories make it easier to compare the status of Korean scientific output and global scientific output(KSCI vs. SCI(E)). Fig. 2 shows the differences of the analysis result of the structures of scientific output (KSCI 718 journals, SCI(E) 7,995 journals in 2010 publication) by 14 subjects, as Leydesdorff suggested, and we mapped the WoS subject categories of the 14 subjects [5].The considerable differences of subjects between KSCI and SCI can be found in below figure 4. First, Biomedical Sciences shares the highest portion with 19.1% in SCI(E) publication. On the other hand, Computer Sciences shares the largest portion with 17.9% in KSCI publication. Second, the disciplines such as Engineering, Environmental Sciences and Agriculture share much higher portions comparatively in KSCI than SCI(E). These meaningful differences could support discussion materials or form an important basis for study concern about scholarly activities. All equations must be typed or written neatly in black. They should be numbered consecutively throughout the text. Equation numbers should be enclosed in parentheses and flushed right. Equations should be referred to as Eq. (X) in the text, where X is the equation number. In multiple-line equations, the number should be given on the last line.

Fig. 4.Comparison of Scientific Output between KSCI and SCI(E)

 

5. CONCLUSION AND FUTURE WORK

This paper suggested the method of automatic subject classification for Korean scientific journals using KSCD. Korean researchers cite global journals, such as SCI, much more than domestic journals. The assumption that the relationship between cited journals would be strong provided the motivation and background to this research.

In particular, the idea for automatic subject classification is to use SCI(E) journals with WoS subject categories cited by Korean journals. The result of using this suggested method showed 87% matching rate through the verification of 90 Korean SCI(E) journals. Also, the considerable differences of subjects between KSCI and SCI(E) were revealed through comparison between KSCD vs. SCI(E) databases, on which the result of automatic subject classification were based.

This paper has limitations due to the utilization of the top 3 cited SCI(E) journals by each Korean journal. In further research, we plan to use all of the cited SCI(E) journals by each Korean journal to improve the precision of the result of subject classification and analyze the specific differences by subsubjects between KSCI vs. SCI(E) in depth.