초록

Abstract

The objective of this research was to quantitatively examine the uses of first person pronouns in academic journal paper abstracts. An approximate total of 144,400 abstracts that comprising of four disciplines (chemistry, computer sciences, social sciences, and medicine) from nine countries (China, Germany, India, Japan, South Korea, France, Spain, United Kingdom, and U.S.) were quantitatively examined. By exploring the use of first person pronoun in abstracts, this paper examined the current practices among academics in the world. The results indicate the norms of each author’s country and the norms of each discipline. Furthermore, the frequency-count result of this study contradicted viewpoints of academics who disapprove the use of personal person expressions in abstracts. An implication of this study is that there is a need for academics to acknowledge the uses of first person pronoun in the real world before forming personal opinions regarding the first person pronoun.

초록

본 연구에서는 R2RML 방식과 Non-R2RML 방식을 각각 적용하여 RDB를 RDF 온톨로지로 변환한 결과를 비교하였다. RDB 기반의 구조적학술용어사전 데이터베이스인 STNet의 데이터를 대상으로, 변환이 완료된 데이터의 규모, 튜플당 변환에 걸리는 시간, 그리고 질의 응답 속도를 측정하였다. 변환 규모의 평가 결과 Non-R2RML 방식이 더 많은 수의 변환을 하였으며, 표현의 풍부성과 추론 가능성 정도를 높이는 변환을 수행한 것으로 나타났다. 튜플당 변환 시간의 경우 Non-R2RML 방식이 미세하지만 더 빠른 것으로 나타났으며, 질의 응답 속도는 두 방식 모두 300회 이상의 질의 횟수부터는 안정적인 성능을 보이면서 유사한 형태의 속도를 보였다. 측정에 대한 종합적 검토 결과 데이터의 변형이 빈번하고 새로운 데이터의 추가나 데이터들 간의 연결관계가 지속적으로 변화하는 STNet과 같은 동적인 RDB에는 Non-R2RML 방식이 적절한 것으로 평가되었다.

Abstract

This study described the results of converting RDB to RDF ontology by each of R2RML method and Non-R2RML method. This study measured the size of the converted data, the conversion time per each tuple, and the response speed to queries. The STNet, a structured terminology dictionary based on RDB, was served as a test bed for converting to RDF ontology. As a result of the converted data size, Non-R2RML method appeared to be superior to R2RML method on the number of converted triples, including its expressive diversity. For the conversion time per each tuple, Non-R2RML was a little bit more faster than R2RML, but, for the response speed to queries, both methods showed similar response speed and stable performance since more than 300 numbers of queries. On comprehensive examination it is evaluated that Non-R2RML is the more appropriate to convert the dynamic RDB system, such as the STNet in which new data are steadily accumulated, data transformation very often occurred, and relationships between data continuously changed.