We propose a method for the automatic classification of research papers in terms of the KAKEN classification index using a machine learning method. This classification index was originally devised to classify reports for the KAKEN research fund in Japan, and it is organized as a three-level hierarchy: Area, Discipline, and Research Field. Traditionally, researcher and conference names are used as cue phrases to classify the research paper efficiently. In addition to these cue phrases, we focus on elemental technologies and their effects, as discussed in each research paper. Examining the use of elemental technology terms used in each research paper and their effects is important for characterizing the research field to which a given research paper belongs. Therefore, we use elemental technology terms and their effects as additional cue phrases for machine-learning-based text classification. To investigate the effectiveness of our method, we conducted some experiments using the KAKEN and CiNii articles data. From the experimental results, we obtained average precision scores of 0.853, 0.712, and 0.615 for the Area, Discipline, and Research Field levels in the KAKEN classification index, respectively. These scores are higher than those for the method not using elemental technologies and their effects. From these results, we confirmed the effectiveness of using elemental technology terms and their effects as cue phrases.

In this research, the information retrieval (IR) skills in web environment are newly defined as professional skills and knowledge for traditional information retrieval in conjunction with digital skills described by van Deursen. To assess the status of IR skills in Japan, an online questionnaire-based survey was conducted in August 2014. Ultimately, 1,551 participants responded. The results showed the following trends among the participants: (1) not using advanced search technique such as Boolean operator, (2) understanding the format of information on web, (3) carefully choosing search terms, (4) evaluating multiple search results based on certain evaluation criteria, and (5) feeling that they receive benefits from the Internet. Moreover, based on the hierarchical clustering analysis of their information retrieval skills, the participants were categorized into eight clusters. One of the clusters comprised the participants with highly performing IR skills. The characteristics of this cluster members included relatively younger generation, males, highly educated, high critical thinking ability, and high self-efficiency. Although this cluster members achieved the highest average scores on each IR skill, their scores on search technique skills were significantly less developed than those on other skills.

Subject headings are assigned in library catalogs on the basis of a list of subject headings. The Basic Subject Headings (BSH) is the standard list of subject headings used for library cataloging in Japan. However, for elementary library catalogs, the Elementary School Subject Headings can also be used. Although the main MARC, created by private distributors, usually uses the BSH, if OPACs were equipped with a function that integrates the Elementary School Subject Headings and the BSH, elementary school students could choose subjects based on easy words while also having access to broader subject searches made possible by using the BSH. The present study examined methods of linking the Elementary School Subject Headings and the BSH by assigning NDC classification numbers and references, and focusing on the variability of terms and notation, which aim to integrate retrieval methods that combine the two lists of subject headings. The results of an experimental prototype system show that assigning NDC classification numbers and reference linking system worked effectively in most cases. However, the results also revealed a few cases with a dependency on the text-based search function.