풍력에너지저널 문헌검토에 의한 주제모델링

초록

In celebration of the 10th anniversary of the publication of the Journal of Wind Energy, we intend to grasp research trends through a literature review of the journal’s papers and seek ways to improve the quality of the journal. The text mining technique was used to extract a document-term matrix, and topic modeling was performed using latent semantic analysis and fuzzy K-means clustering. From the comparison of the topic modeling results with manual categorization by experts, it is anticipated that supervised learning is recommended by including the specific topic classification by author in the bibliography metadata for meaningful topic modeling in the future. We confirmed that it is necessary to apply different weights because the descriptive level of title, keyword, and abstract are different when specifying the topic of the research paper. The characteristic theme of the Journal of Wind Energy was identified as “offshore wind,” and research institutes and universities are participating widely, but it is of concern because the participation of industry is declining.

키워드