基於詞語抽取的圖書與資訊學刊研究主題分析

Term Extracting-based Analysis of Research Subjects in the Bulletin of Library and Information Science

本論文持續《圖書與資訊學刊〉第42期「圖書與資訊學刊論文的高頻詞語抽取與分析」的研究，提出一系列基於自然語言處理與資訊檢索技術的研究主題分析方法。在這個方法中，首先應用關鍵詞語抽取技術，從中英文雙語的論文資料中，抽取出具有意義並且完整的詞語。接著利用隱含語義索引和叢集分析等技術進行主題辨識，將相關的詞語叢集成概念集合，辨識出期刊內重要的研究主題。最後則利用統計上的多維量尺法與電腦繪圖工具將主題繪製在二維的圖形上，便於觀察主題間的結構。我們對《圖書與資訊學刊》第16到45期的論文資料進行分析，結果發現本期刊的6個重要研究主題為（1）圖書館服務與館員訓練、（2）學校圖書館利用教育與資訊素養、（3）圖書資訊學教育與課程、（4）資訊檢索與使用者研究、（5）檔案學和（6）網路與電子技術。

In line with Lin (2003)'s study, this paper proposes a method to analyze research subjects using natural language processing and information retrieval techniques. Keywords are extracted from bilingual corpus, and latent semantic indexing and cluster analysis help identifying research subjects that are co-occurring with keywords. The multidimensional scaling technique is applied to determine thematic structures among research subjects. This study examines 209 articles in the (Bulletin of Library and Information Science) and finds the following six areas of research subjects: (1) library services and librarian training, (2) library user education and information literacy, (3) curriculum in library and information science, (4) studies on information retrieval and users, (5) archive studies, (6) internet and electronic technologies.