會議與期刊文獻對預測主題趨勢之比較研究—以「資訊檢索」領域為例

Simple item page
dc.contributor曾元顯zh_TW
dc.contributorYuen-Hsien Tsengen_US
dc.contributor.author許育聞zh_TW
dc.contributor.authorYu-Wen Shuen_US
dc.date.accessioned2019-08-28T08:10:06Z
dc.date.available2012-8-10
dc.date.available2019-08-28T08:10:06Z
dc.date.issued2009
dc.description.abstract多數進行主題趨勢探測的學者，皆採用期刊文獻作為主要的分析素材，鮮少針對其他類型的文獻進行研究，然而在學術傳播中會議論文的重要性不可小覷，因此本研究以資訊檢索（Information Retrieval）領域為範圍，針對會議文獻與期刊文獻分別進行主題趨勢預測，以觀察不同類型的文獻進行主題趨勢預測時的差異性。 本研究收集1990年至2007年資訊檢索領域具代表性的SIGIR會議文獻及五本核心期刊中收錄主題為「資訊檢索」的期刊文獻，五本核心期刊分別是：Information Processing& management、JASIST&JASIS、Journal of Information science、Journal of Documentation、Information Retrieval，主題歸類的部份是以主題整併和自動化歸類兩種方式進行。為了確保預測的準確性，本研究以相同文獻類型和相異文獻類型分別進行預測，以比較其預測上的準確性，最後分別改變預測集和驗證集之年代範圍以比較其差異性。 研究發現會議文獻和期刊文獻在主題詞彙的用法上有所差異，且各自有較關注探討的主題。會議文獻大部分比期刊文獻較早出現，然而在主題預測上，會議文獻並未佔有優勢，當預測的主題範圍較廣時，期刊文獻預測之效果較佳，且相同類型文獻預測效果優於交叉預測之效果。 最後提出之建議為：期刊文獻之控制詞彙尚未完善，許多單複數詞彙和縮寫詞彙尚需統整；主題預測的部份，若要瞭解較廣泛的領域趨勢，以期刊文獻預測的效果較佳，而要了解細部領域的趨勢則是以相同的文獻類型進行預測較佳；在後續研究的部份，可以針對像是專利或部落格等其他類型的灰色文獻進行研究，或是以文獻之作者群進行社會網絡分析也是一個可行的研究方向。zh_TW
dc.description.abstractMany scholars who study topic trends use journal articles as primary texts for analysis and hardly pay attention to other types of documents. However, the importance of conference papers cannot be neglected in the academic field of Scholarly Communication. Hence, the research focusing on Informal Retrieval puts topic trends into practice in two kinds of literature, conference papers and journal articles, and observes the discrepant results of those in different types of documents. The research collects representative researches on “Information Retrieval” in SIGIR conference papers and five core journals: Information Processing& Management, JASIST&JASIS, Journal of Information Science, Journal of Documentation, and Information Retrieval. The methods of categorizing documents rely on topics of journal articles given in databases, session titles of conference papers, and then the previous articles and papers in automatic categorization. In order to ensure the accuracy of prediction, and prediction is experimented in two groups, the same and the different types of literature. Then, the research changes periods of prediction and validation set to compare the results. The research finds that conference papers and journal articles differ not only in the uses of topic vocabulary but also in the topics of their concerns. Although most conference papers publish earlier than journal articles, the latter possesses more advantages in topic prediction. When the scope of the predicted topic is wider, the predicted results of journal articles are better. The predicted results of documents from the same type also generate superior outcomes than those from the different type. Suggestions are proposed in the end of the research. Control terms of publication papers are defective because plenty of singular/plural vocabulary and abbreviations need arranging. In the part of topic prediction, if understanding trends in wilder fields is needed, the prediction of topic trends in journal articles leads to better effects. The result of using journal articles to predict topic trends is better. To understand trends in detailed field, the prediction of topic trends in same type of documents is more effective. Finally, further studies on Information Retrieval is recommended to study other types of gray literature, such as patents or articles on blogs, or make an social network analysis on authors of documents.en_US
dc.description.sponsorship圖書資訊學研究所zh_TW
dc.identifierGN0695150105
dc.identifier.urihttp://etds.lib.ntnu.edu.tw/cgi-bin/gs32/gsweb.cgi?o=dstdcdr&s=id=%22GN0695150105%22.&%22.id.&
dc.identifier.urihttp://rportal.lib.ntnu.edu.tw:80/handle/20.500.12235/89240
dc.language中文
dc.subject主題趨勢預測zh_TW
dc.subject會議文獻zh_TW
dc.subject共字分析zh_TW
dc.subject自動化歸類zh_TW
dc.subject資訊檢索zh_TW
dc.subjecttopic trends predicten_US
dc.subjectconference paperen_US
dc.subjectco-word analysisen_US
dc.subjectautomatic categorizationen_US
dc.subjectinformation retrievalen_US
dc.title會議與期刊文獻對預測主題趨勢之比較研究—以「資訊檢索」領域為例zh_TW
dc.titleA comparison study on conference papers and journal articles for predicting topic trends – using「Information Retrieval」as an exampleen_US

Files

Original bundle
Now showing 1 - 1 of 1
No Thumbnail Available
Name:
n069515010501.pdf
Size:
2.27 MB
Format:
Adobe Portable Document Format
Download

Collections

學位論文