基於草稿論文上下文及結構關係的論文引文推薦

NCUIR > School of Management at National Central University > Graduate Institute of Information Management > Electronic Thesis & Dissertation > Item 987654321/72229

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/72229

Title:	基於草稿論文上下文及結構關係的論文引文推薦
Authors:	施多加;Shih,Duo-Jia
Contributors:	資訊管理學系
Keywords:	論文引文;文字探勘;推薦;草稿;Draft paper citation;Text mining;Citation recommendation
Date:	2016-08-29
Issue Date:	2016-10-13 14:33:45 (UTC+8)
Publisher:	國立中央大學
Abstract:	大量的研究結果造成研究者在撰寫論文時的困難，搜尋適當的引用文獻成為極耗費人力的工作。但目前的研究大多是從關鍵字、文章的角度進行推薦，無法針對文章中的某個段落文字推薦。就算有以段落文字為基礎的方法，也沒有考慮到文字可能有不存在，或者部分殘缺的問題我們的研究，希望能幫助作者在撰寫論文時，得到合適的引用資料，除了考慮本文之外，還加入標題、摘要、關鍵字等論文結構資訊；並利用Wordnet改良傳統TF-IDF遇到文字量不足時資料疏散的問題；最後加入完整度門檻來調整不同權重對推薦結果的影響，至終從作者的角度給予合適的推薦文章。;The great amount of research results brings researchers recurrent difficulties. Searching for the appropriate citations become an extremely labor-intensive work. However the current study about paper citation recommendation almost recommend from the perspective of keywords or the articles. These methods can not recommend from the perspective of contexts. Even if there are some context-based methods, they didn’t take into account the information completeness of these contexts. In this paper, we hope to help the authors who is writing papers during their research, getting the appropriate citation. In addition to considering the context of articles, but also the structure information of research papers like title, abstract, keywords and so on. We use Wordnet to improve the text information inadequate problem of traditional TF-IDF method. Finally, we take completeness threshold into consideration to adjust the problems of information incompleteness. Using this algorithm, we can recommend the appropriate citation in terms of authors eventually
Appears in Collections:	[Graduate Institute of Information Management] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	507	View/Open

社群 sharing

Loading...