基於影響力分析之意見單元評價的研究

機構典藏 > College of Business and Management > Graduate Institute & Department of Information Management > Thesis > Item 987654321/102368

Please use this identifier to cite or link to this item: https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/102368

Title:	基於影響力分析之意見單元評價的研究
Other Titles:	A study of opinion unit evaluation based on influence analysis
Authors:	黃心宜;Huang, Shin-Yi
Contributors:	淡江大學資訊管理學系碩士班蕭瑞祥;Shaw, Ruey-Shiang
Keywords:	意見探勘;意見單元;影響力分析;Opinion Mining;Opinion unit;Influence analysis
Date:	2014
Issue Date:	2015-05-04 09:54:02 (UTC+8)
Abstract:	意見單元為評價對象與其對應的意見詞的組合，是計算評價產品資訊的基本單位，本研究提出了一種應用於網路論壇的產品評論文章，基於影響力分析的意見單元加權方法。本研究建立以意見領袖的概念找出有效的評論者特徵因素，進而計算文章作者的影響力，並將其與產品的意見單元整合彙總成產品評價之模式與雛型系統，並評估系統之準確性。系統評估文章來自於Mobile01網路論壇，關於智慧型手機的評論文章。藉由未經加權的意見單元與人工評分結果相比較，觀察到意見單元受到評論者因素的影響比例為46%，顯示評論者影響力可能會影響網路使用者觀看產品評價的感受；而基於評論者影響力的意見單元加權方法計算出的意見評價，能與人工閱讀評價文章後給予的意見單元分數有不錯的相似程度，與未加權的意見單元分數比較後，意見單元分析的準確度提升了15.37%，達到69.3%，證明本研究提出的方法使產品評價結果更加準確且能夠有效的衡量評論者的影響力。期望本研究之研究結果能做為後續網路文章評價相關研究與實務之參考。 Opinion unit is a combination of evaluation objects and corresponding opinion words and is a basic unit of calculate the product information. . This study proposes a product review articles used in Internet forum, use the method of weighting based on the influence analysis. This study established the concept of opinion leaders to identify effective reviewer characteristic factors, then calculate the influence of the article author. Integrating the influence of the article author with product review unit aggregated into product evaluation model and prototype system.It can evaluate the accuracy of the system. System Assessment article from Mobile01 Internet forum on "smart phones" products evaluated articles. Unweighted opinion unit compared with manual scoring results. Review unit being observed proportion of factors affecting reviewers was 46%. Show influential commentators might affect Internet users to view product evaluation experience. Calculated the evaluation unit based on the views of influential commentators with weighting method. It has a good similarity with the opinion unit scores that give by people reading the evaluation article. Comparison with the unweighted opinion unit scores, opinion analysis unit to enhance the accuracy of 15.37% to 69.3%. This study proposes a method to prove the product evaluation results more accurate and can effectively measure the reviewers influence. Expected the result of this study can do evaluation research and practice of reference for subsequent Internet articles.
Appears in Collections:	[Graduate Institute & Department of Information Management] Thesis

Files in This Item:

File	Size	Format
index.html	0Kb	HTML	227	View/Open

Loading...