A template approach for summarizing restaurant reviews

NCUIR > School of Management at National Central University > Graduate Institute of Information Management > Electronic Thesis & Dissertation > Item 987654321/83964

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/83964

Title:	A template approach for summarizing restaurant reviews
Authors:	甘哲宇;Gan, Jer-Yeu
Contributors:	資訊管理學系
Keywords:	摘要;餐廳評論;模板;情緒分析;TextRank;Summarization;Restaurant reviews;Template;Template;TextRank
Date:	2020-06-24
Issue Date:	2020-09-02 17:47:10 (UTC+8)
Publisher:	國立中央大學
Abstract:	在社群網路發展越來越快的時代下，使用者們在餐廳評論網站上的評論也逐漸增加，為了要讓使用者可以更快速的了解評論網站上的評論資訊，本論文會實作一種基於模板、主題和情感的餐廳評論摘要化的模板系統。本論文還使用了預定義主題的概念，將評論摘要依照主題來放，因此可以讓使用者一看就非常清楚及明瞭。在評估時會依照資訊豐富度(informativeness)、清晰度(clearness) 、有用性(helpfulness)與Refresh和Gensim的系統在比較，來讓使用者主觀選擇較好的模板。最後，我們發現我們的方法在資訊豐富度和有用性方面優於其他兩種方法。這個結果證明我們的方法可以提供更多的訊息，對用戶有更大的幫助。;In the era of rapid development of social networks, user reviews on restaurant review sites have increased rapidly. In order to enable users to more quickly grasp the focus of the review information on the review site, this article will implement a template method for summarizing restaurant reviews, which is based on templates, topics, and emotions. This article also uses the concept of pre-defined topics applicable to restaurants to summarize reviews so that users can understand the reviews more clearly and accurately. In the evaluation, we compared the template method with the Refresh and Gensim systems according to the criteria of informativeness, clarity and usefulness to evaluate which method can better satisfy the user′s subjective preferences. Finally, we found that our method is superior to the other two methods in terms of informativeness and usefulness. This result proves that our method can provide more information and is more helpful to users.
Appears in Collections:	[Graduate Institute of Information Management] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	119	View/Open

社群 sharing

Loading...