社群媒體協同過濾式英語學習影片推薦系統設計與實作

THE DESIGNING AND IMPLEMENTATION OF SOCIAL MEDIA COLLABORATIVE FILTERING ENGLISH LEARNING VIDEO RECOMMENDATION SYSTEM

葉芯妤(Hsin-Yu Yeh) ； 蔡妤姿(Yu-Tzu Tsai) ； 徐霈倫(Pei-Lun Hsu) ； 張智凱(Chih-Kai Chang)

《國立臺灣科技大學人文社會學報》 13卷2期 (2017/06) Pp. 153-168

摘要

網路學習英語已愈來愈普遍，其中透過網路影片學習英語也隨之蓬勃發展，例如影音平台Voice Tube 為擁有多元免費影片資源的英語影片學習網站，同時能結合社群網路應用。然而為了從豐富的英語影片資源中做選擇，挑選出符合個人喜好的觀看影片，一個良好的推薦系統機制至為重要。因此，本研究希望藉由協同過濾式（collaborative filtering） 的資料過濾方法，從Python網頁爬蟲抓取的使用者資訊中計算分析，找出相似偏好的使用者族群進而得出評分值，並結合Crab 推薦系統篩選出適合的英語學習影片，期望藉此幫助使用者在觀看英文影片的同時能更精確地推薦適合的影片，進而提升影片學習英語的動機。此外，針對使用者學習策略分為「精讀」和「廣讀」模式欲加以探討，系統評估方面使用共同影片收藏人數分佈模擬與叢集分析（cluster analysis），觀察Voice Tube 使用者的影片收藏與學習行為特徵，希望藉此發現合適的使用者族群可作為影片篩選的依據，提高系統的準確性，進而提升影片學習英語的動機。

關鍵字

語言學習 ； 社群媒體 ； Crab 推薦系統 ； 協同過濾式 ； 網頁爬蟲

並列摘要

Learning English through the internet is getting popular. Under this trend, online English learning videos have flourished. For example, video platforms such as VoiceTube is an English learning website that offers a variety of free video resources and can be utilized with social networking sites at the same time. However, in order to select videos that one prefers viewing from the abundant English video resources available, it is crucial to have an outstanding recommendation system mechanism. Therefore, this study utilized the information filtering method of collaborative filtering to calculate and analyze user information obtained by web crawling Python websites to find user groups with similar preferences and further score them. With the combination of the Crab recommendation system, it further filtered appropriate English learning videos with the expectation of accurately recommending users to English videos that are appropriate to them while they are watching them, thus enhancing their motivation in learning English with videos. In addition, this study divided the learning strategy of the users into ＂scanning＂ and ＂skimming＂ in order to further discuss the similarity and difference between these two strategies. As for system evaluation, the simulation and cluster analysis on the population distribution of users who collect the same videos was conducted to observe the video collection and learning behavior features of VoiceTube users. Through the aforementioned procedures, we expect to find the video filtering indicators for suitable user groups, improve systematic accuracy, and further enhance the motivation for learning English through videos.

並列關鍵字

language learning, social media ； Crab recommendation system ； collaborative filtering ； web crawler

參考文獻

李觀、林金城(2009)。停止惡意掃描的陷阱系統(碩士論文)。大同大學資訊工程學系。

周德懋、李舟軍(2009)。高性能網路爬蟲：研究綜述。計算機科學。36(8)，26-29。

彭賡、范明鈺(2010)。基於改進網路爬蟲技術的SQL注入漏洞檢測。計算機應用研究。27(7)

楊亨利、黃仁智(2008)。具整體觀點考量之推薦系統：以家庭親子為例。中華管理評論國際學報。11(3)，1-26。

廖崇勛、黃承龍()。