應用自動化文本分類及電子書推薦提升點擊率;Applying Automatic Text Classification and E-book Recommendation to Improve Click Rate

NCUIR > College of Electrical Engineering & Computer Science > Graduate Institute of Computer Science and Information Engineering > Electronic Thesis & Dissertation > Item 987654321/77534

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/77534

Title:	應用自動化文本分類及電子書推薦提升點擊率;Applying Automatic Text Classification and E-book Recommendation to Improve Click Rate
Authors:	楊智凱;Yang, Zhi-kai
Contributors:	資訊工程學系
Keywords:	文本分類;自動化文本分類;開放教材資源;電子書;機器學習;Text classification;Automatic text classification;Open Educational Resources;E-book;Machine learning
Date:	2018-07-16
Issue Date:	2018-08-31 14:47:25 (UTC+8)
Publisher:	國立中央大學
Abstract:	近年來大量的開放式教育資源已經逐漸融入於學生各階段的學習過程中，開放式教育資源不但能幫助學生實現自我學習，也能減少教師的課前的備課時間，讓教師專注於解決學生在學習過程所遇到的學習困難。然而，隨著開放式教育資源的數量不斷增加，如何提升各類型教材的被使用程度來幫助學生精準獲得所需要的學習教材，成為開放式教育資源需要解決的問題。有鑑於此，本研究以教育部教育大市集為平台，應用機器學習方法及文本分類技術來提升各類型教材的被使用程度，進而幫助學生精準獲得所需要的學習教材。本研究將會透過比較多種不同的分類模型，從中選擇相對適合此資料集的分類模型並透過LDA進行特徵萃取找出最佳特徵集合，最後使用SGD和投票機制來做模型的修正和決定分類模型。實驗的環境本研究使用Spark進行分散式處理，並利用Cassandra資料庫系統儲存前處理過的資料，而利用隨機森林、支持向量機、邏輯迴歸、類神經網路…等分類器分類好的教材及推薦清單會儲存在MySQL關聯式資料庫系統，最後透過PHP以及JavaScript網頁技術進行使用者介面的推薦。;In recent years, a large number of open educational resources have been gradually blend in the learning process of students at all stages. Open educational resources not only can help students achieve self-learning but also let teachers to reduce their lesson preparative time and focus solving the learning problems of students. However, with the number of open educational resources increasing in these year, how to help students precisely get the required learning materials has become a problem that open educational resources need to solve. In view of this, the study takes the Education Market as a platform. Apply machine learning methods and text classification techniques to improve the usage amount of various types teaching materials, and then helps students precisely get the required learning materials. The study will compare the various classification models to select the appropriate classification model for our data set. After that ,The study will use LDA feature extraction to find the best feature component. In addtion to LDA,the study use SGD and voting mechanisms to make model optimize and integrated the classification model. This study use Spark experimental environment to do distributed processing and use Cassandra database system to store pre-processed data. In our study, we apply multiple different classifiers, such as random forests, support vector machines, logistic regression, neural networks to classify teaching materials and establish recommendation list and the result will be stored in MySQL associative database system. Finally, we can through PHP and JavaScript web technologies to do recommend on everyone user interface.
Appears in Collections:	[Graduate Institute of Computer Science and Information Engineering] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	91	View/Open

社群 sharing

Loading...