運用機器學習進行贗品網站分類

THUIR > College of Engineering > Department of Computer Science and Information > Master's Theses > Item 310901/31832

Please use this identifier to cite or link to this item: http://140.128.103.80:8080/handle/310901/31832

Title:	運用機器學習進行贗品網站分類
Other Titles:	Counterfeit Website Classification By Using Machine Learning
Authors:	陳柏翔 CHEN,PO-HSIANG
Contributors:	蔡清欉 TSAI,CHING-TSUNG 資訊工程學系
Keywords:	假網站;贗品網站;機器學習;隨機森林;深度神經網路 Fake Website;Counterfeit Website;Machine Learning;Random Forest;Deep Neural Networks
Date:	2019
Issue Date:	2019-12-16T06:50:36Z (UTC)
Abstract:	近年來社群網站盛行，許多人開始在網路社團上進行買賣交易。所以開始出現大量的贗品網站，使得網路交易招詐騙的風險也逐漸提高。機器學習近年來被廣泛運用在各個領域，為了解決上述問題本研究將提出機器學習對於贗品網站進行分類，藉由訓練合格的網站及贗品網站資料，從中學習到相關特徵，最後進行合法的網站及贗品網站的分類。本研究的資料集為合法的網站及贗品網站各400個，藉由17個特徵進行訓練，例如在網站上看到的貨幣數量、是否在中國註冊、網站是否在一年內註冊等等。我們也會採用統計的方式，如卡方檢測等等對特徵進行評估，之後，我們使用兩種機器學習演算法進行模型訓練：Random Forest、Deep Neural Networks，並將其以8:2的比例分作訓練集與測試集並將結果作比較，最後，進行預測與先前研究結果比較分析。 In recent years, social networking sites have become popular, and many people have started trading transactions on the Internet community. Therefore, a large number of counterfeit websites have begun to appear, which has gradually increased the risk of fraudulent online transactions.In recent years, machine learning has been widely used in various fields. In order to solve the above problems, this study will propose machine learning to classify fake websites, learn the relevant features by training qualified websites and product websites, and finally carry out legal websites and The classification of the product website.The data set of this study is 400 websites of legal websites and counterfeit websites. It is trained by 17 characteristics, such as the amount of money seen on the website, whether it is registered in China, whether the website is registered within one year, and so on. We also use statistical methods such as chi-square detection to evaluate features. After that, we use two machine learning algorithms for model training: Random Forest、Deep Neural Networks, and divide it into training sets and test sets in a ratio of 8:2. The results are compared and, finally, the predictions are compared with previous studies.
Appears in Collections:	[Department of Computer Science and Information ] Master's Theses

Files in This Item:

File	Description	Size	Format
107THU00394012-001.pdf		3828Kb	Adobe PDF	273	View/Open

Loading...