| DC 欄位 |
值 |
語言 |
| DC.contributor | 資訊管理學系 | zh_TW |
| DC.creator | 鄭義穎 | zh_TW |
| DC.creator | Yi- ying Jheng | en_US |
| dc.date.accessioned | 2010-7-2T07:39:07Z | |
| dc.date.available | 2010-7-2T07:39:07Z | |
| dc.date.issued | 2010 | |
| dc.identifier.uri | http://ir.lib.ncu.edu.tw:444/thesis/view_etd.asp?URN=974203038 | |
| dc.contributor.department | 資訊管理學系 | zh_TW |
| DC.description | 國立中央大學 | zh_TW |
| DC.description | National Central University | en_US |
| dc.description.abstract | 隨著網路拍賣的普及，網路拍賣詐欺也逐漸變成犯罪的手法，其常見的手法便是不肖賣家透過拍賣網站所具有高度匿名與進入門檻低的特性，進行帳號與帳號之間相互哄抬評價，以創造高評價分數的假像來引誘買家，此種情形屢見不鮮。但之前大多數的研究僅利用社會網路分析（Social Network Analysis）來偵測哄抬評價之共犯群體，大都不能完整的偵測出整體詐欺群體的關係且非線上即時偵測。
因此，本研究針對此種詐欺共犯群體，提出一個模組化的詐欺共犯群體偵測流程，以彌補網路拍賣評價系統之不足。首先，我們利用K-Core分群演算法的概念來設計我們代理人搜尋路徑以達成可完整與立即地找出潛在的共犯群體；第二，透過本研究定義的資料前處理動作，進行資料清理；第三，使用PageRank演算法找出在群體中具有權威性與重要性的帳號，並計算其有效的指標；第四，使用Auction Fraud Rank演算法計算第二個有效的指標，此演算法是修改自PageRank演算法其目的是為了能夠讓演算法同時考量網絡結構性與帳號本身潛在的危險度；最後，我們使用調適性類神經推論系統(Adaptive-Network-based Fuzzy Inference System, ANFIS)結合社會網路分析與網路結構探勘(Web Structure Mining)來偵測群體中每個帳號的危險性。本研究使用真實案例的方式來檢驗所提出的系統架構是否可以有效幫助使用者找出潛在的共犯群體。
| zh_TW |
| dc.description.abstract | With the popularity of the online purchase, online auction fraud has become a kind of criminal in our daily life. The most common fraud method is that auctioneers use the characteristics of high anonymity and the lower thresholds of the e-Auctions to create multiple accounts and manipulate their reputations. In this way, they can deceive the buyer by their high reputations. But most of the previous researches focus on only using the Social Network Analysis to detect the inflated reputation behaviors of the auction fraud group. Thus, it can’t detect the relationship of whole group fraudsters and the execution process should not be Online real-time.
Therefore, the research proposes a new process which can detect collusive group: First, we use the concept of k-core clustering algorithm to design our searching path for Agent in order to capture the potential collusive group completely and immediately. Second, we define a data preprocessing to clean-up unrelated data. Third, we use the PageRank algorithm to discover authoritative and important accounts in the group and calculate the useful indicator. Four, Auction Fraud Rank algorithm, an extension to the standard PageRank algorithm, takes into account the importance of both Web structure and the potential risk of account in order to calculate the second useful indicator. Finally, we use ANFIS to combine SNA and WSN to detect the risk of each account in the group. In the research, we use real cases to validate whether the proposed system can effectively help auctioneers to find the potential collusive group.
| en_US |
| DC.subject | 調適性類神經模糊推論系統 | zh_TW |
| DC.subject | 社會網路分析 | zh_TW |
| DC.subject | 線上拍賣 | zh_TW |
| DC.subject | 網絡結構探勘 | zh_TW |
| DC.subject | 詐欺共犯群體 | zh_TW |
| DC.subject | PageRank演算法 | zh_TW |
| DC.subject | PageRank Algorithm | en_US |
| DC.subject | Fraud Group Detection | en_US |
| DC.subject | K-Core | en_US |
| DC.subject | Social Network Analysis | en_US |
| DC.subject | Online Auction | en_US |
| DC.subject | Web Structure Mining | en_US |
| DC.subject | ANFIS | en_US |
| DC.title | 結合社會網路分析與網絡結構探勘偵測網路拍賣哄抬評價之共犯群體 | zh_TW |
| dc.language.iso | zh-TW | zh-TW |
| DC.title | Combing Social Network Analysis with Web Structure Mining for detecting collusive fraud group in online auction | en_US |
| DC.type | 博碩士論文 | zh_TW |
| DC.type | thesis | en_US |
| DC.publisher | National Central University | en_US |