儘管先前已經有針對攻擊檢測相關的研究提出，但是大多數主要是利用以用戶為導 向的攻擊檢測，將惡意用戶進行檢測和刪除。過去也有關注檢測受攻擊項目的研究。其中有研究所使用的方法需要密集的 user-item matrix，並透過一些已知的標準先將項目進行分組。本研究基於協作過濾的推薦系統其所使用的 user-item matrix 資料，在沒有先前的那些限制下，利用 Beta Distribution 的方法來檢測受攻擊的項目。其實驗結果表明，整個模擬攻 擊的檢測率大於 80%，錯殺率小於 16%。 ;A recommender system providing appropriate items to the user, effectively helping them to find items that may be of interest. The most common recommendation method is collaborative filtering. However, these recommender systems can be injected with false data to create false ratings to push or nuke specific items. This will greatly affect the user trust in the recommender system. After all, it is important that the recommender system recommends a trusted item.
Although previous studies have investigated how to detect attacks, most focus on user-orientation detection, i.e., detect and remove malicious users. To our best knowledge, only a few previous studies have considered how to detect items under attacks. However, one of the proposed methods requires a dense user-item matrix and clustering items into groups based on some known criteria. Therefore, this study proposes detecting items under attack without these constraints, using beta distribution with the user-item matrix utilized with the collaborative filtering recommender system. Experimental results show that the detection rate of the simulated attack is more than 80%, and the false rate is less than 16%.
