| 摘要: | 最近幾年，雲模型已經成功地整合到協同過濾處理對於解決稀疏性與可擴展性的問題。基於項目雲模型協同過濾演算法是轉換被使用者評分的項目資料成整體項目的特徵。基於使用者雲模型協同過濾演算法是轉換使用者的評分資料成使用者的整體偏好度。
在這篇論文裡，我們提出整合雲模型協同過濾系統，它整合了來自基於項目雲模型協同過濾系統的預測評分及基於項目雲模型協同過濾系統的預測評分。並且，這套系統裡採用特徵權重的機制對於融合計算使用者和項目的預測評分。
實驗結果證實，我們所提出的推薦系統表現優於先前所提出的模型，尤其是在較少鄰居的案例時，它能有效率地解決稀疏性和可擴展性的問題。
In recent years, cloud model has been successfully integrated into the collaborative filtering (CF) process for addressing the sparsity and the scalability problems. The item-based cloud model CF approach converts item’s ratings data by user into a global item’s characteristics. The user-based cloud model CF approach converts user’s ratings data into a global user’s preference.
In this paper, we present a Unified could-model CF system that integrates the pre-dictions from both the item-based and the user-based cloud model CF systems. This system adopts a significance weighting scheme for fusion the user and item prediction computation.
The experimental results showed that the suggested recommender system outper-formed previous models, especially for cases with small neighbors by effectively solv-ing the sparsity and the scalability problems.
Key |
