一個利用假設檢定和協同過濾來自我調整之動態QoS預測法

Abstract

Service-Based Systems的開發中，QoS是一個重要的考量因素來幫助決定要將哪些現存的服務有效地去組成一個新的服務。為了能選擇出合適的服務，許多預測QoS的方法被提出。然而在現今開放且動態的網路環境中，導致許多QoS值偏差而無法真實的反應，進而嚴重影響QoS預測的準確度，因此，在這種情況下也無法對Service的選用提供有效的幫助。本文中，我們提出一個自我調整之動態QoS預測法，利用統計的假設檢定來剔除QoS偏差異常值，並以此改善在動態環境下預測的準確度。我們也做多項實驗來評估提出的方法，實驗數據顯示我們的預測結果在一般的情況下準確度優於現存的方法。

During development of service-based systems (SBS), the quality of services (QoS) plays an important role at helping select more suitable services, based on QoS values. There are several QoS prediction approaches proposed; however, in an open and dynamic network environment, QoS values cannot be predicted accurately because the collected or derived QoS values may not indicate the realistic software quality. The inaccuracy caused by the deviation of QoS values cannot provide users an effective way to select services. In this thesis, we propose a framework to dynamic QoS Prediction via Hypothesis Testing and Collaborative Filtering. Using Hypothesis Testing is to remove deviation value and to improve accuracy of QoS prediction. The experimental results indicate that our approach has higher prediction accuracy than other methods.