應用動態樁技術於合作式網頁應用程式測試

Abstract

現今，在網際網路上大量的免費人力資源通常被運用來減少測試成本與驗證軟體，如：線上遊戲與開放原始碼軟體。然而，傳統的合作測試方法在不考慮大眾測試者有共同的測試偏好與測試者素質參差不齊的情況下，往往會有測試時間難以收斂與測試報告的可信度不高等議題。為了加速收斂網站測試，有必要應用細顆粒的網頁應用程式模型來平行化測試工作。在這篇研究中，首先我們提出狀態轉換圖來為使用者的執行行為建模以達到分解網頁應用程式測試問題的目的。更進一步為了達到加速測試速度與改善測試報告的品質，並且減少合作測試的總成本，我們提出了一個動態樁技術來引導使用者進行測試。動態樁技術結合所提出的細顆粒的網頁應用程式模型可以在大眾測試者沒有察覺的情況下改變測試環境來導引他們解決子問題。實驗結果顯示我們所提出的方法可以減少50﹪的測試成本與增加30﹪的偵測效能。

Nowadays, large volunteers creeping on internet are usually treated as free human resources for reducing test cost and validating software, like online games and open source software. However, traditional collaborative testing design and management approach encounters the long due time and doubting test report resulting from the common preferences of users and unqualified testers, respectively. A fine-grained Web application model is essential to refine job assignments for speeding up test coverage. In this thesis, we first propose State Transition Diagram to model the users‘ runtime behaviors for decomposing Web application testing problem. Then, based on this fine-grained Web application model, a dynamic stubbing technique which allows folk testers contribute themselves in solving sub-problems with barely noticing the change of test environment is proposed for achieving faster test coverage speed and improving the quality of test report, and hence reduces the total cost of collaborative testing. The experimental results show that our proposed approach can reduce 50% test cost and increase 30% detection performance.