基於主題識別之自動化測試技術

Abstract

目前的自動化測試工具在開始測試之前仍然有許多前置作業需要處理，才有辦法產生出易讀的測試報告，很大一部分的原因在於只有人類有辦法定義畫面所包含的語意，以及畫面上的每一個按鈕、輸入欄所代表的意義。這篇論文提出了一套方法讓測試工具自動辨別畫面的語意，自動產生相對應的測試序列，並判別測試序列是否通過測試。藉由這個方法，可以在沒有測試人員介入的情況下，產生易讀的測試報告。

Current automated testing tools still have a lot of preparation to process before they start testing to produce an easy-to-read test report. The reason is that only humans have a way to define the meaning contained on the screen, as well as the meaning of each button and input field on the screen. In this thesis, we present a method for the testing tool to automatically distinguish the topic of each page of webpage, automatically generate the corresponding test sequence, and determine whether the test sequence passed the test. With this method, an easy-to-read test report can be generated without the intervention of a tester.