摘要
為了防止機器人程式，透過模擬瀏覽器的行為，進而破解網頁進入系統，執行某些惡意註冊或留言動作，許多網站都是特過扭曲變形文字或圖案辨識來阻擋機器人程式進入系統執行惡意操作，這些方式多數是透過需要人的識別與辨認能力來阻擋機器人程式，但現在Big Data與cloud computing盛行，根據Google研究，現在最高科技的人工智慧技術，可以讓機器輕易辨識出各類扭曲文字CAPTCHA驗證碼內容，而且準確率可以到達99.8%，本篇文章提出的方法是透過人的認知關聯能力來阻擋機器人程式，依據Chris Argyris所提出的「推論階梯（Ladder of Inference）」心智模型來分析本篇方法，本篇方法可以完整運用到全部的七個心智活動，目前AI人工智慧功能尚無法完美模擬出完整人類心智活動，故本篇提出的認知關聯方法，是可以有效阻擋機器人入侵程式。
並列摘要
CAPTCHA technique is now widely used to prevent web robots intentionally hacking into web sites by twisting letters and background graphics. However, according to the latest research, this verification solution is no longer reliable. A Google algorithm is proven to be able to bust CAPTCHA with 99.8 percent accuracy. We propose a new method here which is based on human cognitive relevance ability to block web robots and analyze it with Ladder of Inference model, developed by Chris Argyris,1990. The method we propose here uses all 7 mental activities. With the fact that no AI programs has been proven to be able to simulate them all simultaneously, we believe our solution is capable of preventing malicious program invention effectively.