以機器學習特徵應用於學習診斷之研究

Abstract

為了增進老師與學生之間的互動以提高學生學習的意願，最主要的重點是讓老師能迅速的了解學生的學習狀況，藉此改進教學的內容。在先前本實驗室(Tsao, 2010)的研究中，我們能夠藉著智慧型電腦系統自動評量分數的幫助，讓老師們能減少批改學生成績的時間。因此，接下來的研究便是希望能夠利用電腦系統進行自動診斷學生答案，指出學生缺少的概念，配合先前發展的自動評估系統，讓師生雙方都能更瞭解學生的學習情況，以增進學習的品質。 一般的學習診斷系統，通常是直接將學生的答案與老師提供的標準答案做比較，即可直接顯示出學生所缺少的部份；但此方法受限於必須先知道標準答案且因標準答案時常會參雜過多的字，而導致無法確定哪些是學生真正缺少的部份。 因此，本實驗的目的在於不需要老師提供標準答案，能夠自動的從學生答案中進行分析而產生題目的核心概念，進而診斷出學生缺少的部份。

For enhancing the interaction between students and teachers, it is important to allow teachers to understand the student’s learning level and improve the teaching materials. In the previous research by our laboratory, we can decrease the time of grading via an intelligent computer automatic assessment system. Therefore, we expect the following research can diagnose the answers and indicate the inadequate concept by automatic computer systems, and corresponding to the previous automatic assessment system, we can understand students’ learning situation for improving the learning quality. General learning diagnosis systems contrast the students’ response and standard answers. Unfortunately, this method is restricted by known answers and the answers which contain too many words not to ensure which part the student lacks. As a result, the goal of this experiment is to analyze the core concept of the subject automatically without additional standard answers to diagnose the students’ response.