運用蟻群最佳化方法實現個人化行動學習系統

Applying Ant Colony Optimization to the Implementation of a Personalized Mobile Learning System

江傳文(CHUAN-WEN CHIANG)

《科學與工程技術期刊》 14卷1期 (2018/03) Pp. 13-24

摘要

本文中，吾人主要探討試卷組成問題之求解。在滿足多重評估準則的要求下，此一問題之目標旨在由大型題庫中選取適當的試題進而組成最佳的試卷內容。由於試卷組成問題具有NP-hard之特性，於是許多研究便以各種近似演算法為解決此一問題之方案，試圖在合理的計算時間內獲致令人滿意的問題解。然而，目前大多數的解決方案在問題求解效能上的表現仍存在著可供改進之空間，我們因此提出一種以螞蟻搜尋技術為基礎的建構式演算法。此一演算法的主要特色在於採用一種嶄新的建構圖形，藉此導引人工螞蟻在決策過程中得以選取有效之構成問題解的元件。實驗結果顯示本研究所提出之方法在問題求解效能方面有顯著的優異表現。此外，為了驗證本文所提出方法的實用性兼且擴展其應用層面，吾人也將整合數位學習、無線網路以及行動裝置等元素，實作出一款具體可用的個人化行動學習系統。

關鍵字

蟻群最佳化方法 ； 行動學習 ； 試卷組成問題

並列摘要

In this paper, we consider the test-sheet composition problem. The objective of this problem is to compose an optimal test sheet that meets multiple assessment criteria from a large item bank. The test-sheet composition problem is known to be NP-hard. Due to the intractability of the problem, research efforts have focused on approximation algorithms to acquire satisfactory suboptimal solutions within a reasonable computation cost. However, most realistic approaches for solving the test-sheet composition problem can still be improved. We therefore propose a novel constructive algorithm based on ant colony optimization. The proposed algorithm adopts a new type of constructive graph for leading artificial ants in decision-making to select effective solution components. Experimental results demonstrated that the proposed approach was efficacious for test-sheet composition. A personalized mobile learning system is also implemented to demonstrate the practicality of the proposed algorithm.

並列關鍵字

ant colony optimization (ACO) ； mobile learning ； test-sheet composition