Efficient Learning Strategy of Chinese Characters Based on Usage Frequency and Structural Relationship

Abstract

In this paper, we propose an efficient learning strategy to optimize the learning order of Chinese characters. First, we construct a network of Chinese characters according to their hierarchical structural relationships. Then, we develop a new measure of character importance as weight that incorporates the cost and the benefit of learning one character based on its usage frequency and structural relationship. During the learning process, the unlearned character weight is updated dynamically while its ancestor character is learned. Following this strategy, the learning order of Chinese characters is optimized in order of characters weights. The results show that our strategy is significantly efficient for learners to recognize more Chinese characters or reach higher cumulative usage frequency with less learning cost.