Research Of Algorithm Visualization Instruction Based On Three-Layer Structure Model

Data structures and algorithms are important core issues in computerscience education. However, people usually have difficulty tounderstand, because of their abstract and dynamic features. One way toimprove their understanding is to provide visualizations to make theabstract concepts more concrete.This thesis first introduces the origin of algorithm visualization,common animation techniques and some foreign representatives ofalgorithm visualization systems. Then the paper analyzes the mainreasons which block these systems widely used in teaching from survey:(1) Developing visualizations costs a lot of time; (2) The currentalgorithm visualization systems lack interaction .with users, so learnerscan only view animations passively without active engagement. This maylead to reduce the visual teaching effect; (3) Most of the systems give nofeedback on student's performance. Based on engagement taxonomy andBloom's taxonomy, metrics for determining effectiveness of visualizationare discussed and an experiment is given to measure the relationshipbetween engagement levels and learner's depth of understanding.In order to overcome the present deficiency, the concept ofalgorithm simulation is applied and a three-Layer Structure Model isprovided. The execution of algorithm can be regarded as changing performance on special data structure. Algorithms can be studied bydiverse states of data structure. According to the hierarchy of differentconcepts, the data structures are divided into three structural levels:Abstract data types (ADT) which correspond to logic structure of actuallife; Basic data types (FDT) which correspond to physical structurerealized in computer; Concept data types (CDT) which are theintermediate layer between ADT and FDT. Algorithm simulation andteaching application are based on the three abstract levels.Based on three-Layer Structure Model, not only algorithmsimulation possesses the advantage of the traditional algorithmvisualization, but also joins the new characteristic such as the interaction,easily creating algorithm animation and offering automatic feedback.Algorithm visualization based on this model has positive meanings inpopularizing visualization in computer instruction and improving theteaching results.