評估使用者所繪製之叢集圖形繪圖之美學之研究

Abstract

文獻中已存在許多不同的方法被用來判斷完整圖形繪圖的可讀性。本研究著重於分析使用者根據給定的一個鄰接表與一個叢集分群所創造一個叢集圖形繪圖，當中我們尤其重視叢集圖形繪圖美學的相關信息。這些美學分析分為三個部分：第一，分析使用者所產生的叢集圖形繪圖的代表性，包括識別某些類型的圖形繪圖，例如局部對稱性、叢集內彎曲的邊的數目、網格圖、和交叉線; 第二，分析特定視覺特徵的重要性，例如顯示橋和叢集的聚合力; 第三，分析繪圖於創建過程和個人畫圖的行為。實驗結果顯示使用者在提供叢集分群的資訊下能對叢集圖形繪圖做一些改善：減少交叉線數、叢集內部更密集、叢集更明顯的存在、以及對稱最大化顯示。從本質上說，本文顯示在圖形繪圖製美學和視覺效果的整體性均能有顯著改善。

Many different approaches have been employed to judge the readability of presented graph layouts. The main focus of this study is to analyze how the users create clustered graph drawing based on a given cluster set and a given adjacency list of the clustered graph, in which the analysis on underlying aesthetic information of clustered graph drawings is focused. These aesthetic analyses are divided into the following three parts. First, we analyze the representation of clustered graph drawings generated by participants, which includes recognizing certain types of graph drawings, such as local symmetries, intra-inter cluster bends, grid drawings, and edge crossings. Second, we analyze the importance of particular visual features, such as representation of bridge and cluster cohesion. Third, we analyze the graph creation process and individual drawing behavior. The results indicate that our participants made improvement in several factors such as a smaller number of edge crossings, higher cluster cohesion, more obvious presence of clusters and maximizing symmetric display. In essence, this study shows a significant improvement in the overall performance of graph drawing aesthetics and visualizations.