分析具節點權重之圖形之手繪繪圖美學之研究

Abstract

摘要 一個好的圖形繪圖可以幫助閱讀者快速理解其中的意涵且容易記憶，因此過去有許多學者針對圖形繪圖美學進行研究。早期的研究方法是直接請受測者判讀已由學者設計完成的圖形，近年的研究方法轉變為讓受測者重新調整初始化圖形繪圖以求完美的結構，直到近期的研究是只提供受測者一張連結關係表並依據實驗任務在平板電腦上繪製圖形，這個方法才真正讓受測者自由地表達出他們的繪圖邏輯；然而，這類的研究方法並未考量到權重的特性，權重的特性在圖形繪圖中就是某些節點的重要性高於其他節點或是某些連結的重要性高於其他連結，而這樣的特性存在各種現實環境中，舉例而言，城市的規模有大小之分，大城市的重要性顯然高於小城市，所以我們應給予大城市較高的權重，如此才能真實表達出各城市之間的關係。因此，本研究是應用前人的研究方法，並加以考量節點具有權重的特性，我們除了關注權重高的節點如何在圖形繪圖中呈現以外，還設計了四階段的實驗繪圖部分來驗證受測者的圖形繪圖是否呈現出學習效應。在實驗結束後，我們針對受測者所繪製的圖形繪圖進行分析，這些分析共有四個部分：第一，分析最終圖形繪圖具有的特徵，例如：交叉線的存在、節點與連結呈現棋盤式的結構；第二，分析受測者繪製圖形的過程及其繪圖策略；第三，利用實驗後問卷調查受測者的繪圖喜好；第四，依據不同屬性將受測者分成兩個群組，分析這兩個群組在各種特徵上的表現是否有差異。結果顯示在本研究中交叉線數最小化[5]仍然是受測者極重視的美學，此外當圖形繪圖中的節點具有不同權重的情況下，受測者更重視棋盤式結構的美學，所以我們建議繪圖軟體未來就應以這兩種美學作為設計的主要考量；最後我們也證實受測者的圖形繪圖的確表現出良好的學習效應。

關鍵詞：圖形繪圖美學、節點權重、學習效應、手繪圖形繪圖、視覺化

Abstract A good graph drawing enables fast reading and understanding of the information represented in the graph, hence numerous scholars study the aesthetics of graph drawing. Earlier works were only limited to ask participants to interpret graph designed by scholars. Later works require participants to rearrange the nodes in the initial graph drawings in order to represent the best layout. Recent works have asked participants to draw the graphs on a tablet PC according to adjacency list and task so that it allows participants to freely express their interpretation. However, weighted vertices were not considered in recent works. Weighted vertices refer to the assignment of weights to nodes in a graph to reflect the differences in importance of each node. Weighted vertices can be applied to many real world scenarios. For example, cities can be assigned weights based on their varying levels of importance where large cities are of greater importance than small cities. Larger cities are assigned with greater weights so that it could represent the actual relationship between the cities. Therefore, the work presented here builds on the prior work and the application of weighted vertices. We not only focus on how should vertices with high-weight be presented in the graph, but also design four-part experiments to detect the occurrence of the learning effect. After completing the experiments, we discussed four aspects on the graph drawings generated by users. First, we analyze the characteristics of the final graph drawings such as edge crossings and grid drawings. Second, we analyze the process of graph creation and participants’ strategies for creating the graph. Third, we analyze the drawing preferences of participants through the questionnaires. Fourth, we divide participants into two groups according to different attributes and analyze the difference between the groups on the above aspects. The results indicate that minimizing the number of edge crossings is still the most important aesthetic for users in this work and participants prefer the aesthetic of grid drawings in the condition that weight-vertices exists in graph drawings. It suggests that two aesthetics of minimizing number of edge crossings and grid drawings should be the main consideration for designing the graph drawing software. Finally, we verified that learning effect existed apparently when participants created several graph drawings.

Keywords： Graph drawing aesthetics, vertex-weighted, learning effect, user-sketched graph drawings, visualization