空間型構理論預測行為分析實證研究 -展場平面配置型態與參觀者路徑選擇之影響關係

步入21世紀之現在，有關空間中行為分佈及其形式，吾等擁有足夠之工具可對以往謎題一一解惑，由英國倫敦大學Bill Hillier 教授所創可量化空間資訊之空間型構理論（Space Syntax）（動線、視覺、人潮移動… …等）掌握了實質環境被建置前或後之空間分佈傾向（劉秉承，2018）。本文欲理解展場空間配置形式（矩陣均一分佈、中軸線對稱、完全不對稱）之行為移動分佈趨勢，並藉由空間型構理論視域分析方法（Visibility Graph Analysis）之代理人群聚分析（Gate Counts）預測與掌握。本研究透過實地展出三種展場配置類型之觀展者移動分佈記錄，並將觀測值匯入分析軟體中，執行群聚分析分佈預測數值與實地觀測值，兩者之統計分析關連分佈比對。藉以驗證空間型構理論對於空間配置形式與路徑選擇等行為分佈移動之預測準確度。本實驗一共執行三種不同展場配置類型之觀測，分別是配置A『矩陣均一分佈』、配置B『中軸線對稱』，以及配置C『完全不對稱』。此三種展場配置類型，在設計階段會先使用視域分析軟體（Depthmap）進行空間資訊運算與行為分佈傾向預測，再與實際觀測結果交叉比對分析。初步觀測結果，三種代表配置A、B、C的實際觀測結果與代理人群聚分析(Gate Counts) 兩者之統計關聯分佈決定係數R-square(Rn-Cn)值分別是0.089、0.236、0.527，只有第三種『完全不對稱』之配置略高於0.5，為一正相關之實驗記錄，佐證當空間配置若是『中軸線對稱』或『矩陣均一分佈』排列時，皆較容易造成使用者路徑選擇困難與方位迷失現象，並降低觀展品質，反而在『完全不對稱』之配置形式中，其空間組構之行為分佈能藉由分析軟體預測，且群聚軟體分析數值顯示『完全不對稱』配置類型，為一種較高之空間組構自明性之空間配置形式，空間觀展者較能透過『局部』視覺資訊狀態推斷『整體』空間組構，而能有相對優質的觀展品質。

關鍵字

空間型構理論；空間配置組構形式；行為分佈預測；視域分析

並列摘要

In the 21st century, we have enough analytical tools to solve the problem and answer the questions between the human behavior distribution tendency and spatial configuration form based on Space Syntax theory founded by UCL Professor Bill Hillier (axial lines、 visual integration、 convex space etc.) before or after a real environment been built. In this research, it is the main intention to figure out the relationship between the spatial configuration form and human behavior distribution tendency (“matrix equal distributon layout” 、“symmertrical layout” 、 “asymmetrical layout”) of an exhibition space. Can we use the analytical tools of Space Syntax (visibility graph analysis and gate counts) to forecast the behavior? What is the difference of spatial information between those different spatial configurations? For this subject, a practical research is setting the rules of basical spatial form and observing three different typical layout of exhibition space with human behavior distribution tendency and crowds moving record. We input record values into analytical tool and run the R-square paramenters of gate counts values vs. observed ones, in order to prove the predicted values and the observed ones prediction accuracy. There are three different typical layouts of spatial configuration experiments in this research, project A、B and C, which were from the three different typical cases “matrix equal distributon layout” “symmertrical layout” and “asymmetrical layout” respectively. We used a spatial quantified analytical tool, UCL Depthmap 10 to forecast the human behavior trends and crowds moving then compare forecasted values and the observed values. The R-square(observation-Agent) values of project A、B and C values are 0.089、0.236、0.527 respectively, only the third project “asymmetrical layout” is highter than 0.5，it’s a positive correlation experimental record that can prove the theory hypothesis.The other two typical project “matrix equal distributon layout” “symmertrical layout” are easier to make user get lost and hard for way-finding, and also reduce the quality of exhibition visiting. But the quality is increased in project “asymmetrical layout”. Visitor can control the whole spatial configurations by the visual information in local part of whole space. The quality of exhibition visiting would be better relatively.