地理空间中的地理要素往往具有模糊性,这种模糊源于人对现实世界的概念化过程,因此具有主观特性.基于模糊集理论,尽管有很多途径对模糊地理要素对应的隶属度函数进行探讨,但是基于认知实验的方法最直接的反映了人们对相应要素的概念化过程中的模糊性.以中关村地区为例,设计了基于地标的问卷调查,并计算每个地标属于"中关村地区"这一概念的隶属度,进而采用支持向量回归方法,得到该要素的隶属度函数.该方法具有实验实施简单,结果便于管理的特点.最后,我们分析了中关村的隶属度函数的一些空间分布特征及其原因.

In practice, vagueness is a common phenomenon of geographical features. The vagueness of a feature often\ncomes from human conceptualization of the realworld. Modeling of vague features will undoubtedly contribute to more\nprecisely handling spatial knowledge. In recent studies, a number of theoreticalmethods have been employed tomodel\nvague features, where fuzzy set theory is in common use. Following that theory, the degree that an elementbelongs to a\nfuzzy set can be expressed by a number between 0 and 1. We can thus establish a correspondingmembership function\n(MF) for a fuzzy set. Recently, much literature focusson vague features and proposes approaches to establish correspond-\ningMFs. They include approaches based on cognitive expermi ent, remotely sensed data andGIR (Geographical Informa-\ntion Retrieval). Due to the subjectivity ofvagueness, spatial cognitive expermi ents provide a directway to represent the\nvagueness of individuals conceptualization of corresponding features. However, previousmethods based on cognition cost\nhighly and are somewhathard to control the result, since subjects in such expermi ents are asked to delineate boundariesof\nvague arealobjects. Landmarksplay an mi portantrole in individuals developmentofspatial cognition. It is thus relatively\neasy for individuals to perceive a landmark and decidewhether it iswithin a given region ornot. In this research, we took\nZhongguancun in Beijing city as an example, since it is complexwith differentmeanings, such as educationa,l political and historicalmeanings. A questionnaire is designed to collectmembership degrees of 30 landmarks which are in the\nregion ofZhongguancun. These 30 landmarks, which are selected from the maxmi um potential region corresponding to\nZhongguancun, can be abstracted to point features. Theybelong to different types, such asoffice building, hote,l schoo,l\nrecreation place, and natural feature. For each landmark, the subjects are askedwhether it iswithin Zhongguancun, for\nwhich three optional answers are provided: YES, NO, andNOT SURE. By collecting all answer sheets, we can compute\na score ofeach landmark. Such a value can be viewed as themembership degree that the corresponding position belongs to\nthe conceptof“Zhongguancun”. However, since Zhongguancun is a two-dmi ensionalvague object, itshould be represen-\nted by amembership function (MF) likez=f(x,y). We thusneed to find outan appropriate interpolationmethod to ob-\ntain theMF. In this research, supportvector regression (SVR) is adopted to compute theMF. Comparedwith convention-\nal interpolationmethods, such as IDW (InverseDistanceWeighted) method, the proposed approach is easy to mi plement\nand the results are convenient to be managed. Additionally, SVR provides a mechanism to obtain a trade-off between\ngoodness of fit and generality by adjusting some parameters, such asγfor radialbasis kernel functions. Based on the re-\nsultofmembership function, we also investigate spatialdistribution propertiesofZhongguancun and find outsome interest-\ning points. Since Zhongguancun hasbeen viewed as“the silicon valley inChina”, it is closely relatedwith such concepts\nas high-tech industry, university, and so on. Consequently, the landmarks associated with these concepts always have\nhighermembership degrees. On the contrary, lower scores are assigned to some recreation places and natural features. In\nsummary, from a pointofview ofbehavior, the currentconcept of Zhongguancun is farbeyond the scope as an administra-\ntive unit, both spatially and functionally, sincemany factors influence its internal representation of individuals.