A mereological theory of frames of reference

In this paper a mereological theory of frames of reference is presented. It shows that mereology extended by the notion of granularity and approximation is sufficient to provide a theory for location based features of frames of reference. More complex theories, taking also into account orientation and metric properties can be built as extensions of the presented theory. In order to take the hierarchical organization of frames of reference into account we introduce the notion of stratified approximation to facilitate transformations between different levels of granularity. This paper shows that the ontological grounding of the theory of frames of reference into mereology allows us to give a clear semantics to the notion ‘degree of parthood’ which is central to the notion of approximation. It also shows how epistemic aspects which affect the use of frames of reference help us to understand the feature of epistemic vagueness and the way it affects the notions of approximation and of degree of parthood. This paper proposes a mereological theory of frames of reference. There is a broad body of literature on frames of reference in at least four major areas: (i) in physics, in particular in the contexts of mechanics and relativity "!$#&%(') *+!$,.-) ; (ii) in linguistics, where people are interested in the semantics of spatial prepositions and in the insights into human cognition, which language can provide /0 1(23 4 5 6)7983:4 ;< =38 ; (iii) in psychology where, for instance in work in the distinction between figure and ground in human cognition ,?>I83J , the formal theory of parthood, has not, hitherto, been held to provide a basis for theorizing about frames of reference. This is probably because the most salient features of frames of reference are characterized by ordering relations and metric properties, features which require theories much stronger than mereology for their formalization. Examples of metric frames of reference are the ones used in physics to measure distance, weight, temperature, etc. Examples of frames of reference based on ordering relations are those of cardinal directions (N,S,E, and W), the distinction between left and right relative to your body axis, clock and calendar time, etc. However it will turn out that mereology with only some extensions will be sufficient to formally describe locational frames of reference. 1 This paper will show that mereology can help us to understand the more basic features of frames of reference which are already pointed to by the Gestalt psychologists in their distinction between figure and ground ,?><!$8 % . Here the ground is seen as a frame of reference in which the figure is located. An important point here is that the ground, if it is to provide a frame of reference must have a relatively simple structure, which, as we shall show, can be understood in terms of mereology. In order to use mereology as a formal foundation, however, it must be extended by the feature of granularity. We will refer to this theory as ‘granular mereology’. (See also ,?K"-) for further arguments along these lines.) Since frames of reference are often quite coarse, as in the case of the frame of reference which divides your surroundings into the part in front of you and the part behind, the specification of location of the figure within its ground (or of the referenced object within the frame of reference) is often rough or approximate. In this respect we will build upon the theory of granular partitions by KL,.-3:)6 . It will turn out that the resulting formalism is quite close to the notions of rough sets M"6( 2(' and rough mereology M ,.23 , which are extensions of set theory and mereology developed in the context of data analysis and data mining. One important feature of frames of reference is heir hierarchical organization. An example is the tree-like structure formed by the subdivision of London into Borroughs such as Westminster, Camden, etc at one level of granularity and parks and neighborhoods like Hyde Park, Soho, etc. at another level of granularity. Within a frame of reference based on this tree structure the relation between entities and the frame of reference, i.e., between figure and ground, can be specified at multiple levels of granularity. Consider the sentences A ‘John is in Hyde Park’ B ‘The Rocky Mountains are in the Western United States’. Here hierarchically ordered systems of places are used as frames of reference. As an alternative to (A) one can say, for example: A N John is in London. And to (B) one can say B N The Rocky Mountains are partly located in Montana, Idaho, Wyoming, Nevada, Colorado, Utah, Arizona, and New Mexico. Given formal representations of the location of entities within a frame of reference it is often necessary to transform approximations between different levels of granularity. The notion of stratified approximation introduced below will facilitate these kinds of transformations. It is important to distinguish between: (a) the study of the nature and the formal structure of frames of reference, and (b) the study of their application in certain contexts. This is because (a) refers to the study of the formal ontology of frames of reference. (b), on the other hand, refers to the question of how epistemic issues, for example issues pertaining to the limits on human knowledge and the representation of human knowledge, affect the specification of location in frames of reference and our capability to switch between levels of granularity in the ways illustrated in (A, A N ) and (B, B N ). From the perspective of formal ontology approximations with respect to a frame of reference are rough but crisp. Epistemic issues then give, as we shall see, raise to 2 vagueness. Vagueness hereby is understood in the sense that there are multiple, equally good, approximations which are consistent with the knowledge at hand. This epistemic understanding of vagueness needs to be distinguished from semantic vagueness which affects the ways names like ‘Mount Everest’ refer to parts of the surface of Earth B !$#&J3;%4 O 6(= -4') KL,.-3:)P . We refer to the latter as to semantic vagueness and to the former as to epistemic vagueness. Since in this paper we exclusively focus on epistemic vagueness it will be often sufficient to use the term vagueness. In order to provide a mereological theory that addresses the points raised above we start by extending mereology in order to take into account the feature of granularity. We then give a very general formal account of levels of granularity. In the following section we introduce the notion of approximation and show how stratified approximations can be defined across multiple levels of granularity. We then discuss epistemic issues that give raise to vagueness and extend the crisp formalism of stratified approximation in order to take epistemic vagueness into account. In the end we discuss related work and give the conclusions. The examples we use throughout the paper will be mostly spatial in nature. Due to its general nature, the underlying theory can easily be extended to domains of other sorts. Q R" SUTWVDXY [Z VDZ\V? ]VD VDX^SUXY_[ ]S0 `aSb dcLSb ] X In this section I give a formal definition of frames of reference as an extension of the theory of granular partitions which was originally introduced in KL, -):)6 . Q feLgih[jlk.mon]pogik)p.k.g]k qsr]k A frame of reference is a triple, tvuxwzy\{ | }~(| y€‚| ƒ~(| „l…‡† y\{ | }~ is a cell structure with a partial ordering defined by } which forms a finite tree. y€d| ƒ~ is the target domain which is a partial ordering which satisfies the axioms of extensional closure mereology (CEM) O 64= 23 . The projection mapping „zˆ {Š‰‹€ is an order-homomorphism from { into € . Before going into details, let us consider some examples. In figure 1 some cell structures of frames of reference popular in artificial intelligence are shown. Part (i) shows nine cells of a cell structure with ten cells (the tenth, which has all cells represented in the figure as subcells, is omitted here) are shown. When projected onto the surface of the Earth in such a way that the horizontal lines are in line with lines of longitude, the vertical lines are in line with lines of latitude, and the cell labeled Œ coincides with the location of some entity  , then the regions carved out by projecting of the cells onto the surface of the Earth are called ‘North of  ’, ‘Northeast of  , and so on. The entity  and the portions of Earth carved out in the way described are parts of the target domain € . The spatial projection of the cells onto the surface of Earth is a specific form of the projection „ . The frame of reference as a whole is formed by the three components: cell structure, target domain, projection. For an extended discussion of this specific family of frames of reference see B =)6(23 .