Towards a Model for Spatio-Temporal Schema Selection John F. Roddick School of Computer and Information Science, University of South Australia, The Levels Campus, Mawson Lakes, South Australia 5095, Australia. Email: roddick@cis.unisa.edu.au Fabio Grandi, Federica Mandreoli and Maria Rita Scalas Dipartimento di Elettronica, Informatica e Sistemistica, Universit` a di Bologna, Viale Risorgimento, 2, I-40136, Bologna, Italy. Email: fgrandi, fmandreoli, mrscalas @deis.unibo.it Abstract Schema versioning provides a mechanism for handling change in the structure of database systems and has been investigated widely, both in the context of static and tem- poral databases. With the growing interest in spatial and spatio-temporal data as well as the mechanisms for holding such data, the spatial context within which data is format- ted also becomes an issue. This paper presents a gener- alised model that accommodates schema versioning within static, temporal, spatial and spatio-temporal relational and object-oriented databases. Keywords: Schema Evolution, Schema Versioning, Spatio-Temporal Databases. 1. Introduction The motivation for schema versioning (which can be re- garded for some purposes as the application of temporal se- mantics to the schema of a database) is to provide a mech- anism to handle change in the real world in a data indepen- dent fashion. For example, structural changes such as the addition or deletion of attributes, the splitting or coalesc- ing of relations, the addition or deletion of methods, etc. should be achievable with the minimum impact to both the data and to the applications. It also provides a mechanism for versions of a database structure to facilitate alternative structural representations to the same data. A previous sur- vey of the field [22] provides a review of the area and a glos- sary of temporal database concepts [11] gives the following definitions: Schema Evolution A database system supports schema evolution if it permits modification of the database schema without the loss of extant data. No support for previous schemata is required. Schema Versioning A database system accom- modates schema versioning if it allows the query- ing of all data, both retrospectively and prospec- tively, through user-definable version interfaces. The development of spatial and, more recently, spatio- temporal databases [1, 9, 10, 16] is an important research topic. Recent research has shown that there are similar problems that could be considered as the spatial equiva- lent to the problems experienced by time-related data and changing schemata and for which temporal databases and schema versioning provide some solutions. For example, changes in database structure in time and the ability to ver- sion schemata have their spatial analogue in the locations that a given schema is to apply and the ability to apply dif- ferent schemata according to where the data is to be appli- cable. For example, the differences applicable across land administration authorities (qv. [18]). This paper therefore investigates the lessons learnt from the theory and development of temporally oriented schema versioning and examines whether these can be applied to spatial and spatio-temporal databases. In doing so, it is ac- knowledged that the spatial domain differs in several funda- mental ways and solutions appropriate to handling time will have to be carefully re-examined for the spatial context. The paper is structured as follows. Section 2 reviews temporal schema versioning and provides some background to the model to be presented later. Section 3 then inves- tigates the analogies between the temporal and spatial do- mains and provides some motivation for spatial schema ver- sioning. Section 4 proposes a generalised spatio-temporal schema selection model and further discussion and research directions are provided in Section 5.