Object-relational spatio-temporal databases

We present an object-relational model for uniform handling of dimensional data. Spatial, temporal, spatio-temporal and ordinary data are special cases of dimensional data. The said uniformity is achieved through the concept of dimension alignment, which automatically allows lower dimensional data and queries to be used in a higher dimensional context. Unlike ordinary data, dimensional objects are interwoven. We introduce object identity (oid) fragments to circumvent data redundancy at logical level. Computed types are placed appropriately in a type hierarchy to allow maximal use of existing methods. A query language for spatio-temporal data is presented for associative navigation. A framework for algebraic optimization of the query language is suggested. A pattern matching language is designed for complex querying of spatio-temporal data which seamlessly extends the associative navigation in our query language. The pattern matching language recognizes special features of time and space providing an appropriate level of abstraction for application development compared to traditional languages. This reduces the need for embedding the query language in a lower level language such as C++. The pattern matching language is also dimensionally extensible. The pattern matching allows query of data with multiple granularities and continuous data. It also provides hooks for direct query of scientific data (observations). Our model is dimensionally extensible, and also an extension of a relational model for dimensional data. Moreover the dimensionality and addition of oids are mutually orthogonal concepts. Thus starting from classical ordinary data, one may migrate to higher forms of relational or object-relational data in any sequence, without having to recode application software. Our model does not deal with complex objects, which is left as a future extension.