The development of extended pattern matching operators and a supporting operator framework for relational database management systems

Many database users rely on relational database management systems (RDBMSs) for the storage, retrieval and analysis of their data. RDBMSs support the retrieval and analysis of data by providing pattern matching operators that match target data against one or more specified patterns. However, support for pattern matching operations is insufficient in current RDBMS products, as these systems support a small number of pattern matching operations with limited functionality. Users are asking increasingly difficult pattern matching questions regarding more complex data such as protein sequences and medical images. Current RDBMSs provide some support for storing and manipulating newer types of data such as video, audio, spatial and temporal data, but support for advanced general pattern matching operations on relational data remains lacking.; The contributions of this thesis in addressing this problem are as follows. (1) We have developed a unifying framework for pattern matching operators. This framework assists users in identifying the proper operator for a given query. The framework also assists operator developers by providing an organizational matrix for understanding the generalization relationships among existing pattern matching operators and by identifying potential new operators through areas of the framework that are not currently supported. (2) We have developed an efficient procedural implementation for the Set HAS operator as part of a Relational Algebra package implemented using Oracle's PL/SQL language, showing that sophisticated pattern matching operators like Set HAS can be practically used on large data sets. We have also implemented dynamic functionality supporting Set HAS variants such as Range HAS and Vary HAS, showing that dynamic operator implementations are feasible. We have used this implementation of Set HAS to develop an implementation of the more general and powerful MATCH operator. (3) We have extended Set HAS into new dimensions, developing and implementing Associated Value HAS, Weighting HAS, and Self HAS variants. (4) We have designed and implemented Bag HAS, a new pattern matching operator that extends advanced pattern matching from set data to bag (multiset) data, thereby allowing users to directly answer queries regarding bag data that were previously difficult or impossible.