Jerome Robinson

Data Analysis for Query Processing

Data analysis is needed in connection with query processing, to produce data summary information in the form of rules or assertions that allow semantic query optimisation or direct query answering without consulting the data itself. The goal of an intelligent analyser in this context is to produce robust rules, stable in the presence of data changes, which allow easy rule maintenance as data changes, and provide rapid query reformulation, refutation or answering. It must also limit the rule set to rules useful for query processing.

Constructing Inter-relational Rules for Semantic Query Optimisation

Semantic query optimisation is the process by which a user query is transformed into a set of alternative queries each of which returns the same answer as the original. The most efficient of these alternatives is then selected, for execution, using standard cost estimation techniques. The query transformation process is based on the use of semantic knowledge in the form of rules which are generated either during the query process itself or are constructed according to defined heuristics. Previous research has tended to focus on constructing rules applicable to single relations and does not take advantage of the additional semantic knowledge, inherent in most databases, associated with relational joins. Our paper seeks to address this weakness by showing how the rule derivation process may be extended to the generation of inter-relational rules using an approach based on inductive learning.

A Statistical Approach to Rule Selection in Semantic Query Optimisation

Semantic Query Optimisation makes use of the semantic knowledge of a database (rules) to perform query transformation. Rules are normally learned from former queries fired by the user. Over time, however, this can result in the rule set becoming very large thereby degrading the efficiency of the system as a whole. Such a problem is known as the utility problem. This paper seeks to provide a solution to the utility problem through the use of statistical techniques in selecting and maintaining an optimal rule set. Statistical methods have, in fact, been used widely in the field of Knowledge Discovery to identify and measure relationships between attributes. Here we extend the approach to Semantic Query Optimisation using the Chi-square statistical method which is integrated into a prototype query optimiser developed by the authors. We also present a new technique for calculating Chi-square, which is faster and more efficient than the traditional method in this situation.

Attribute-Pair Range Rules

This paper examines the properties of metadata in the form of IF THEN rules which contain two predicates on attributes of a relational database table. For example: a(15. 30) ⟹ d(243. 271), which means “if the value of attribute ‘a’ in a tuple is in the range 15 to 30 then the value of attribute ‘d’ will be in the range 243 to 271.” Metadata of this kind is useful in Semantic Query Optimisation and Remote Cache Management. The two predicates (antecedent and consequent) in each rule are Selection Conditions or constraints of the type found in database queries. Each condition therefore denotes a subset of a database table. Rules can be cascaded, using subrange containment as the link between successive rules. The set of rules can therefore be regarded as a set of edges in a Condition Dependency Graph, and using the rule-set is path discovery in the graph. The purpose of the current paper is to introduce some of the properties of attribute-pair range rules.

A transitive closure and magic functions machine

An extended version of our SIMD Relational Algebraic Processor is presented. In addition to the usual relational and set operations the new machine has the ability to recycle its responder sets internally. This allows it to perform repeated joins, for example, without external intervention and so achieve operations such as path discovery and transitive closure in graphs stored as relations, and to evaluate various types of recursive query. The many compiled methods for recursive query evaluation are applicable in this system as in any other relational database, and can be efficiently evaluated because of the in-built recursive and iterative capability of our machine. The Magic Functions approach has a clear connection with the machine since it uses relations as magic functions.

A High Performance Relational Algebraic Processor for Large Knowledge Bases

The principal requirement for knowledge-based systems is the ability to perform pattern-directed search over large numbers of variously-sized objects. The precise interpretation of the word ’object’ depends upon the knowledge representation formalism being used. Examples of objects range from simple numbers or variables, through tuples or logic clauses, to quite large objects such as relations, semantic nets or procedures. Objects are conceptually held in some form of associative (i.e. content-addressable) store. Pattern-directed search of this store is a process of recognition or selection, which in general results in a set of objects being discovered to match the search criteria. The output from a search operation may be called the responder set. AI-related processing consists, implicitly or explicitly, of further operations on responder sets. Amongst these operations, set intersection and transitive closure occur so frequently that they have been proposed, along with pattern-directed search, as prime candidates for hardware support for AI (Fahlman et al 1983). Furthermore, set intersection is one of a family of relational algebraic operations which is central to all database-like computation. Finally, transitive closure is just one of a family of graph-manipulation operations (other examples are shortest-path and activation-propagation) which find a use throughout Computer Science.

Data extraction from Web data sources

An explanation is given of the basic data structures used in a new page analysis technique to create wrappers (data extractors) for the result pages produced by Web sites in response to user qeries via Web page forms. The key structure called a tpGrid is a representation of the web page, which is easier to analyse than the raw HTML code. The analysis looks for repetition patterns of sets of tagSets, which are defined in the paper.

Using a Network of Workstations to Enhance Database Query Processing Performance

Query processing in database systems may be improved by applying parallel processing techniques. One reason for improving query response time is to support the increased number queries when databases are made accessible from the Internet.

Improved Data Retrieval Using Semantic Transformation

Semantic query optimisation uses knowledge about properties of the data, represented as a set of subset descriptor rules, to transform a query into another form that can be executed in a more efficient manner but still yields the same result as the original query. Commonly this ’semantic knowledge’ in the form of rules is generated either during the query process itself or else is constructed in advance according to defined heuristics. Over a period of time the rule set may, therefore, become very large and the number of semantically equivalent queries that may be derived rises exponentially. Each rule use creates a new equivalent query. The problem is to identify one near optimal alternative query in a time that is minimal and also short relative to the overall query execution time. In this paper we propose a method for measuring the effectiveness of each rule and present a fast algorithm which selects the most cost effective transformations to directly yield the optimal alternative query. Experiments carried out on a large publicly available dataset show worthwhile savings using the approach.

Utilising Multiple Computers in Database Query Processing and Descriptor Rule Management

A fundamental problem to be solved in systems that derive rules from database tables to use in query optimisation is the workload involved. If the data server has to do the work it can interfere with query processing and cause slower query answering, which is the opposite of the required effect. This paper reports our investigation of the use of multiple workstations in the same local network as the data server to derive and maintain sets of rules describing data subsets. These rules are used in query optimisation. In a local area network of workstations, one computer accepts SQL queries and data manipulation commands from networked clients. This computer provides an interface to one or more database management systems located on computers in the network. It uses a collection of subset-descriptor rules for query reformulation before forwarding the semantically optimised query. It manages a set of workstations in the network, to derive and maintain the rules. The workstations are ordinary networked computers whose spare computing capacity is utilised by spawning background programs on them.