Simon H. Lavington

The Manchester Mark I and atlas: a historical perspective

In 30 years of computer design at Manchester University two systems stand out: the Mark I (developed over the period 1946-49) and the Atlas (1956-62). This paper places each computer in its historical context and then describes the architecture and system software in present-day terminology. Several design concepts such as address-generation and store management have evolved in the progression from Mark I to Atlas. The wider impact of Manchester innovations in these and other areas is discussed, and the contemporary performance of the Mark I and Atlas is evaluated.

Speeding up Knowledge Discovery in Large Relational Databases by Means of a New Discretization Algorithm

Most of the KDD (Knowledge Discovery in Databases) algorithms proposed in the literature have been applied to relatively small datasets and do not permit any integration with a DBMS. Hence, the application of these algorithms to the huge amounts of data found in current databases and data warehouses faces serious scalability problems, particularly the problem of excessive learning time. This paper investigates a way of improving the scalability of KDD algorithms, via discretization of ordinal or continuous attributes. This work has two novel aspects. First, we map a generic discretization primitive into an SQL query. Second, we propose a new discretization algorithm for classification tasks. We show how the new discretization algorithm can be implemented with good effect via the SQL primitive.

Interfacing knowledge discovery algorithms to large database management systems

Abstract The efficient mining of large, commercially credible, databases requires a solution to at least two problems: (a) better integration between existing Knowledge Discovery algorithms and popular DBMS; (b) ability to exploit opportunities for computational speedup such as data parallelism. Both problems need to be addressed in a generic manner, since the stated requirements of end-users cover a range of data mining paradigms, DBMS, and (parallel) platforms. In this paper we present a family of generic, set-based, primitive operations for Knowledge Discovery in Databases (KDD). We show how a number of well-known KDD classification metrics, drawn from paradigms such as Bayesian classifiers, Rule-Induction/Decision Tree algorithms, Instance-Based Learning methods, and Genetic Programming, can all be computed via our generic primitives. We then show how these primitives may be mapped into SQL and, where appropriate, optimised for good performance in respect of practical factors such as client–server communication overheads. We demonstrate how our primitives can support C4.5, a widely-used rule induction system. Performance evaluation figures are presented for commercially available parallel platforms, such as the IBM SP/2.

Hardware memory management for large knowledge bases

Large knowledge bases form an important applications area for parallel architectures. The special problems of this domain concern the movement and protection of large amounts of complex data in a hierarchy of storage devices and processing elements, in a manner which is sympathetic to the logical structure of the information. This structure is a reflection of the underlying knowledge representation formalism. An analysis of candidate formalisms leads to the specification of a memory architecture for large knowledge bases. The particular problems of paging are studied, and a scheme for semantic caching is proposed. The design of a fast cache employing highly-parallel pattern-directed searching is described. The performance of this cache in a multi-level memory hierarchy is given.

The performance of layered video over an IP network

We evaluate the quality of full-motion video, when transmitted over IP via a two-layer codec and CBQ routers in the presence of TCP traffic. We investigate the effects of packet size and distribution of I-frames. We derive guidelines for the cost-effective divisions of bandwidth between base and enhancement layers, so as to maintain perceived quality in the presence of known amounts of packet loss.

An Appreciation of Dina St Johnston (1930–2007) Founder of the UK's First Software House

In the 1950s there was no software industry. Dina St Johnston, who had learned to program whilst working for the computer manufacturer Elliott-Automation, founded Vaughan Programming Services in 1959. The company began to specialise in on-line systems for digital process control at a time when industrial automation was in its infancy. In due course the company developed its own platform-independent, timesharing, mini-operating system (MACE) and, in 1970, the Vaughan 4M microprocessor. Vaughan went on to become specialists in the supply of real time controllers for passenger railways. Dina St Johnston remained an active programmer until 1996.

Parallel Data Mining for Very Large Relational Databases

Data mining, or Knowledge Discovery in Databases (KDD), is of little benefit to commercial enterprises unless it can be carried out efficiently on realistic volumes of data. Operational factors also dictate that KDD should be performed within the context of standard DBMS. Fortunately, relational DBMS have a declarative query interface (SQL) that has allowed designers of parallel hardware to exploit data parallelism efficiently. Thus, an effective approach to the problem of efficient KDD consists of arranging that KDD tasks execute on a parallel SQL server. In this paper we devise generic KDD primitives, map these to SQL and present some results of running these primitives on a commercially-available parallel SQL server.

In the footsteps of Colossus: a description of Oedipus

A number of cryptanalysis projects were developed in the UK during the postwar years 1945-1955. One of the most significant of these was Oedipus, a special-purpose rapid analytical machine using novel digital storage. Oedipus was developed by GCHQ and the UK companies Elliott and Ferranti. Oedipus contained a large semiconductor associative memory, a magnetic drum with on-the-fly searching, and a high-speed RAM cache. Its history has only recently been made publicly available.

Knowledge Discovery from Client-Server Databases

The subject of this paper is the implementation of knowledge discovery in databases. Specifically, we assess the requirements for interfacing tools to client-server database systems in view of the architecture of those systems and of “knowledge discovery processes”. We introduce the concept of a query frontier of an exploratory process, and propose a strategy based on optimizing the current query frontier rather than individual knowledge discovery algorithms. This approach has the advantage of enhanced genericity and interoperability. We demonstrate a small set of query primitives, and show how one example tool, the well-known decision tree induction algorithm C4.5, can be rewritten to function in this environment.

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.