An intelligent agent, operating in an external world which cannot be fully described in its internal world model, must be able to monitor the success of a previously generated plan and to respond to any errors which may have occurred. The process of error analysis requires the ability to reason in an expert fashion about time and about processes occurring in the world. Reasoning about time is needed to deal with causality. Reasoning about processes is needed since the direct effects of a plan action can be completely specified when the plan is generated, but the indirect effects cannot. For example, the action `open tap' leads with certainty to `tap open', whereas whether there will be a fluid flow and how long it might last is more difficult to predict. The majority of existing planning systems cannot handle these kinds of reasoning, thus limiting their usefulness. This thesis argues that both kinds of reasoning require a complex internal representation of the world. The use of Qualitative Process Theory and an interval-based representation of time are proposed as a representation scheme for such a world model. The planning system which was constructed has been tested on a set of realistic planning scenarios. It is shown that even simple planning problems, such as making a cup of coffee, require extensive reasoning if they are to be carried out successfully. The final Chapter concludes that the planning system described does allow the correct solution of planning problems involving complex side effects, which planners up to now have been unable to solve.

The IRDS standard is an international standard produced by the International Organisation for Standardisation (ISO). In this work the process for producing standards in formal standards organisations, for example the ISO, and in more informal bodies, for example the Object Management Group (OMG), is examined. This thesis examines previous models and classifications of standards. The previous models and classifications are then combined to produce a new classification. The IRDS standard is then placed in a class in the new model as a reference anticipatory standard. Anticipatory standards are standards which are developed ahead of the technology in order to attempt to guide the market. The diffusion of the IRDS is traced over a period of eleven years. The economic conditions which affect the diffusion of standards are examined, particularly the economic conditions which prevail in compatibility markets such as the IT and ICT markets. Additionally the consequences of the introduction of gateway or converter devices into a market where a standard has not yet been established is examined. The IRDS standard did not have an installed base and this hindered its diffusion. The thesis concludes that the IRDS standard was overtaken by new developments such as object oriented technologies and middleware. This was partly because of the slow development process of developing standards in traditional organisations which operate on a consensus basis and partly because the IRDS standard did not have an installed base. Also the rise and proliferation of middleware products resulted in exchange mechanisms becoming dominant rather than repository solutions. The research method used in this work is a longitudinal study of the development and diffusion of the ISO/EEC IRDS standard. The research is regarded as a single case study and follows the interpretative epistemological point of view.

Most parametric software cost estimation models used today evolved in the late 70's and early 80's. At that time, the dominant software development techniques being used were the early 'structured methods'. Since then, several new systems development paradigms and methods have emerged, one being Jackson Systems Development (JSD). As current cost estimating methods do not take account of these developments, their non-universality means they cannot provide adequate estimates of effort and hence cost. In order to address these shortcomings two new estimation methods have been developed for JSD projects. One of these methods JSD-FPA, is a top-down estimating method, based on the existing MKII function point method. The other method, JSD-COCOMO, is a sizing technique which sizes a project, in terms of lines of code, from the process structure diagrams and thus provides an input to the traditional COCOMO method. The JSD-FPA method allows JSD projects in both the real-time and scientific application areas to be costed, as well as the commercial information systems applications to which FPA is usually applied. The method is based upon a three-dimensional view of a system specification as opposed to the largely data-oriented view traditionally used by FPA. The method uses counts of various attributes of a JSD specification to develop a metric which provides an indication of the size of the system to be developed. This size metric is then transformed into an estimate of effort by calculating past project productivity and utilising this figure to predict the effort and hence cost of a future project. The effort estimates produced were validated by comparing them against the effort figures for six actual projects.

We address the question of how to obtain effective fusion of identification information such that it is robust to the quality of this information. As well as technical issues data fusion is encumbered with a collection of (potentially confusing) practical considerations. These considerations are described during the early chapters in which a framework for data fusion is developed. Following this process of diversification it becomes clear that the original question is not well posed and requires more precise specification. We use the framework to focus on some of the technical issues relevant to the question being addressed. We show that fusion of hard decisions through use of an adaptive version of the maximum a posteriori decision rule yields acceptable performance. Better performance is possible using probability level fusion as long as the probabilities are accurate. Of particular interest is the prevalence of overconfidence and the effect it has on fused performance. The production of accurate probabilities from poor quality data forms the latter part of the thesis. Two approaches are taken. Firstly the probabilities may be moderated at source (either analytically or numerically). Secondly, the probabilities may be transformed at the fusion centre. In each case an improvement in fused performance is demonstrated. We therefore conclude that in order to obtain robust fusion care should be taken to model the probabilities accurately; either at the source or centrally.

The topic of this thesis is the development of knowledge based statistical software. The shortcomings of conventional statistical packages are discussed to illustrate the need to develop software which is able to exhibit a greater degree of statistical expertise, thereby reducing the misuse of statistical methods by those not well versed in the art of statistical analysis. Some of the issues involved in the development of knowledge based software are presented and a review is given of some of the systems that have been developed so far. The majority of these have moved away from conventional architectures by adopting what can be termed an expert systems approach. The thesis then proposes an approach which is based upon the concept of semantic modelling. By representing some of the semantic meaning of data, it is conceived that a system could examine a request to apply a statistical technique and check if the use of the chosen technique was semantically sound, i.e. will the results obtained be meaningful. Current systems, in contrast, can only perform what can be considered as syntactic checks. The prototype system that has been implemented to explore the feasibility of such an approach is presented, the system has been designed as an enhanced variant of a conventional style statistical package. This involved developing a semantic data model to represent some of the statistically relevant knowledge about data and identifying sets of requirements that should be met for the application of the statistical techniques to be valid. Those areas of statistics covered in the prototype are measures of association and tests of location.

Requirements for systems to continue to operate satisfactorily in the presence of faults has led to the development of techniques for the construction of fault tolerant software. This thesis addresses the problem of error detection and recovery in distributed systems which consist of a set of communicating sequential processes. A method is presented for the `a priori' design of conversations for this class of distributed system. Petri nets are used to represent the state and to solve state reachability problems for concurrent systems. The dynamic behaviour of the system can be characterised by a state-change table derived from the state reachability tree. Systematic conversation generation is possible by defining a closed boundary on any branch of the state-change table. By relating the state-change table to process attributes it ensures all necessary processes are included in the conversation. The method also ensures properly nested conversations. An implementation of the conversation scheme using the concurrent language occam is proposed. The structure of the conversation is defined using the special features of occam. The proposed implementation gives a structure which is independent of the application and is independent of the number of processes involved. Finally, the integrity of inter-process communications is investigated. The basic communication primitives used in message passing systems are seen to have deficiencies when applied to systems with safety implications. Using a Petri net model a boundary for a time-out mechanism is proposed which will increase the integrity of a system which involves inter-process communications.

The ERS-1 Satellite was launched in July 1991 by the European Space Agency into a polar orbit at about 800 km, carrying a C-band scatterometer. A scatterometer measures the amount of backscatter microwave radiation reflected by small ripples on the ocean surface induced by sea-surface winds, and so provides instantaneous snap-shots of wind flow over large areas of the ocean surface, known as wind fields. Inherent in the physics of the observation process is an ambiguity in wind direction; the scatterometer cannot distinguish if the wind is blowing toward or away from the sensor device. This ambiguity implies that there is a one-to-many mapping between scatterometer data and wind direction. Current operational methods for wind field retrieval are based on the retrieval of wind vectors from satellite scatterometer data, followed by a disambiguation and filtering process that is reliant on numerical weather prediction models. The wind vectors are retrieved by the local inversion of a forward model, mapping scatterometer observations to wind vectors, and minimising a cost function in scatterometer measurement space. This thesis applies a pragmatic Bayesian solution to the problem. The likelihood is a combination of conditional probability distributions for the local wind vectors given the scatterometer data. The prior distribution is a vector Gaussian process that provides the geophysical consistency for the wind field. The wind vectors are retrieved directly from the scatterometer data by using mixture density networks, a principled method to model multi-modal conditional probability density functions. The complexity of the mapping and the structure of the conditional probability density function are investigated. A hybrid mixture density network, that incorporates the knowledge that the conditional probability distribution of the observation process is predominantly bi-modal, is developed. The optimal model, which generalises across a swathe of scatterometer readings, is better on key performance measures than the current operational model. Wind field retrieval is approached from three perspectives. The first is a non-autonomous method that confirms the validity of the model by retrieving the correct wind field 99% of the time from a test set of 575 wind fields. The second technique takes the maximum a posteriori probability wind field retrieved from the posterior distribution as the prediction. For the third technique, Markov Chain Monte Carlo (MCMC) techniques were employed to estimate the mass associated with significant modes of the posterior distribution, and make predictions based on the mode with the greatest mass associated with it. General methods for sampling from multi-modal distributions were benchmarked against a specific MCMC transition kernel designed for this problem. It was shown that the general methods were unsuitable for this application due to computational expense. On a test set of 100 wind fields the MAP estimate correctly retrieved 72 wind fields, whilst the sampling method correctly retrieved 73 wind fields.

An implementation of a Lexical Functional Grammar (LFG) natural language front-end to a database is presented, and its capabilities demonstrated by reference to a set of queries used in the Chat-80 system. The potential of LFG for such applications is explored. Other grammars previously used for this purpose are briefly reviewed and contrasted with LFG. The basic LFG formalism is fully described, both as to its syntax and semantics, and the deficiencies of the latter for database access application shown. Other current LFG implementations are reviewed and contrasted with the LFG implementation developed here specifically for database access. The implementation described here allows a natural language interface to a specific Prolog database to be produced from a set of grammar rule and lexical specifications in an LFG-like notation. In addition to this the interface system uses a simple database description to compile metadata about the database for later use in planning the execution of queries. Extensions to LFG's semantic component are shown to be necessary to produce a satisfactory functional analysis and semantic output for querying a database. A diverse set of natural language constructs are analysed using LFG and the derivation of Prolog queries from the F-structure output of LFG is illustrated. The functional description produced from LFG is proposed as sufficient for resolving many problems of quantification and attachment.

The assessment of the reliability of systems which learn from data is a key issue to investigate thoroughly before the actual application of information processing techniques to real-world problems. Over the recent years Gaussian processes and Bayesian neural networks have come to the fore and in this thesis their generalisation capabilities are analysed from theoretical and empirical perspectives. Upper and lower bounds on the learning curve of Gaussian processes are investigated in order to estimate the amount of data required to guarantee a certain level of generalisation performance. In this thesis we analyse the effects on the bounds and the learning curve induced by the smoothness of stochastic processes described by four different covariance functions. We also explain the early, linearly-decreasing behaviour of the curves and we investigate the asymptotic behaviour of the upper bounds. The effect of the noise and the characteristic lengthscale of the stochastic process on the tightness of the bounds are also discussed. The analysis is supported by several numerical simulations. The generalisation error of a Gaussian process is affected by the dimension of the input vector and may be decreased by input-variable reduction techniques. In conventional approaches to Gaussian process regression, the positive definite matrix estimating the distance between input points is often taken diagonal. In this thesis we show that a general distance matrix is able to estimate the effective dimensionality of the regression problem as well as to discover the linear transformation from the manifest variables to the hidden-feature space, with a significant reduction of the input dimension. Numerical simulations confirm the significant superiority of the general distance matrix with respect to the diagonal one. In the thesis we also present an empirical investigation of the generalisation errors of neural networks trained by two Bayesian algorithms, the Markov Chain Monte Carlo method and the evidence framework; the neural networks have been trained on the task of labelling segmented outdoor images.

The satellite ERS-1 was launched in July 1991 in a period of high solar activity. Sparse laser tracking and the failure of the experimental microwave system (PRARE) compounded the orbital errors which resulted from mismodelling of atmospheric density and hence surface forces. Three attempts are presented here to try and refine the coarse laser orbits of ERS-1, made prior to the availability of the full altimetric dataset. The results of the first attempt indicate that by geometrically modelling the satellite shape some improvement in orbital precision may be made for any satellite; especially one where no area tables already exist. The second and third refinement attempts are based on the introduction of data from some second satellite; in these examples SPOT-2 and TOPEX/Poseidon are employed. With SPOT-2 the method makes use of the orbital similarities to produce along-track corrections for the more fully tracked SPOT-2. Transferring these corrections to ERS-1 produces improvements in the precise orbits thus determined. With TOPEX/Poseidon the greater altitude results in a more precise orbit (gravity field and atmospheric errors are of less importance). Thus, by computing height differences at crossover points of the TOPEX/Poseidon and ERS-1 ground tracks the poorer orbit of ERS-1 may be improved by the addition of derived radial corrections. In the positive light of all three results several potential modification are suggested and some further avenues of investigation indicated.