Markov Chain Monte Carlo algorithms are often used to sample combinatorial structures such as matchings and independent sets in graphs. A Markov chain is defined whose state space includes the desired sample space, and which has an appropriate stationary distribution. By simulating the chain for a sufficiently large number of steps, we can sample from a distribution arbitrarily close to the stationary distribution. The number of steps required to do this is known as the mixing time of the Markov chain. In this thesis, we consider a number of Markov chains for sampling matchings, both in general and more restricted classes of graphs, and also for sampling independent sets in claw-free graphs. We apply techniques for showing rapid mixing based on two main approaches: coupling and conductance. We consider chains using single-site moves, and also chains using large block moves. Perfect matchings of bipartite graphs are of particular interest in our community. We investigate the mixing time of a Markov chain for sampling perfect matchings in a restricted class of bipartite graphs, and show that its mixing time is exponential in some instances. For a further restricted class of graphs, however, we can show subexponential mixing time. One of the techniques for showing rapid mixing is coupling. The bound on the mixing time depends on a contraction ratio b. Ideally, b < 1, but in the case b = 1 it is still possible to obtain a bound on the mixing time, provided there is a sufficiently large probability of contraction for all pairs of states. We develop a lemma which obtains better bounds on the mixing time in this case than existing theorems, in the case where b = 1 and the probability of a change in distance is proportional to the distance between the two states. We apply this lemma to the Dyer-Greenhill chain for sampling independent sets, and to a Markov chain for sampling 2D-colourings.

From a system developer's perspective, designing a spoken dialogue system can be a time-consuming and difficult process. A developer may spend a lot of time anticipating how a potential user might interact with the system and then deciding on the most appropriate system response. These decisions are encoded in a dialogue strategy, essentially a mapping between anticipated user inputs and appropriate system outputs. To reduce the time and effort associated with developing a dialogue strategy, recent work has concentrated on modelling the development of a dialogue strategy as a sequential decision problem. Using this model, reinforcement learning algorithms have been employed to generate dialogue strategies automatically. These algorithms learn strategies by interacting with simulated users. Some progress has been made with this method but a number of important challenges remain. For instance, relatively little success has been achieved with the large state representations that are typical of real-life systems. Another crucial issue is the time and effort associated with the creation of simulated users. In this thesis, I propose an alternative to existing reinforcement learning methods of dialogue strategy development. More specifically, I explore how XCS, an evolutionary reinforcement learning algorithm, can be used to find dialogue strategies that cover large state spaces. Furthermore, I suggest that hand-coded simulated users are sufficient for the learning of useful dialogue strategies. I argue that the use of evolutionary reinforcement learning and hand-coded simulated users is an effective approach to the rapid development of spoken dialogue strategies. Finally, I substantiate this claim by evaluating a learned strategy with real users. Both the learned strategy and a state-of-the-art hand-coded strategy were integrated into an end-to-end spoken dialogue system. The dialogue system allowed real users to make flight enquiries using a live database for an Edinburgh-based airline. The performance of the learned and hand-coded strategies were compared. The evaluation results show that the learned strategy performs as well as the hand-coded one (81% and 77% task completion respectively) but takes much less time to design (two days instead of two weeks). Moreover, the learned strategy compares favourably with previous user evaluations of learned strategies.

This thesis presents Topological Domain Theory as a powerful and flexible framework for denotational semantics. Topological Domain Theory models a wide range of type constructions and can interpret many computational features. Furthermore, it has close connections to established frameworks for denotational semantics, as well as to well-studied mathematical theories, such as topology and computable analysis. We begin by describing the categories of Topological Domain Theory, and their categorical structure. In particular, we recover the basic constructions of domain theory, such as products, function spaces, fixed points and recursive types, in the context of Topological Domain Theory. As a central contribution, we give a detailed account of how computational effects can be modelled in Topological Domain Theory. Following recent work of Plotkin and Power, who proposed to construct effect monads via free algebra functors, this is done by showing that free algebras for a large class of parametrised equational theories exist in Topological Domain Theory. These parametrised equational theories are expressive enough to generate most of the standard examples of effect monads. Moreover, the free algebras in Topological Domain Theory are obtained by an explicit inductive construction, using only basic topological and set-theoretical principles. We also give a comparison of Topological and Classical Domain Theory. The category of omega-continuous dcpos embeds into Topological Domain Theory, and we prove that this embedding preserves the basic domain-theoretic constructions in most cases. We show that the classical powerdomain constructions on omega-continuous dcpos, including the probabilistic powerdomain, can be recovered in Topological Domain Theory. Finally, we give a synthetic account of Topological Domain Theory. We show that Topological Domain Theory is a specific model of Synthetic Domain Theory in the realizability topos over Scott's graph model. We give internal characterisations of the categories of Topological Domain Theory in this realizability topos, and prove the corresponding categories to be internally complete and weakly small. This enables us to show that Topological Domain Theory can model the polymorphic lambda-calculus, and to obtain a richer collection of free algebras than those constructed earlier. In summary, this thesis shows that Topological Domain Theory supports a wide range of semantic constructions, including the standard domain-theoretic constructions, computational effects and polymorphism, all within a single setting.

No description available.

No description available.

In this paper, we show how online management of both quality of service (QoS) and provider revenue can be performed in CDMA cellular networks by adaptive control of system parameters to changing traffic conditions. The key contribution is the introduction of a novel call admission control and bandwidth degradation scheme for real-time traffic as well as the development of a Markov model for the admission controller. This Markov model incorporates important features of 3G cellular networks, such as CDMA intra- and inter-cell interference, different call priorities and soft handover. From the results of the Markov model the threshold for maximal call degradation is periodically adjusted according to the currently measured traffic in the radio access network. As a consequence, QoS and revenue measures can be optimized with respect to a predefined goal. To illustrate the effectiveness of the proposed QoS/revenue management approach, we present quantitative results for the Markov model and a comprehensive simulation study considering a half-day window of a daily usage pattern.

In this paper, we identify and analyze structural properties which reflect the functionality of a Web site. These structural properties consider the size, the organization, the composition of URLs, and the link structure of Web sites. Opposed to previous work, we perform a comprehensive measurement study to delve into the relation between the structure and the functionality of Web sites. Our study focuses on five of the most relevant functional classes, namely Academic, Blog, Corporate, Personal, and Shop. It is based upon more than 1,400 Web sites composed of 7 million crawled and 47 million known Web pages. We present a detailed statistical analysis which provides insight into how structural properties can be used to distinguish between Web sites from different functional classes. Building on these results, we introduce a content-independent approach for the automated coarse-grained classification of Web sites. A naïve Bayesian classifier with advanced density estimation yields a precision of 82% and recall of 80% for the classification of Web sites into the considered classes.

Ontology matching has been widely studied. However, the resulting on-tology mappings can be rather unstable when the participating ontologies or util-ized secondary sources (e.g., instance sources, thesauri) evolve. We propose an evolution-based approach for assessing ontology mappings by annotating their cor-respondences by information about similarity values for past ontology versions. These annotations allow us to assess the stability of correspondences over time and they can thus be used to determine better and more robust ontology mappings. The approach is generic in that it can be applied independently from the utilized match technique. We define different stability measures and show results of a first evaluation for the life science domain.

Parallel database systems have to support the effective parallelization of complex queries in multi-user mode, i.e. in combination with inter-query/inter-transaction parallelism. For this purpose, dynamic scheduling and load balancing strategies are necessary that consider the current system state for determining the degree of intra-query parallelism and for selecting the processors for executing subqueries. We study these issues for parallel hash join processing and show that the two subproblems should be addressed in an integrated way. Even more importantly, however, is the use of a multiresource load balancing approach that considers all potential bottleneck resources, in particular memory, disk and CPU. We discuss basic performance tradeoffs to consider and evaluate the performance of several load balancing strategies by means of a detailed simulation model. Simulation results will be analyzed for multi-user configurations with both homogeneous and heterogeneous (query/OLTP) workloads.

We classify data quality problems that are addressed by data cleaning and provide an overview of the main solution approaches. Data cleaning is especially required when integrating heterogeneous data sources and should be addressed together with schema-related data transformations. In data warehouses, data cleaning is a major part of the so-called ETL process. We also discuss current tool support for data cleaning.