Chris Mellish

Bio: Chris Mellish is an academic researcher from University of Aberdeen. The author has contributed to research in topics: Natural language generation & Natural language. The author has an hindex of 38, co-authored 162 publications receiving 6660 citations. Previous affiliations of Chris Mellish include University of Sussex & University of Edinburgh.

Programming in Prolog

Abstract: Since the first publication of ''Programming in Prolog'' in 1981, Prolog has continued to attract an unexpectedly great deal of interest in the computer science community and is now seen as a potential basis for an important new generation of programming languages and systems. In this second edition, the authors have improved the presentation and corrected various minor errors to provide a textbook as well as a reference work for everyone who wants to study and use Prolog as a practical programming language. Various examples show how useful programs can be written with the Prolog system that exists today. The authors concentrate on teaching the ''core'' Prolog; all examples conform to this standard and will run on most existing Prolog implementations. Some of the existing Prolog implementations are listed in the appendices with indications as to how diverge from the standard.

Advances in Instance Selection for Instance-Based Learning Algorithms

Abstract: The basic nearest neighbour classifier suffers from the indiscriminate storage of all presented training instances. With a large database of instances classification response time can be slow. When noisy instances are present classification accuracy can suffer. Drawing on the large body of relevant work carried out in the past 30 years, we review the principle approaches to solving these problems. By deleting instances, both problems can be alleviated, but the criterion used is typically assumed to be all encompassing and effective over many domains. We argue against this position and introduce an algorithm that rivals the most successful existing algorithm. When evaluated on 30 different problems, neither algorithm consistently outperforms the other: consistency is very hard. To achieve the best results, we need to develop mechanisms that provide insights into the structure of class definitions. We discuss the possibility of these mechanisms and propose some initial measures that could be useful for the data miner.

Current research in natural language generation

Abstract: Current Research in Natural Language Generation is derived from the Second European Natural Language Generation Workshop, which was held in Edinburgh in April 1989. The papers included in this volume were selected from revised versions of some of the papers presented at the workshop. The book provides a snapshot of the current research in NLG, with particular emphasis on the work conducted by the European research community. It is aimed at an audience already familiar with NLG. Even though the different papers provide introductory technical material where necessary, in general, this material alone seemed insufficient to enable the uninitiated to understand completely the different arguments. However, adequate references are provided. The book is divided into four main sections: text planning (four papers), linguistic realization (two papers), building descriptions (three papers), and connectionist approaches (two papers). The first section contains contributions by Hovy, Scott and de Souza, Cawsey, and McKeown et al. The first paper, entitled "Unresolved issues in paragraph planning," by Hovy, is a position paper that raises seven unresolved problems in discourse planning, mainly from the perspective of Rhetorical Structure Theory (RST) (Mann and Thompson 1988). These problems are divided into two groups: problems concerning the theory and representation of coherence relations, and algorithmic problems. The importance of this paper is that it focuses the discussion on NLG on crucial issues to which researchers have to address themselves. The second paper, "Getting the message across in RST-based generation," by Scott and de Souza, addresses the problem of generating text that achieves a communicative goal effectively. To this end, the authors look upon style in terms of a reader's ease of processing a text, rather than in terms of aesthetics. The main contribution of the paper is that it presents explicit heuristics grounded in psycholinguistic evidence to control the realization of RST discourse relations. I found Section 3.2, "Making the text sensitive to the communicative setting," where the authors link the generation of textual markers to context sensitivity, somewhat problematic. In addition, since the heuristics presented in the paper were not implemented, some analysis of how they interact would have been useful. The third paper, "Generating explanatory discourse," by Cawsey, discusses an interactive content and discourse planner whose output is tailored to the changing capabilities of the user, and which can also handle interruptions and remedial discourse. The paper offers a novel approach that addresses specific criteria in discourse planning. Its contributions include: the use of an approach similar to Litman's (1985) to separate content and discourse planning; the hierarchical decomposition of schemata; and the use of an agenda for content planning. This paper provides a clear exposition

Choosing the content of textual summaries of large time-series data sets

Abstract: Natural Language Generation (NLG) can be used to generate textual summaries of numeric data sets. In this paper we develop an architecture for generating short (a few sentences) summaries of large (100KB or more) time-series data sets. The architecture integrates pattern recognition, pattern abstraction, selection of the most significant patterns, microplanning (especially aggregation), and realisation. We also describe and evaluate SumTime-Turbine, a prototype system which uses this architecture to generate textualsummaries of sensor data from gas turbines.

Introduction to Multiagent Systems

Abstract: From the Publisher: An agent is an entity with domain knowledge, goals and actions. Multi-agent systems are a set of agents which interact in a common environment. Multi-agent systems deal with the construction of complex systems involving multiple agents and their coordination. A multi-agent system (MAS) is a distributed computing system with autonomous interacting intelligent agents that coordinate their actions so as to achieve its goal(s) jointly or competitively.

Toward integrating feature selection algorithms for classification and clustering

Abstract: This paper introduces concepts and algorithms of feature selection, surveys existing feature selection algorithms for classification and clustering, groups and compares different algorithms with a categorizing framework based on search strategies, evaluation criteria, and data mining tasks, reveals unattempted combinations, and provides guidelines in selecting feature selection algorithms. With the categorizing framework, we continue our efforts toward-building an integrated system for intelligent feature selection. A unifying platform is proposed as an intermediate step. An illustrative example is presented to show how existing feature selection algorithms can be integrated into a meta algorithm that can take advantage of individual algorithms. An added advantage of doing so is to help a user employ a suitable algorithm without knowing details of each algorithm. Some real-world applications are included to demonstrate the use of feature selection in data mining. We conclude this work by identifying trends and challenges of feature selection research and development.

Supervised Machine Learning: A Review of Classification Techniques

Abstract: The goal of supervised learning is to build a concise model of the distribution of class labels in terms of predictor features. The resulting classifier is then used to assign class labels to the testing instances where the values of the predictor features are known, but the value of the class label is unknown. This paper describes various supervised machine learning classification techniques. Of course, a single chapter cannot be a complete review of all supervised machine learning classification algorithms (also known induction classification algorithms), yet we hope that the references cited will cover the major theoretical issues, guiding the researcher in interesting research directions and suggesting possible bias combinations that have yet to be explored.