ALBERT

Notes: The properties of a modular neural network system designed to forecast movements of the long gilt futures contract are examnined. The examination illustrates that diagnostic measures can effectively inform end-users of the system when appropriately to switch modules given the likelihood of the degradation of forecasts. The modelling is based upon periods of market sentiment, and the inductive data models encapsulated, within the modules of the neural network system, are investigated. The models track, and also provide knowledge of , underlying market characteristics in different periods of market sentiment. The research provides deductions concerning the (dis-)similarity of underlying characteristics in differing periods of market sentiment.

Notes: The paper presents a method for helping knowledge engineers in modelling the knowledge involved in sequential diagnosis. In particular, we consider sequential diagnosis as a process which occurs in the following type of scenario: (1) there exist some candidate hypotheses which are to be pursued; (2) for each hypothesis there are some alternative tests which might be executed to pursue it; (3) the available knowledge about the world precludes projecting into the future (in other words, the available knowledge about the world does not afford the information needed for planning sequences of tests, i.e. for projecting the consequences of possible sequences of actions and picking the best sequence); (4) the choice of the next test (or tests) is made on the basis of the current state of knowledge; (5) the outcome of a test execution is affected by uncertainty. The suggested method addresses the problem of engineering the knowledge experts use for making decisions under uncertainty. A practical example of the method is also presented: at any time of the diagnostic process the expert is provided with a dynamically updated list of suggested tests in order to support him or her in the decision-making problem about which test to execute next.

Notes: The Internet offers a large potential for delivery of various information-based services, including the services of intelligent applications. As the availability of the Internet has grown, its value as a medium for the delivery of expert systems in particular has increased. There are now a large number of expert systems available on the Internet, including applications in industry, medicine, science and government. Though the Internet provides several advantages for expert system development, it also presents some special problems. These advantages and disadvantages are explored in more detail in this paper. The paper also presents a review of several Internet-based expert systems with a representative sample of publicly available applications, and a discussion of typical tools for developing Internet-based expert systems. A case study of an Internet-based expert system is presented as well.

Notes: This paper describes a connectionist inference architecture which performs standard symbolic inference on a subclass of first-order predicate calculus. This work first involves constructing efficient connectionist mechanisms to represent basic symbol components, dynamic bindings and basic symbolic inference procedures, and devising a set of algorithms which automatically translates input descriptions to localist neural networks. These connectionist mechanisms are built by taking an existing phase-locking mechanism and extending it further to obtain desirable features to represent and manipulate basic symbol structures. The existing phase-locking mechanism represents dynamic bindings very efficiently using temporal synchronous activity between neuron elements but has fundamental limitations in supporting standard symbolic inference. The extension addresses these limitations. The extended system encodes a significant subset of a Horn clause language in a connectionist style.

Notes: Connectionist network learning of context-free languages has so far been applied only to very simple cases and has often made use of an external stack. Learning complex context-free languages with a homogeneous neural mechanism looks like a much harder problem. The current paper takes a step toward solving this problem by analyzing context-free grammar computation (without addressing learning) in a class of analog computers called dynamical automata, which are naturally implemented in connectionist networks. The result is a widely applicable method of using fractal sets to organize infinite-state computations in a bounded state space. An appealing consequence is the development of parameter-space maps, which locate various complex computers in spatial relationships to one another. An example suggests that such a global perspective on the organization of the parameter space may be helpful for solving the hard problem of getting connectionist networks to learn complex grammars from examples.

Notes: Holographic reduced representations (HRRs) are a method for encoding nested relational structures in fixed-width vector representations. HRRs encode relational structures as vector representations in such a way that the superficial similarity of the vectors reflects both superficial and structural similarity of the relational structures. HRRs also support a number of operations that could be very useful in psychological models of human analogy processing: fast estimation of superficial and structural similarity via a vector dot-product; finding corresponding objects in two structures; and chunking of vector representations. Although similarity assessment and discovery of corresponding objects both theoretically take exponential time to perform fully and accurately, with HRRs one can obtain approximate solutions in constant time. The accuracy of these operations with HRRs mirrors patterns of human performance on analog retrieval and processing tasks.

Notes: Rough set theory (RST) offers an interesting and novel approach both to the generation of rules for use in expert systems and to the traditional statistical task of classification. The method is based on a novel classification metric, implemented as upper and lower approximations of a set and more generally in terms of positive, negative and boundary regions. Classification accuracy, which may be set by the decision maker, is measured in terms of conditional probabilities for equivalence classes, and the method involves a search for subsets of attributes (called ’reducts’) which do not require a loss of classification quality. To illustrate the technique, RST is employed within a state level comparison of education expenditure in the USA.

Notes: The main goal of this paper is to show how relatively minor modifications of well-known algorithms (in particular, back propagation) can dramatically increase the performance of an artificial neural network (ANN) for time series prediction. We denote our proposed sets of modifications as the ’self-momentum’, ’Freud’ and ’Jung’ rules. In our opinion, they provide an example of an alternative approach to the design of learning strategies for ANNs, one that focuses on basic mathematical conceptualization rather than on formalism and demonstration. The complexity of actual prediction problems makes it necessary to experiment with modelling possibilities whose inherent mathematical properties are often not well understood yet. The problem of time series prediction in stock markets is a case in point. It is well known that asset price dynamics in financial markets are difficult to trace, let alone to predict with an operationally interesting degree of accuracy. We therefore take financial prediction as a meaningful test bed for the validation of our techniques. We discuss in some detail both the theoretical underpinnings of the technique and our case study about financial prediction, finding encouraging evidence that supports the theoretical and operational viability of our new ANN specifications. Ours is clearly only a preliminary step. Further developments of ANN architectures with more and more sophisticated ’learning to learn’ characteristics are now under study and test.

Notes: This paper describes the design and implementation of an intelligent real-time supervision system for the monitoring and control of wastewater treatment plants. The system presents a hybrid structure which integrates knowledge engineering techniques and methodologies (an expert system based on production rules embodying non-mathematical or semiquantitative knowledge of unit operation) and algorithmic process-related methods (data acquisition, signal preprocessing, calculation of plant operation parameters etc.). The fundamental design aspects consist of temporal information treatment, artifact detection and a client- server structure which allows the telesupervision of various plants by a central system.

Notes: Although many knowledge-based systems (KBSs) focus on single-paradigm approaches to encoding knowledge (such as production rules), human experts rarely use a single type of knowledge to solve a real-world problem. A human expert usually combines a number of reasoning mechanisms. In recent years, rule-based reasoning (RBR), case-based reasoning (CBR) and model-based reasoning (MBR) have emerged as important and complementary reasoning methodologies in the intelligent systems area. For complex problem solving, it is useful to integrate RBR, CBR and MBR. In this paper, a hybrid epidemic screening KBS which integrates a deductive RBR system, an inductive CBR system and a quantitative MBR system is proposed. The system has been tested using real epidemic screening variables and data.

Notes: While knowledge-based systems are being used extensively to assist in making decisions, a critical factor that affects their performance and reliability is the quantity and quality of the knowledge bases. Knowledge acquisition requires the design and development of an in-depth comprehension of knowledge modeling and of applicable domain. Many knowledge acquisition tools have been developed to support knowledge base development. However, a weakness that is revealed in these tools is the domain-dependent and complex acquisition process. Domain dependence limits the applicable areas and the complex acquisition process makes the tool difficult to use. In this paper, we present a goal-driven knowledge acquisition tool (GDKAT) that helps elicit and store experts’ declarative and procedural knowledge in knowledge bases for a user-defined domain. The designed tool is implemented using the object-oriented design methodology under C++ Windows environment. An example that is used to demonstrate the GDKAT is also delineated. While the application domain for the example presented is reflow soldering in surface mount printed circuit board assembly, the GDKAT can be used to develop knowledge bases for other domains also.

Notes: In this paper we propose a connectionist model for variable binding. The model is topology dependent on the graph it builds based on the predicates available. The irregular connections between perceptron-like assemblies facilitate forward and backward chaining. The model treats the symbolic data as a sequence and represents the training set as a partially connected network using basic set and graph theory to form the internal representation. Inference is achieved by opportunistic reasoning via the bidirectional connections. Consequently, such activity stabilizes to a multigraph. This multigraph is composed of isomorphic subgraphs which all represent solutions to the query made. Such a model has a number of advantages over other methods in that irrelevant connections are avoided by superimposing positionally dependent sub-structures that are identical, variable binding can be encoded and multiple solutions can be extracted simultaneously. The model also has the ability to adapt its existing architecture when presented with new clauses and therefore add new relationships/rules to the model explicitly; this is done by some partial retraining of the network due to the superimposition properties.

Notes: Modus ponens is used in forward inference and backward inference, where the truth of the conclusion is inferred from the truth of the premise. In modus tollens, the falseness of the premise is inferred from the falseness of the conclusion. Although modus ponens is used in general connectionist production systems, modus tollens is rarely used, except in Quinlan’s proposed INFERNO system and in the system proposed by Thornber. A connectionist production system called ConnPS that can perform both modus ponens and modus tollens simultaneously is described. Compared to the INFERNO system, one of the advantages of ConnPS is its supervised learning ability. The rules and examples given as external knowledge are often erroneous and incomplete. In ConnPS, these rules can be refined by using the supervised learning. Both positive and negative examples are presented to ConnPS, onto which the external rules and observations are mapped. Moreover, ConnPS’s implementations of implications, conjunctions, disjunctions and negation are intuitively consistent with Boolean logic.

Notes: The usefulness of ‘user centred design’ as a concept in the design of software systems is considered. It is concluded that the concept is frequently used in a way so vacuous as to be devoid of meaning, and would be better discarded. Some suggestions are made for further research in this area; some modest suggestions are also offered to those who wish to develop systems of a type better suited to the needs of civilized society.

Notes: Increasing evidence over the past decade indicates that financial markets exhibit nonlinear dynamics in the form of chaotic behavior. Traditionally, the prediction of stock markets has relied on statistical methods including multivariate statistical methods, autoregressive integrated moving average models and autoregressive conditional heteroskedasticity models. In recent years, neural networks and other knowledge techniques have been applied extensively to the task of predicting financial variables.This paper examines the relationship between chaotic models and learning techniques. In particular, chaotic analysis indicates the upper limits of predictability for a time series. The learning techniques involve neural networks and case–based reasoning. The chaotic models take the form of R/S analysis to measure persistence in a time series, the correlation dimension to encapsulate system complexity, and Lyapunov exponents to indicate predictive horizons. The concepts are illustrated in the context of a major emerging market, namely the Polish stock market.

Notes: This paper concerns the study of destination choice modelling, more specifically identifying within some area (e.g. a city) the region where a particular store is the most favourable to be visited by individuals. An influence measure is constructed for each individual, which incorporates the modern technique known as Dempster–Shafer theory. Based on the evidence of the shopping destinations of individuals, geographical regions are found for levels of largest belief and plausibility (within Dempster–Shafer theory) for specific stores being the most favourable to visit. Additionally, this method may be used to identify the possible position of new stores, based on regions of most uncertainty or conflict in store choice. A prototype choice modelling system is introduced to enable the series of associated results to be easily visualized and analysed.

Notes: Fuzzy logic principles and neural networks, both being computational intelligence technologies, can be combined to produce synergetic effects through the formation of a unified approach which takes advantage of the benefits and at the same time counterbalances the flaws of the two technologies. In this paper, a fuzzy neural approach, which is characterized by its ability to suggest the appropriate adjustment of product quantity from various suppliers with different quality standards in a supply chain network, is presented. This approach is particularly useful in situations where multiple supply chain partners are involved to achieve the common objective of producing products to the best satisfaction of customer demands at the lowest possible cost. To validate the feasibility of this approach, a test has been conducted based on the proposed fuzzy neural approach with the objective of suggesting the appropriate selection of suppliers and the optimal quantity allocated to them to meet the required quality standards. This paper describes the methodology for the deployment of this proposed hybrid approach to enhance the machine intelligence of a supply chain network with the description of a case study to exemplify its underlying principles.

Notes: This paper focuses on the knowledge representation framework utilized by an integrated wastewater treatment expert system for diagnosing operational problems of an activated sludge plant. The system deals with events that may occur in all the units of an activated sludge plant and exploits on-line measurements, observations formed from data provided by laboratory analyses and empirical observations in an integrated manner. The system provides assistance to human experts to control the activated sludge process. It has been tested and evaluated in the pilot activated sludge plant of the Water and Air Quality Laboratory in the University of the Aegean.

Notes: The paper investigates the impact of trust on market efficiency and bilateral contracts. We prove that a market in which agents are trusted to the degree they deserve to be trusted is as efficient as a market with complete trustworthiness. In other words, complete trustworthiness is not a necessary condition for market efficiency. We prove that distrust could significantly reduce market efficiency, and we show how to solve the problem by using appropriately designed multiagent contracts. The problem of trust is studied in the context of a bilateral negotiation game between a buyer and a seller. It is shown that if the seller’s trust equals the buyer’s trustworthiness, then the social welfare, the amount of trade, and the agents’ utility functions are maximized. The paper also studies the efficiency of advance payment contracts as a tool for improving trustworthiness. It is proved that advance payment contracts maximize the social welfare and the amount of trade. Finally, the paper studies the problem of how to make agents truthfully reveal their level of trustworthiness. An incentive–compatible contract is defined, in which agents do not benefit from lying about their trustworthiness. The analysis and the solutions proposed in this paper could help agent designers avoid many market failures and produce efficient interaction mechanisms.

Notes: This paper uses the notion of control from programming languages to look at the organization of mental code. Data for the analysis comes principally from language breakdown. The paper first outlines the well known distinction between logic and control in algorithms and argues that the same distinction holds in mental code. Discussion then focuses mainly on control—the management of data flow—and shows that a variety of language disorders affect either the logic component of the mental algorithms for language (e.g., Specific Language Impairment) or the control component (e.g., Williams syndrome and Turner syndrome). A comparative study of the loss of morphology in Williams syndrome and Specific Language Impairment reinforces the logic/control split as an accurate guide to the explanation of linguistic behavior in these disorders. The data, moreover, are not accountable to sheer performance factors, but to the way the disorders disrupt the structure of mental algorithms. The paper closes with a discussion of how control and the management of cross-domain computation fit into recent theories of modular mental architecture and proposals about the explicitness of representations and their availability to working memory.

Notes: We discuss the use of probability-based natural language processing for Chinese text retrieval. We focus on comparing different text extraction methods and probabilistic weighting methods. Several document processing methods and probabilistic weighting functions are presented. A number of experiments have been conducted on large standard text collections. We present the experimental results that compare a word-based text processing method with a character-based method. The experimental results also compare a number of term-weighting functions including both single-unit weighting and compound-unit weighting functions.

Notes: Recently, the prevalence of information retrieval engines has created an important application of the automatic summarization as the display of retrieval results, whereby the user can quickly and accurately judge the relevance of texts returned as a result of a query. Here, rather than producing a generic summary, the summary that reflects the user's topic of interest (information need) expressed in the query would be considered as more suitable. This type of summary is often called query-biased summary.In this paper we present a method for producing query-biased summaries using lexical chains. Lexical chains are sequences of words that are in lexical cohesion relation with each other, and tend to indicate fragments of a text that form a semantic unit. Using lexical chains would enable to produce more coherent and readable summaries than previous approaches to query-biased summarization.To evaluate the effectiveness of our method, a task-based evaluation scheme is adopted. The results from the experiments show that query-biased summaries by lexical chains outperform others in the accuracy of subjects' relevance judgments.

Notes: Time-series prediction is important in physical and financial domains. Pattern recognition techniques for time-series prediction are based on structural matching of the current state of the time-series with previously occurring states in historical data for making predictions. This paper describes a Pattern Modelling and Recognition System (PMRS) which is used for forecasting benchmark series and the US S&P financial index. The main aim of this paper is to evaluate the performance of such a system on noise free and Gaussian additive noise injected time-series. The results show that the addition of Gaussian noise leads to better forecasts. The results also show that the Gaussian noise standard deviation has an important effect on the PMRS performance. PMRS results are compared with the popular Exponential Smoothing method.

Notes: Expert systems constitute a branch of artificial intelligence, and their unique characteristics enable computer systems to perform at the level of a human expert by the use of algorithms that can capture domain-specific knowledge. Through an extensive survey of research about ship-design expert systems, we conclude that although development and implementation of expert systems for ship design offer substantial barriers, the shipbuilding industry needs them nonetheless. We examine the major reasons that use of expert systems in this industry has lagged behind. These include the problems of integration with existing computer-aided design systems, problems of knowledge representation and sizable development costs. We discuss in detail some key factors that can lead to the solution of these problems.

Notes: A methodology based on case-based reasoning is proposed to build a topological-level influence diagram. It is then applied to a project proposal review process. The formulation of decision problems requires much time and effort, and the resulting model, such as an influence diagram, is applicable only to one specific problem. However, some prior knowledge from the experience in modeling influence diagrams can be utilized to resolve other similar decision problems. The basic idea of case-based reasoning is that humans reuse the problem-solving experience to solve new problems.In this paper, we suggest case-based decision class analysis (CB-DCA), a methodology based on case-based reasoning, to build an influence diagram. CB-DCA is composed of a case retrieval procedure and an adaptation procedure. Two measures are suggested for the retrieval procedure, one a fitting ratio and the other a garbage ratio. The adaptation procedure is based on decision-analytic knowledge and decision participants' domain-specific knowledge. Our proposed methodology has been applied to an environmental review process in which decision-makers need decision models to decide whether a project proposal is accepted or not. Experimental results show that our methodology for decision class analysis provides decision-makers with robust knowledge-based support.

Notes: We report on work concerning the use of object-oriented analysis and design (OAD) methods in the development of artificial intelligence (AI) software applications, in which we compare such techniques to software development methods more commonly used in AI, in particular CommonKADS. As a contribution to clarifying the role of OAD methods in AI, in this paper we compare the analysis models of the object-oriented methods and the CommonKADS high-level expertise model. In particular, we study the correspondences between generic tasks, methods and ontologies in methodologies such as CommonKADS and analysis patterns in object-oriented analysis. Our aim in carrying out this study is to explore to what extent, in areas of AI where the object-oriented paradigm may be the most adequate way of conceiving applications, an analysis level ‘pattern language’ could play the role of the libraries of generic knowledge models in the more commonly used AI software development methods. As a case study we use the decision task — its importance arising from its status as the basic task of the intelligent agent — and the associated heuristic multi-attribute decision method, for which we derive a corresponding decision pattern described in the unified modelling language, a de facto standard in OAD.

Notes: Maintenance issues comprise an important, though frequently overlooked, area of manufacturing operations. In smaller companies in particular, expertise for effective maintenance project planning may be lacking. In this paper we discuss the development of an object-oriented expert system to support such project planning. The system was developed to meet the needs of an automotive components manufacturer, and focuses solely on its material-handling-oriented projects. The microcomputer-based expert system considers issues of schedule, inventory availability and cost.

Notes: This paper describes the development and implementation of a real-time knowledge-based system (RTKBS) carried out in collaboration with Eli Lilly in the UK. The RTKBS was developed in order to provide consistent advice to the operators of a large-scale antibiotic production plant in real time. The ultimate objective was to reduce variability and improve the process yield. A wealth of process information exists in the experience of the process operators, engineers and scientists, and it was capitalized upon in this application. However, this information tends primarily to be qualitative in nature and its elicitation traditionally represents a bottleneck in developing an RTKBS. The paper demonstrates that this type of process knowledge can be effectively and rapidly captured and then coded within an RTKBS. The project was undertaken in two stages. During the knowledge elicitation stage the specific application area was identified in consultation with the industrial partner and qualitative knowledge in the selected area was extracted from experts/operators using KAT™ (a structured technique for knowledge elicitation). Subsequently this knowledge was converted into a rule base and implemented within G2, which then acted as an advisory/decision support system.

Notes: Search is a fundamental problem-solving method in artificial intelligence. Traditional off-line search algorithms attempt to find an optimal solution whereas real-time search algorithms try to find a suboptimal solution more quickly than traditional algorithms to meet real-time constraints. In this work, a new multi-agent real-time search algorithm is developed and its effectiveness is illustrated on a sample domain, namely maze problems. Searching agents can see their environment with a specified visual depth and hence can partially observe their environment. An agent makes use of its partial observation to select a next move, instead of using only one-move-ahead information. Furthermore agents cooperate through a marking mechanism to be able to search different parts of the search space. When an agent selects its next move, it marks its direction of move before executing the move. When another agent comes to this position, it sees this mark and, if possible, moves in a different direction than the previously selected direction. In this way, marking helps agents coordinate their moves with other agents. Although coordination brings an overhead, from experiments we observe that this mechanism is effective in both search time and solution length in maze problems.

Notes: Sequential decision models are an important component of expert systems since, in general, the cost of acquiring information is significant and there is a trade-off between the cost and the value of information. Many expert systems in various domains (business, engineering, medicine etc.), needing costly inputs that are not known until the system operates, have to face this problem. In the last decade the field of sequential decision models based on decision theory (sequential decision-theoretic models) have become more and more important due to both the continuous progress made by research in Bayesian networks and the availability of modern powerful tools for building Bayesian networks and for probability propagation. This paper provides readers (especially knowledge engineers and expert system designers) with a unified and integrated presentation of the disparate literature in the field of sequential decision-making based on decision theory, in order to improve comprehensibility and accessibility. Besides the presentation of the general theory, a view of sequential diagnosis as an instance of the general concept of sequential decision-theoretic models is also shown.

Notes: The present work describes an operational knowledge-based system able to reason with uncertain knowledge, by using methodologies and formalisms that provide for representation of facts and rules with various degrees of certainty and precision. The user interface is windows based, and its design allows high flexibility both in describing the facts in the knowledge base and in adapting the reasoning mechanism. It can therefore be used to evaluate user-generated hypotheses on plant disorders and diagnostic scenarios, as well as provide an environment for education and training in phytopathology.

Notes: Book reviewed:Nils J. Nilsson, Artificial Intelligence: A New Synthesis

Notes: A fuzzy torque distribution controller for energy management (and emission control) of a parallel hybrid electric vehicle is proposed. The proposed controller is implemented in terms of a hierarchical architecture which incorporates the mode of operation of the vehicle as well as empirical knowledge of energy flow in each mode. Moreover, the rule set for each mode of operation of the vehicle is designed in view of an overall energy management strategy that ranges from maximal emphasis on battery charge sustenance to complete reliance on the electrical power source. The proposed control system is evaluated via computational simulations under the FTP75 urban drive cycle. Simulation results reveal that the proposed fuzzy torque distribution strategy is effective over the entire operating range of the vehicle in terms of performance, fuel economy and emissions.

Notes: This paper presents the design, tuning and performance analysis of a new predictive fuzzy controller structure for higher order plants with large time delays. The designed controller consists of a fuzzy proportional-integral (PI) part and a fuzzy predictor. The fuzzy predictive PI controller combines the advantages of fuzzy control while maintaining the simplicity and robustness of a conventional PI controller. The dynamics of the prediction term are adaptive to the system’s time delay. The prediction term has two parts: a fuzzy predictor that uses the system time delay as an input for calculating the prediction horizon and an exponential term that uses the prediction horizon as its positive power. The prediction term also introduces phase lead into the system which compensates for the phase lag due to the time delay in the plant, thereby stabilizing the closed-loop configuration. The performance of the proposed controller is compared with the responses of the conventional predictive PI controller, showing many advantages of the new design over its conventional counterpart.

Notes: The development of online drug administration strategies in operating theatres represents a highly safety-critical situation. The usefulness of different levels of simulation prior to clinical trials has been shown in previous studies in muscle relaxant anaesthesia. Thus, in earlier work on predictive self-tuning control for muscle relaxation a dual computer real-time simulation was undertaken, subsequent to algorithm validation via off-line simulation. In the present approach a supervised rule-based control algorithm is used. The control software was implemented on the actual machine to be used in theatre, while another computer acted as a real-time patient simulator. This set-up has further advantages of providing accurate timing and also finite data accuracy via the ADC/DAC interface, or the equivalent digital lines. Also, it provides for controller design fast simulation studies compared to the real-time application. In this paper, a new architecture which combines several hierarchical levels for control (a Mamdani-type fuzzy controller), adaptation (self-organizing fuzzy logic control) and performance monitoring (fault detection, isolation and accommodation) is developed and applied to a computer real-time simulation platform for muscle relaxant anaesthesia. Experimental results showed that the proposed algorithm fulfilled successfully the requirements for autonomy, i.e. automatic control, adaptation and supervision, and proved effective in dealing with the faults and disturbances which are normally encountered in operating theatres during surgery.

Notes: Fuzzy controller design includes both linear and non-linear dynamic analysis. The knowledge base parameters associated within the fuzzy rule base influence the non-linear control dynamics while the linear parameters associated within the fuzzy output signal influence the overall control dynamics. For distinct identification of tuning levels, an equivalent linear controller output and a normalized non-linear controller output are defined. A linear proportional-integral-derivative (PID) controller analogy is used for determining the linear tuning parameters. Non-linear tuning is derived from the locally defined control properties in the non-linear fuzzy output. The non-linearity in the fuzzy output is then represented in a graphical form for achieving the necessary non-linear tuning. Three different tuning strategies are evaluated. The first strategy uses a genetic algorithm to simultaneously tune both linear and non-linear parameters. In the second strategy the non-linear parameters are initially selected on the basis of some desired non-linear control characteristics and the linear tuning is then performed using a trial and error approach. In the third method the linear tuning is initially performed off-line using an existing linear PID law and an adaptive non-linear tuning is then performed online in a hierarchical fashion. The control performance of each design is compared against its corresponding linear PID system. The controllers based on the first two design methods show superior performance when they are implemented on the estimated process system. However, in the presence of process uncertainties and external disturbances these controllers fail to perform any better than linear controllers. In the hierarchical control architecture, the non-linear fuzzy control method adapts to process uncertainties and disturbances to produce superior performance.

Notes: The traditional unit of analysis in knowledge acquisition and requirements acquisition is the method, e.g. the interview, the repertory grid etc. There are practical and theoretical difficulties with this division. A more useful concept is the method fragment, described in this paper. The method fragment is a discrete component of a method which can be used in one or more methods. The discreteness of method fragments makes it much easier to describe their strengths and weaknesses than is the case with methods, enabling the elicitor to put together a customized method to suit a particular elicitation need. In this paper we outline the concept of method fragments using worked examples, discuss issues raised and make recommendations for further work.

Notes: This paper presents the development of a knowledge-based system for the evaluation of industrial equipment in terms of preventive maintenance criticality. A deterministic rating methodology based on the principles of reliability centered maintenance constitutes the logic of the analytical model of the system. The final output generated by the expert system consists of a listing of the pieces of equipment that should receive special consideration for maintenance purposes, along with a description of the possible failure modes arranged by machine component. The system was developed and delivered on an IBM-compatible personal computer through an object-oriented computer shell.

Notes: Many practitioners and theoreticians feel uncomfortable in using classic self-report, knowledge acquisition methods for assessing an expert’s feelings of confidence or certainty about rules, choices or decisions. This paper proposes and tests an alternative approach using response latency, or the time it takes an expert to respond to a paired choice inquiry. The question is to learn if there is any merit in considering the response latency approach as a replacement for the classic method. This investigation is concerned with using the analytic hierarchy process (AHP) to compare two methods for estimating a subject’s feeling of certainty or confidence about their rule, choice or decision. One method is the older or classic method where a subject self-reports both the choice selected and the level of certainty associated with a choice-pair on the 1-to-9 scale used with the AHP. The second is proposed by the authors and based on using only the choice and decision time or response latency. In the second case, the inverse of response latency is used to estimate a subject’s degree of certainty or confidence in the selected choice in a paired comparison experiment. The response latency method has the advantage of being unobtrusive, less prone to conscious censure, quicker to perform, requiring less effort, and possibly being less expensive to administer. Employing AHP methodology, priorities are used to represent the degree of confidence among choices or rules, and consistency ratios are used to estimate the degree of consistency detected by each method. To compare the two methods, 21 subjects were evaluated in a paired sample design. Employing a computer-administered questionnaire, subjects expressed relative levels of confidence for a set of potential recommendations and the time for each subject to respond was recorded unobtrusively. Using paired sample t tests, sufficient evidence was found to reject the null hypothesis (α= 0.025) that the mean of the difference in the priorities between the two methods was zero (p 〈 0.0001). In a second test, there was sufficient evidence to be at least 95% confident of accepting an alternative hypothesis that response latency produces a lower consistency ratio (higher consistency of the reports) than the self-report method (p 〈 0.0001). In addition to the tests of significance, the results were found to be practically significant and are discussed in the paper.

Notes: The generically used expert scheduling system (GUESS) intelligent scheduling toolkit has been built and applied in various scheduling domains. Previously, GUESS included three major scheduling approaches: a heuristic-based (suggestion tabulator) approach, a hill-climbing algorithm, and a genetic algorithm approach. GUESS has now been expanded to include a neural network scheduling approach. This paper discusses the development, implementation and testing results of the neural network scheduling method within GUESS.

Notes: One of the major challenges for electronic commerce is how to establish a relationship of trust between different parties. Establishing trust is nontrivial, because the traditional physical or social means of trust cannot apply directly in virtual settings. In many cases, the parties involved may not ever have interacted before. Reputation systems seek to address the development of trust by recording the reputations of different parties. However, most existing reputation systems are restricted to individual market websites. Further, relevant information about a party may come from several websites and from interactions that were not mediated by any website.This paper considers the problem of automatically collecting ratings about a given party from others. Our approach involves a distributed agent architecture and adapts the mathematical theory of evidence to represent and propagate the ratings that participants give to each other. When evaluating the trustworthiness of a given party, a peer combines its local evidence (based on direct prior interactions with the party) with the testimonies of others regarding the same party. This approach satisfies certain important properties of distributed reputation management and is experimentally evaluated through simulations.

Notes: We envision a future economy where e–markets will play an essential role as exchange hubs for commodities and services. Future e–markets should be designed to be robust to manipulation, flexible, and sufficiently efficient in facilitating exchanges. One of the most important aspects of designing an e–market is market mechanism design. A market mechanism defines the organization, information exchange process, trading procedure, and clearance rules of a market. If we view an e–market as a multi–agent system, the market mechanism also defines the structure and rules of the environment in which agents (buyers and sellers) play the market game. We design an e–market mechanism that is strategy–proof with respect to reservation price, weakly budget–balanced, and individually rational. Our mechanism also makes sellers unlikely to underreport the supply volume to drive up the market price. In addition, by bounding our market’s efficiency loss, we provide fairly unrestrictive sufficient conditions for the efficiency of our mechanism to converge in a strong sense when (1) the number of agents who successfully trade is large, or (2) the number of agents, trading and not, is large. We implement our design using the RETSINA infrastructure, a multi–agent system development toolkit. This enables us to validate our analytically derived bounds by numerically testing our e–market.

Notes: We propose an alternative method of machine–aided translation: Structure–Based Machine Translation (SBMT). SBMT uses language structure matching techniques to reduce complicated grammar rules and provide efficient and feasible translation results. SBMT comprises the following four features: (1) source language input sentence analysis; (2) source language sentence transformation into target language structure; (3) dictionary lookup; and (4) semantic disambiguation or word sense disambiguation (WSD) for correct output selection. SBMT has been designed and a prototype system has been implemented that generates satisfactory translations.

Notes: In this article we present the design of an ACL for a dynamic system of agents. The ACL includes a set of conversation performatives extended with operations to register, create, and terminate agents. The main design goal at the agent–level is to provide only knowledge–level primitives that are well integrated with the dynamic nature of the system. This goal has been achieved by defining an anonymous interaction protocol which enables agents to request and supply knowledge without considering symbol–level issues concerning management of agent names, routing, and agent reachability. This anonymous interaction protocol exploits a distributed facilitator schema which is hidden at the agent–level and provides mechanisms for registering capabilities of agents and delivering requests according to the competence of agents. We present a formal specification of the ACL and of the underlying architecture, exploiting an algebra of actors, and illustrate it with the help of a graphical notation. This approach provides the basis for discussing dynamic primitives in ACL and for studying properties of dynamic multi agent systems, for example concerning the behavior of agents and the correctness of their conversation policies.

Notes: In this paper, we propose a reputation–oriented reinforcement learning algorithm for buying and selling agents in electronic market environments. We take into account the fact that multiple selling agents may offer the same good with different qualities. In our approach, buying agents learn to avoid the risk of purchasing low–quality goods and to maximize their expected value of goods by dynamically maintaining sets of reputable sellers. Selling agents learn to maximize their expected profits by adjusting product prices and by optionally altering the quality of their goods. Modeling the reputation of sellers allows buying agents to focus on those sellers with whom a certain degree of trust has been established. We also include the ability for buying agents to optionally explore the marketplace in order to discover new reputable sellers. As detailed in the paper, we believe that our proposed strategy leads to improved satisfaction for buyers and sellers, reduced communication load, and robust systems. In addition, we present preliminary experimental results that confirm some potential advantages of the proposed algorithm, and outline planned future experimentation to continue the evaluation of the model.

Notes: This paper presents eMediator, an electronic commerce server prototype that demonstrates ways in which algorithmic support and game–theoretic incentive engineering can jointly improve the efficiency of e–commerce. eAuctionHouse, the configurable auction server, includes a variety of generalized combinatorial auctions and exchanges, pricing schemes, bidding languages, mobile agents, and user support for choosing an auction type. We introduce two new logical bidding languages for combinatorial markets: the XOR bidding language and the OR–of–XORs bidding language. Unlike the traditional OR bidding language, these are fully expressive. They therefore enable the use of the Clarke–Groves pricing mechanism for motivating the bidders to bid truthfully. eAuctionHouse also supports supply/demand curve bidding. eCommitter, the leveled commitment contract optimizer, determines the optimal contract price and decommitting penalties for a variety of leveled commitment contracting mechanisms, taking into account that rational agents will decommit strategically in Nash equilibrium. It also determines the optimal decommitting strategies for any given leveled commitment contract. eExchangeHouse, the safe exchange planner, enables unenforced anonymous exchanges by dividing the exchange into chunks and sequencing those chunks to be delivered safely in alternation between the buyer and the seller.

Notes: Our approach for automating the negotiation of business contracts proceeds in three broad steps. First, determine the structure of the negotiation process by applying general knowledge about auctions and domain–specific knowledge about the contract subject along with preferences from potential buyers and sellers. Second, translate the determined negotiation structure into an operational specification for an auction platform. Third, after the negotiation has completed, map the negotiation results to a final contract. We have implemented a prototype which supports these steps by employing a declarative specification (in courteous logic programs) of (1) high–level knowledge about alternative negotiation structures, (2) general–case rules about auction parameters, (3) rules to map the auction parameters to a specific auction platform, and (4) special–case rules for subject domains. We demonstrate the flexibility of this approach by automatically generating several alternative negotiation structures for the domain of travel shopping in a trading agent competition.

Notes: We propose that through the formalization of concepts related to trust, a more accurate model of trust can be implemented. This paper presents a new model of trust that is based on the formalization of reputation. A multidisciplinary approach is taken to understanding the nature of trust and its relation to reputation. Through this approach, a practical definition of reputation is adopted from sociological contexts and a model of reputation is designed and presented.Reputation is defined as role fulfillment. To formalize reputation, it is necessary to formalize the expectations placed upon an agent within a particular multi–agent system (MAS). In this case, the agents are part of an information–sharing society. Five roles are defined along with the ways in which these roles are objectively fulfilled. Through the measurement of role fulfillment, a vector representing reputation can be developed. This vector embodies the magnitude of the reputation and describes the patterns of behavior associated with the direction of the vector.Experiments are conducted to verify the sensibility of the proposed models for role fulfillment and overall reputation. The simulation results show that the roles, defined for building reputation in an information–sharing MAS environment, react to different agent and user actions in a manner consistent with the formal definitions.

Notes: This article presents a formal framework and outlines a method that autonomous agents can use to negotiate the semantics of their communication language at run–time. Such an ability is needed in open multi–agent systems so that agents can ensure they understand the implications of the utterances that are being made and so that they can tailor the meaning of the primitives to best fit their prevailing circumstances. To this end, the semantic space framework provides a systematic means of classifying the primitives along multiple relevant dimensions. This classification can then be used by the agents to structure their negotiation (or semantic fixing) process so that they converge to the mutually agreeable semantics that are necessary for coherent social interactions.

Notes: In this paper, we discuss how recent theoretical linguistic research focusing on the Minimalist Program (MP) (Cho95, Mar95, Zwa94) can be used to guide the parsing of a useful range of natural language sentences and the building of a logical representation in a principles-based manner. We discuss the components of the MP and give an example derivation. We then propose parsing algorithms that recreate the derivation structure starting with a lexicon and the surface form of a sentence. Given the approximated derivation structure, MP principles are applied to generate a logical form, which leads to linguistically based algorithms for determining possible meanings for sentences that are ambiguous due to quantifier scope.

Notes: This paper addresses the problem of granularity in temporal representation in the context of text analysis. In contrast with other fields where granularity levels are essentially quantitative, in natural language the different levels are not always precisely defined and granularity is of a more subtle and qualitative nature. A model is proposed for representing such phenomena, based on time units and time units intervals. Time units represent chunks of time which are considered indivisible at a given granularity level, but which may be refined and contain other time units or intervals at a higher granularity level. The structure of the set of time units and intervals is studied and a relation algebra is defined which extends the traditional Allen's Point or Interval Algebra. Weak and strong representations of this algebra are proposed. We claim that such representations, called grained temporal structures, are adequate for coping with dynamic qualitative changes of granularity. A logic with restricted quantifiers is proposed for formalizing temporal knowledge and examples are discussed which show the relevance of the model for natural language analysis.

Notes: Effective manipulation of temporal information about periodic events is required for solving complex problems such as long-range scheduling or querying temporal information. Furthermore, many problems involving repeating events involve the optimization of temporal aspects of these events (e.g., minimizing make-span in job-shop scheduling). In this paper, a constraint-based formulation of reasoning problems with repeating events is presented, and its complexity is analyzed for a range of problems. Optimization constraints are interpreted formally using the Semiring CSPs (SCSP) representation of optimization in constraint reasoning. This allows for familiar algorithms such as branch-and-bound to be applied to solving them.

Notes: The measurement of information is potentially as important to an information engineer as the measurement of physical quantities is to a civil or mechanical engineer. This article introduces semantic measures, representing the distance between the meanings of two messages. We demonstrate one possible application by giving a small-scale example in which a semantic measure was used to retrieve information by meaning rather than by word-occurrence. The distance function between the meanings of two messages can be generalized to cover fuzzy meanings. A possible application, the processing of free responses to opinion polls, is described.

Notes: Kowalski and Sergot's Event Calculus (EC) is a simple temporal formalism that, given a set of event occurrences, derives the maximal validity intervals (MVIs) over which properties initiated or terminated by these events hold. In this paper, we conduct a systematic analysis of EC by which we gain a better understanding of this formalism and determine ways of augmenting its expressive power. The keystone of this endeavor is the definition of an extendible formal specification of its functionalities. This formalization has the effects of casting determination of MVIs as a model checking problem, of setting the ground for studying and comparing the expressiveness and complexity of various extensions of EC, and of establishing a semantic reference against which to verify the soundness and completeness of implementations.We extend the range of queries accepted by EC, which is limited to Boolean combinations of MVI verification or computation requests, to support arbitrary quantification over events and modal queries. We also admit specifications based on preconditions. We demonstrate the added expressive power by encoding a number of diagnosis problems. Moreover, we provide a systematic comparison of the expressiveness and complexity of the various extended event calculi against each other. Finally, we propose a declarative encoding of these enriched event calculi in the logic programming language λProlog and prove the soundness and completeness of the resulting logic programs.

Notes: The basic nearest-neighbor rule generalizes well in many domains but has several shortcomings, including inappropriate distance functions, large storage requirements, slow execution time, sensitivity to noise, and an inability to adjust its decision boundaries after storing the training data. This paper proposes methods for overcoming each of these weaknesses and combines the methods into a comprehensive learning system called the Integrated Decremental Instance-Based Learning Algorithm (IDIBL) that seeks to reduce storage, improve execution speed, and increase generalization accuracy, when compared to the basic nearest neighbor algorithm and other learning models. IDIBL tunes its own parameters using a new measure of fitness that combines confidence and cross-validation accuracy in order to avoid discretization problems with more traditional leave-one-out cross-validation. In our experiments IDIBL achieves higher generalization accuracy than other less comprehensive instance-based learning algorithms, while requiring less than one-fourth the storage of the nearest neighbor algorithm and improving execution speed by a corresponding factor. In experiments on twenty-one data sets, IDIBL also achieves higher generalization accuracy than that reported for sixteen major machine learning and neural network models.

Notes: We address long–term coalitions that are formed of both customer and vendor agents. We present a coalition formation mechanism designed at the agent level as a decision problem. The proposed mechanism is analyzed at both system and agent levels. Our results show that the coalition formation mechanism is beneficial for both the system—it reaches an equilibrium state—and for the agents—their gains highly increase over time.

Notes: Two similar multi–agent systems have been designed to address the issue of information sharing within a multi–agent system. This paper examines the architectural components that have been added to our information–sharing societies, ACORN and MP3. Through this exploration, we conclude that these components and their underlying concepts can be added to other information–retrieval societies.ACORN consists of a set of information–sharing locations referred to as cafés. Cafés are defined as meeting locations for like–minded agents. Like–minded agents are defined as agents that share a common set of interests. As an example, a café may contain agents that are interested in information relating to cars. A dynamic café clustering method is developed. The performance evaluation of the proposed structure for the café is presented. The concept of a fat/thin agent architecture is introduced. This agent architecture allows for minimizing network traffic as agents traverse the network in search of or distribution of knowledge. The directory server component is presented along with its relation to the fat/thin agent architecture. Finally, an anonymity service provider which allows anonymity for users is introduced.The MP3 society exists with the sole purpose of finding MP3s throughout a given network. Through this society, the core design issues of agent verification and agent validation are addressed and solutions are presented through respective interface components.

Notes: We propose Generate and Repair Machine Translation (GRMT), a constraint–based approach to machine translation that focuses on accurate translation output. GRMT performs the translation by generating a Translation Candidate (TC), verifying the syntax and semantics of the TC and repairing the TC when required. GRMT comprises three modules: Analysis Lite Machine Translation (ALMT), Translation Candidate Evaluation (TCE) and Repair and Iterate (RI). The key features of GRMT are simplicity, modularity, extendibility, and multilinguality.An English–Thai translation system has been implemented to illustrate the performance of GRMT. The system has been developed and run under SWI–Prolog 3.2.8. The English and Thai grammars have been developed based on Head–Driven Phrase Structure Grammar (HPSG) and implemented on the Attribute Logic Engine (ALE). GRMT was tested to generate the translations for a number of sentences/phrases. Examples are provided throughout the article to illustrate how GRMT performs the translation process.

Notes: Agent technology is an exciting and important new way to create complex software systems. Agents blend many of the traditional properties of AI programs—knowledge–level reasoning, flexibility, proactiveness, goal–directedness, and so forth—with insights gained from distributed software engineering, machine learning, negotiation and teamwork theory, and the social sciences. An important part of the agent approach is the principle that agents (like humans) can function more effectively in groups that are characterized by cooperation and division of labor. Agent programs are designed to autonomously collaborate with each other in order to satisfy both their internal goals and the shared external demands generated by virtue of their participation in agent societies. This type of collaboration depends on a sophisticated system of inter–agent communication. The assumption that inter–agent communication is best handled through the explicit use of an agent communication language (ACL) underlies each of the articles in this special issue. In this introductory article, we will supply a brief background and introduction to the main topics in agent communication.

Notes: Conversation protocols are used to achieve certain goals or to bring about certain states in the world. Therefore, one may identify the landmarks or the states that must be brought about during the goal–directed execution of a protocol. Accordingly, the landmarks, characterized by propositions that are true in the state represented by that landmark, are the most important aspect of a protocol. Families of conversation protocols can be expressed formally as partially ordered landmarks after the landmarks necessary to achieve a goal have been identified. Concrete protocols represented as joint action expressions can, then, be derived from the partially ordered landmarks and executed directly by joint intention interpreters. This approach of applying Joint Intention theory to protocols also supports flexibility in the actions used to get to landmarks, shortcutting protocol execution, automatic exception handling, and correctness criterion for protocols and protocol compositions.

Notes: This paper is a response to commentaries on my target paper for Computational Intelligence, “Control and Cross-Domain Mental Computation: Evidence from Language Breakdown.” In this response, I acknowledge certain errors in my initial construal of control and dismiss unwarranted criticisms. I then reexamine both control and certain language disorders in light of the explicitness of cross-domain communication and the visibility of representations to each other. In the end, I reassert the validity of the logic/control (visibility) distinction in mental computation and argue that the contrasts between Specific Language Impairment and Williams syndrome parallel this distinction.

Notes: A complete analysis of an English sentence includes syntactic, semantic, and pragmatic components. Presupposition belongs to the pragmatic component. How to determine the presuppositions of multiple-clause sentences has been the focus of much work. Projection of clausal presuppositions is one method to determine the presuppositions of multiple-clause sentences. In this paper we present a new approach to the projection problem. Drawing heavily on the theoretical techniques originating with Montague semantics, our system maps sentences of a category-based grammar into a set of expressions of intensional logic: one expression corresponding to the literal interpretation of the sentence and the remaining expressions corresponding to the presuppositions of the sentence. The new approach correctly predicts the presuppositions of a larger range of multiple-clause sentences than previous projection approaches.

Notes: This paper describes a system for the summarization of multiple text-only news-like documents. We address two main issues: clustering of documents in order to find the main topics that should be mentioned in the multidocument summary and organization of the information in order to create a summary that presents the information in a logical way and is easy to read. The system is based on a single-document summarizer that uses text-extraction.

Notes: This work examines important issues in probabilistic temporal representation and reasoning using Bayesian networks (also known as belief networks). The representation proposed here utilizes temporal (or dynamic) probabilities to represent facts, events, and the effects of events. The architecture of a belief network may change with time to indicate a different causal context. Probability variations with time capture temporal properties such as persistence and causation. They also capture event interaction, and when the interaction between events follows known models such as the competing risks model, the additive model, or the dominating event model, the net effect of many interacting events on the temporal probabilities can be calculated efficiently. This representation of reasoning also exploits the notion of temporal degeneration of relevance due to information obsolescence to improve the efficiency.