Applications and Innovations in Intelligent Systems XIII

Knowledge engineers have been working in the legal domain since the rise of their discipline in the mid-eighties of the last century. Traditionally their main focus was capturing and distributing knowledge by means of the knowledge-based systems, thus improving legal access. More and more legal knowledge engineering has become an analytical approach that helps to improve legal quality. An example is the POWER-approach developed in a research programme that is now finished. This programme was run by the Dutch Tax and Customs Administration (DTCA in Dutch: Belastingdienst) and some partners (see e.g. Van Engers et al., 1999, 2000, 2001, 2003 and 2004). The POWER-approach helped to improve quality of (new) legislation and codify the knowledge used in the translation processes in which legislation and regulations are transformed into procedures, computer programs and other designs. We experienced that despite these clear benefits implementation proved to be far from easy. In fact the implementation phase still continues. Adapting research results in public administrations is a tedious process that takes lots and lots of energy and requires continuous management attention. Learning at organisational level proved to be much harder than we thought.

In this paper, we present ISOR, a Case-Based Reasoning system for long-term therapy support in the endocrine domain and in psychiatry. ISOR performs typical therapeutic tasks, such as computing initial therapies, initial dose recommendations, and dose updates. Apart from these tasks ISOR deals especially with situations where therapies become ineffective. Causes for inefficacy have to be found and better therapy recommendations should be computed. In addition to the typical Case-Based Reasoning knowledge, namely former already solved cases, ISOR uses further knowledge forms, especially medical histories of query patients themselves and prototypical cases (prototypes). Furthermore, the knowledge base consists of therapies, conflicts, instructions etc. So, retrieval does not only provide former similar cases but different forms and steps of retrieval are performed, while adaptation occurs as an interactive dialog with the user. Since therapy inefficacy can be caused by various circumstances, we propose searching for former similar cases to get ideas about probable reasons that subsequently should be carefully investigated. We show that ISOR is able to successfully support such investigations.

We apply a variant of Simulated Annealing (SA) as a standard black-box optimisation algorithm to the colour quantisation problem. The main advantage of black-box optimisation algorithms is that they do not require any domain specific knowledge yet are able to provide a near optimal solution. To further improve the performance of the algorithm we combine the SA technique with a standard k-means clustering technique. We evaluate the effectiveness of our approach by comparing its performance with several specialised colour quantisation algorithms. The results obtained show that our hybrid SA algorithm clearly outperforms standard quantisation algorithms and provides images with superior image quality.

The concept of the Knowledge Bazaar as a paradigm for the development of Expert Systems, whereby knowledge bases are created dynamically using knowledge supplied by self appointed Internet communities is proposed. The idea espouses the creation of individual Knowledge Bazaars, operating in specific domains, but all operating through a generic Knowledge Bazaar XML Web application. Issues addressed include the provision of the service, XML rule representations and rule integrity. The concept is illustrated with a demonstration gardening Knowledge Bazaar that is currently operational.

SkillSum is an Artificial Intelligence (AI) and Natural Language Generation (NLG) system that produces short feedback reports for people who are taking online tests which check their basic literacy and numeracy skills. In this paper, we describe the SkillSum system and application, focusing on three challenges which we believe are important ones for many systems which try to generate feedback reports from Web-based tests: choosing content based on very limited data, generating appropriate texts for people with varied levels of literacy and knowledge, and integrating the web-based system with existing assessment and support procedures.

The aim of this study is to evaluate the effectiveness of using external indicators, such as commodity prices and currency exchange rates, in predicting movements in the Dow Jones Industrial Average index. The performance of each technique is evaluated using different domain specific metrics. A comprehensive evaluation procedure is described, involving the use of trading simulations to assess the practical value of predictive models, and comparison with simple benchmarks that respond to underlying market growth. In the experiments presented here, basing trading decisions on a neural network trained on a range of external indicators resulted in a return on investment of 23.5% per annum, during a period when the DJIA index grew by 13.03% per annum. A substantial dataset has been compiled and is available to other researchers interested in analysing financial time series.

Engineering design requires extensive decomposition and integration activities relying on a multidisciplinary basis. A design structure matrix (DSM) can be used as a representation and analysis tool in order to manage the design process under diverse perspectives. The design outcome is always subject to the abstract nature, the subjectivity and the low availability of the required design knowledge. This paper addresses the DSM as a communicating design tool among multiple designers. A fuzzy-logical inference mechanism permits the collaboration among designers on the qualitative definition of the interrelations among the design problem’s entities or tasks and the resulting DSM may be then utilized for various tasks (partitioning, clustering, tearing etc.) depending on the problem under consideration. A DSM is deployed for the case of the parametric design of an oscillating conveyor where two (2) designers are collaboratively involved.

This paper details the application of geometric proportional analogies in the sub-classification of polygons within a topographic (land cover) map. The first part of this paper concerns geometric proportional analogies that include attributes (e.g. fill-pattern and fill-colour). We describe an extension to the standard theory of analogy that incorporates attributes into the analogical mapping process. We identify two variants on this “attribute matching” extension, which is required to solve different types of geometric proportional analogy problems. In the second part of this paper we describe how we use the simpler of these algorithms to generate inferences in topographic maps. We detail the results of identifying a number of different structures on a sample topographic map.

Ripple-Down Rules (RDR) is an approach to building knowledge-based systems (KBS) incrementally, while the KBS is in routine use. Domain experts build rules as a minor extension to their normal duties, and are able to keep refining rules as KBS requirements evolve. Commercial RDR systems are now used routinely in some Chemical Pathology laboratories to provide interpretative comments to assist clinicians make the best use of laboratory reports. This paper presents usage data from one laboratory where, over a 29 month period, over 16,000 rules were added and 6,000,000 cases interpreted. The clearest evidence that this facility is highly valuable to the laboratory is the on-going addition of new knowledge bases and refinement of existing knowledge bases by the chemical pathologists.

Bayesian Networks (BNs) have been recently employed to solve meteorology problems. In this paper, the application of BNs for mining a real-world weather dataset is described. The employed dataset discriminates between “wet fog” instances and “other weather conditions” instances, and it contains many missing data. Therefore, BNs were employed not only for classifying instances, but also for filling missing data. In addition, the Markov Blanket concept was employed to select relevant attributes. The efficacy of BNs to perform the aforementioned tasks was assessed by means of several experiments. In summary, more convincing results were obtained by taking advantage of the fact that BNs can directly (i.e. without data preparation) classify instances containing missing values. In addition, the attributes selected by means of the Markov Blanket provide a simpler, faster, and equally accurate classifier.

WISE Expert is a general-purpose system that can be used for monitoring or controlling, in real time, complex systems that have recurring sub-structures. The system has been developed using a unique schematic development tool that ensures coherency of structure during design and construction. The design of the Expert System takes advantage of a distinction between the monitored system structure and expert knowledge so that the structure description can be used to generate specific rules for the system automatically. The system has been tested as an overseer during the running of trainee mariner exercises with a liquid cargo simulator and is now operational at over 35 customer sites throughout the world.

Objects of interest are represented in the brain simultaneously in different frames of reference. Knowing the positions of one’s head and eyes, for example, one can compute the body-centred position of an object from its perceived coordinates on the retinae. We propose a simple and fully trained attractor network which computes head-centred coordinates given eye position and a perceived retinal object position. We demonstrate this system on artificial data and then apply it within a fully neurally implemented control system which visually guides a simulated robot to a table for grasping an object. The integrated system has as input a primitive visual system with a what-where pathway which localises the target object in the visual field. The coordinate transform network considers the visually perceived object position and the camera pan-tilt angle and computes the target position in a body-centred frame of reference. This position is used by a reinforcement-trained network to dock a simulated PeopleBot robot at a table for reaching the object. Hence, neurally computing coordinate transformations by an attractor network has biological relevance and technical use for this important class of computations.

In this paper, we present an application of Artificial Intelligence to the implementation of Virtual Automobile Manufacturing Enterprise. We provide a multi autonomous agent based framework. Our agent based architecture leads to flexible design of a spectrum of virtual enterprises by distributing computation and by providing a unified interface to data and programs. Autonomous agents are intelligent enough and provide autonomy, simplicity of communication, computation, and a well developed semantics. The steps of design and implementation are discussed in depth, in particular an ontology, the agent model, and interaction pattern between agents are given. We have developed mechanisms for coordination between agents using a language, which we call Virtual Enterprise Modeling Language (VEML). VEML is a dialect of Java and includes Knowledge Query and Manipulation Language (KQML) primitives. We have implemented a multi autonomous agent based system, which we call VE System. VE System provides application programmers with potential to globally develop different kinds of VEs based on their requirements and applications. We provide case study of automobile manufacturing enterprise and demonstrate efficacy of our system by discussing its salient features.

Web-based teaching via Intelligent Tutoring Systems (ITSs) is considered as one of the most successful enterprises in artificial intelligence. Indeed, there is a long list of ITSs that have been tested on humans and have proven to facilitate learning, among which we may find the well-tested and known tutors of algebra, geometry, and computer languages. These ITSs use a variety of computational paradigms, as production systems, Bayesian networks, schema-templates, theorem proving, and explanatory reasoning. The next generation of ITSs are expected to go one step further by adopting not only more intelligent interfaces but will focus on integration. This article will describe some particularities of a tutoring system that we are developing to simulate conversational dialogue in the area of Medicine, that enables the integration of highly heterogeneous sources of information into a coherent knowledge base, either from the tutor’s point of view or the development of the discipline in itself, i.e. the system’s content is created automatically by the physicians as their daily work goes on. This will encourage students to articulate lengthier answers that exhibit deep reasoning, rather than to deliver straight tips of shallow knowledge. The goal is to take advantage of the normal functioning of the health care units to build on the fly a knowledge base of cases and data for teaching and research purposes.

Meridian-collateral knowledge is a profound and complex part of the whole traditional Chinese medicine (TCM). It is the basis for many TCM-related computer applications. This work aimed to develop a sharable knowledge base of the human meridian-collateral system for those applications. We began the work by building a frame ontology of the meridian-collateral system (called OMCAP); and then developed a large-scale sharable instance base (called IMCAP), which was, with the aid of the tool OKEE. The OMCAP consists of 89 categories and 38 slots, and the IMCAP contains 1549 instance frames.

This paper presents the results of an investigation into the use of machine learning methods for the identification of narcotics from Raman spectra. The classification of spectral data and other high dimensional data, such as images, gene-expression data and spectral data, poses an interesting challenge to machine learning, as the presence of high numbers of redundant or highly correlated attributes can seriously degrade classification accuracy. This paper investigates the use of Principal Component Analysis (PCA) to reduce high dimensional spectral data and to improve the predictive performance of some well known machine learning methods. Experiments are carried out on a high dimensional spectral dataset. These experiments employ the NIPALS (Non-Linear Iterative Partial Least Squares) PCA method, a method that has been used in the field of chemometrics for spectral classification, and is a more efficient alternative than the widely used eigenvector decomposition approach. The experiments show that the use of this PCA method can improve the performance of machine learning in the classification of high dimensionsal data.