IRIS Research Publications


2007 2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996 1995 1994 1993 1992 1991 1990 1989 1988 1987

Journal

B. van Gils, H.A. (Erik) Proper, P. van Bommel, and Th.P. van der Weide. Typing and transformational effects in complex information supply. In: International Journal of Cooperative Information Systems, Nr: 2, Vol: 16, Pages: 229-270, June, 2007.

Information plays an increasingly important role in our lives. Often we retrieve this information by querying the web: data resources found on the web may provide the information that were looking for. This implies that the Web may be seen as an information market: authors supply information and searchers may find it. In this article we present a formal framework for the syntactic aspects of the information market. We explore the information landscape using a modeling approach. An important part of this model is a (syntactic) framework for transformations, which allows us to deal with the heterogeneity of the data resources found on the Web. Last but not least we attempt to give an outline how our framework, which in essence focuses on the data resources on the Web, may lead to a better understanding of how information is supplied via the Web. For this we use an example from the field of information retrieval.

[ PDF ] [ Bibtex ]

F. Arbab, F.S. de Boer, M. Bonsangue, and M.M. Lankhorst. Integrating Architectural Models. In: Enterprise Modelling and Information Systems Architectures, Nr: 1, Vol: 2, Pages: 40-57, January, 2007.

The diversity of architectural models in enterprise architecture is a problem for their integration. In this paper we distinguish three kinds of models from each other and their visualization, and we illustrate how the distinctions can be used for model integration within the architectural approach. Symbolic models express properties of architectures of systems, semantic models interpret the symbols of semantic models, and subjective models are purposely abstracted conceptions of a domain. Building on results obtained in the ArchiMate project, we illustrate how symbolic models can be integrated using an architectural language, how integrated models can be updated using the distinction between symbolic models and their visualization, and how semantic models can be integrated using a new kind of enterprise analysis called semantic analysis. We also suggest that subjective models can be integrated using techniques from natural language analysis.

[ PDF ] [ Bibtex ]

A.J. Hommersom, P. Groot, P.J.F. Lucas, M. Balser, and J. Schmitt. Combining Task Execution and Background Knowledge for the Verification of Medical Guidelines. In: Knowledge-Based Systems, Vol: 20, July, 2007.

The use of a medical guideline can be seen as the execution of computational tasks, sequentially or in parallel, in the face of patient data. It has been shown that many of such guidelines can be represented as a `network of tasks`, i.e., as a number of steps that have a specific function or goal. To investigate the quality of such guidelines we propose a formalization of criteria for good practice medicine a guideline should comply to. We use this theory in conjunction with medical background knowledge to verify the quality of a guideline dealing with diabetes mellitus type 2 using the interactive theorem prover KIV. Verification using task execution and background knowledge is a novel approach to quality checking of medical guidelines.

[ PDF ] [ Bibtex ]

S.J. Overbeek. Improving Efficiency and Effectiveness of Knowledge Exchange between Knowledge Workers. In: ICFAI Journal of Knowledge Management, Nr: 1, Vol: 5, Pages: 24-38, January, 2007.

Information technology increasingly influences the way we work and live. Contemporary businesses demonstrate significant concerns on how increasing amounts of available information can be converted into knowledge. An increasing need for new knowledge concerning the development of new services which an organization offers to the customers in order to be competitive in the market is but an example of how important the dissemination of knowledge within organizations is. The growth in the relative size of people working in the knowledge economy stresses these developments. The research discussed in this paper focuses on improving the efficiency and effectiveness of knowledge exchange between knowledge workers by means of automated support so that dissemination of knowledge within organizations improves.

[ PDF ] [ Bibtex ]

C.A. Albers, T. Heskes, and H.J. Kappen. Haplotype inference in general pedigrees using the cluster variation method. In: Genetics, Vol: 177, Pages: 1101-1118, 2007.

We present CVMHAPLO, a probabilistic method for haplotyping in general pedigrees with many markers. CVMHAPLO reconstructs the haplotypes by assigning in every iteration a fixed number of the ordered genotypes with the highest marginal probability, conditioned on the marker data and ordered genotypes assigned in previous iterations. CVMHAPLO makes use of the cluster variation method (CVM) to efficiently estimate the marginal probabilities. We focused on single-nucleotide polymorphism (SNP) markers in the evaluation of our approach. In simulated data sets where exact computation was feasible, we found that the accuracy of CVMHAPLO was high and similar to that of maximum-likelihood methods. In simulated data sets where exact computation of the maximum-likelihood haplotype configuration was not feasible, the accuracy of CVMHAPLO was similar to that of state of the art Markov chain Monte Carlo (MCMC) maximum-likelihood approximations when all ordered genotypes were assigned and higher when only a subset of the ordered genotypes was assigned. CVMHAPLO was faster than the MCMC approach and provided more detailed information about the uncertainty in the inferred haplotypes. We conclude that CVMHAPLO is a practical tool for the inference of haplotypes in large complex pedigrees.

[ PDF ] [ Bibtex ] [ External URL ]

B. Bakker, and T. Heskes. Learning and approximate inference in dynamic hierarchical models. In: Computational Statistics & Data Analysis, Vol: 52, Pages: 821-839, 2007.

A new variant of the dynamic hierarchical model (DHM) that describes a large number of parallel time series is presented. The separate series, which may be interdependent, are modeled through dynamic linear models (DLMs). This interdependence is included in the model through the definition of a `top-level` or `average` DLM. The model features explicit dependences between the latent states of the parallel DLMs and the states of the average model, and thus the many parallel time series are linked to each other. The combination of dependences within each time series and dependences between the different DLMs makes the computation time that is required for exact inference cubic in the number of parallel time series, however, which is unacceptable for practical tasks that involve large numbers of parallel time series. Therefore, two methods for fast, approximate inference are proposed: a variational approximation and a factorial approach. Under these approximations, inference can be performed in linear time, and it still features exact means. Learning is implemented through a maximum likelihood (ML) estimation of the model parameters. This estimation is realized through an expectation maximization (EM) algorithm with approximate inference in the E-step. Examples of learning and forecasting on two data sets show that the addition of direct dependences has a `smoothing` effect on the evolution of the states of the individual time series, and leads to better prediction results. The use of approximate instead of exact inference is further shown not to lead to inferior results on either data set.

[ PDF ] [ Bibtex ] [ External URL ]

P. van Bommel, B. van Gils, H.A. (Erik) Proper, M. van Vliet, and Th.P. van der Weide. Value and the information market. In: Data and Knowledge Engineering, Nr: 1, Vol: 61, Pages: 153-175, April, 2007.

In this paper we explore how (micro)economic theory can be used to analyze and model the exchange of information on the Web. More specifically, we consider searchers for information who engage in transactions on the Web. Searchers will engage in web transactions only if they gain something in such a transaction. To this end we develop a formal model for markets, based on the notions of value and transaction. This model enables us to examine transactions on an information market. In this market we have a dual view on transactions, creating a dichotomy of transactors and transactands.

[ PDF ] [ Bibtex ] [ External URL ]

M.A.J. van Gerven, R. Jurgelenaite, B.G. Taal, T. Heskes, and P.J.F. Lucas. Predicting carcinoid heart disease with the noisy-threshold classifier. In: Artificial Intelligence in Medicine, Vol: 40, Pages: 45-55, May, 2007.

Objective: To predict the development of carcinoid heart disease (CHD), which is a life-threatening complication of certain neuroendocrine tumors. To this end, a novel type of Bayesian classifier, known as the noisy-threshold classifier, is applied. Materials and methods: Fifty-four cases of patients that suffered from a low-grade midgut carcinoid tumor, of which 22 patients developed CHD, were obtained from the Netherlands Cancer Institute (NKI). Eleven attributes that are known at admission have been used to classify whether the patient develops CHD. Classification accuracy and area under the receiver operating characteristics (ROC) curve of the noisy-threshold classifier are compared with those of the naive-Bayes classifier, logistic regression, the decision-tree learning algorithm C4.5, and a decision rule, as formulated by an expert physician. Results: The noisy-threshold classifier showed the best classification accuracy of 72% correctly classified cases, although differences were significant only for logistic regression and C4.5. An area under the ROC curve of 0.66 was attained for the noisy-threshold classifier, and equaled that of the physician’s decision-rule. Conclusions: The noisy-threshold classifier performed favorably to other state-of-the-art classification algorithms, and equally well as a decision-rule that was formulated by the physician. Furthermore, the semantics of the noisy-threshold classifier make it a useful machine learning technique in domains where multiple causes influence a common effect.

[ PDF ] [ Bibtex ] [ External URL ]

F.A. Grootjen, and Th.P. van der Weide. Information Parallax. , Nr: 3, Vol: 1, Pages: 34-55, March, International Journal of Intelligent Information T, Radboud University Nijmegen2007.

To effectively use and exchange information among AI systems, a formal specification of the representation of their shared domain of discourse - called an ontology - is indispensable. In this paper we introduce a special kind of knowledge representation based on a dual view on the universe of discourse and show how it can be used in human activities such as searching, in-depth exploration and browsing. After a formal definition of dualistic ontologies we exemplify this definition with three different (well known) kinds of ontologies, based on the vector model, on formal concept analysis and on fuzzy logic respectively. The vector model leads to concepts derived by latent semantic indexing using the singular value decomposition. Both the set model as the fuzzy set model lead to Formal Concept Analysis, in which the fuzzy set model is equipped with a parameter that controls the fine-graining of the resulting concepts. We discuss the relation between the resulting systems of concepts. Finally, we demonstrate the use of this theory by introducing the dual search engine. We show how this search engine can be employed to support the human activities addressed above.

[ PDF ] [ Bibtex ]

A.J. Hommersom, P. Groot, P.J.F. Lucas, M. Balser, and J. Schmitt. Verification of Medical Guidelines using Background Knowledge in Task Networks. In: IEEE Transactions on Knowledge and Data Engineering, June, 2007.

The application of a medical guideline to the treatment of the patient`s disease can be seen as the execution of tasks, sequentially or in parallel, in the face of patient data. It has been shown that many of such guidelines can be represented as a `network of tasks`, i.e., as a sequence of steps that have a specific function or goal. In this paper a novel methodology for verifying the quality of such guidelines is introduced. To investigate the quality of such guidelines we propose to include medical background knowledge to task networks and to formalise criteria for good practice medicine a guideline should comply to. This framework was successfully applied to a guideline dealing with the management of diabetes mellitus type 2 using the interactive theorem prover KIV.

[ PDF ] [ Bibtex ]

Chapter

P. van Bommel, S.J.B.A. (Stijn) Hoppenbrouwers, H.A. (Erik) Proper, and Th.P. van der Weide. On the Use of Object-Role Modeling For Modeling Active Domains . Advances in Database Research, 2007.

This chapter is about how the Object Role Modeling (ORM) language and approach can be used for integration, at a deep and formal level, of various domain modeling representations and viewpoints, with a focus on the modeling of active domains. The authors argue that ORM is particularly suited for enabling such integration because of its generic conceptual nature, its useful, existing connection with natural language and controlled languages, and its formal rigor. They propose the Logbook Paradigm as an effective perspective in active domains modeling and for the derivation of domain grammars. They show how standard ORM can be extended to an Object Role Calculus (ORC), including temporal concepts and constraints that enable the modeling of active domains. A suggestion for graphical representation is also provided. The authors hope to contribute to integration of domain models and viewpoints in an academic and educational context rather than proposing ORM and ORC as new modeling tools in an industrial setting.

[ Missing PDF ] [ Bibtex ]

Ildiko Flesch, and Peter Lucas. Markov Equivalence in Bayesian Networks. Pages: 3-38, Springer, 2007.

Probabilistic graphical models, such as Bayesian networks, allow representing conditional independence information of random variables. These relations are graphically represented by the presence and absence of arcs and edges between vertices. Probabilistic graphical models are non-unique representations of the independence information of a joint probability distribution. However, the concept of Markov equivalence of probabilistic graphical models is able to offer unique representations, called essential graphs. In this survey paper the theory underlying these concepts is reviewed.

[ PDF ] [ Bibtex ]

Conference

Adriana Birlutiu, and T. Heskes. Expectation Propagation for Rating Players in Sports Competitions. In: 11th European Conference on Principles and Practice of Knowledge Discovery in Databases (PKDD) , 2007.

Rating players in sports competitions based on game results is one example of paired comparison data analysis. Since an exact Bayesian treatment is intractable, several techniques for approximate inference have been proposed in the literature. In this paper we compare several variants of expectation propagation (EP). EP generalizes assumed density filtering (ADF) by iteratively improving the approximations that are made in the filtering step of ADF. Furthermore, we distinguish between two variants of EP: EP-Correlated, which takes into account the correlations between the strengths of the players and EP-Independent, which ignores those correlations. We evaluate the different approaches on a large tennis dataset to find that EP does significantly better than ADF (iterative improvement indeed helps) and EP-Correlated does significantly better than EP-Independent (correlations do matter).

[ PDF ] [ Bibtex ]

P. van Bommel, P. Buitenbuis, S.J.B.A. (Stijn) Hoppenbrouwers, and H.A. (Erik) Proper. Architecture Principles -- A Regulative Perspective on Enterprise Architecture. In: Enterprise Modelling and Information Systems Architectures (EMISA2007), Lecture Notes in Informatics, Vol: 119, Pages: 47-60, Gesellschaft fur Informatik, Bonn, Germany, EU, 2007.

Increasingly, organizations make use of enterprise architectures to direct the development of the enterprise as a whole and its IT portfolio in particular. In this paper we investigate the regulative nature of enterprise architecture. We aim to develop a fundamental understanding of the regulative needs that underly an enterprise architecture, and then take these needs as a starting point to arrive at requirements on the language (architecture principles) used to denote enterprise architectures. We furthermore discuss the process of formulating principles as well as their semantics.

[ PDF ] [ Bibtex ]

Ildiko Flesch, Peter Lucas, and Th.P. van der Weide. Conflict-based Diagnosis: Adding Uncertainty to Model-based Diagnosis. In: International Joint Conference on Artificial Intelligence 2007, Vol: I., Pages: 380-385, January, 2007.

[ PDF ] [ Bibtex ]

Ildiko Flesch, Antonio Fernandez, and Antonio Salmeron. Incremental Supervised Classification for the MTE Distribution: a Preliminary Study. In: Congreso Espanol de Informatica (CEDI-2007), Pages: 217-224, 2007.

In this paper we propose an incremental method for building classifiers in domains with very large amounts of data or for data streams. The method is based on the use of mixtures of truncated exponentials, so that continuous and discrete variables can be handled simultaneously.

[ PDF ] [ Bibtex ]

Ildiko Flesch, and Peter Lucas. Independence Decomposition in Dynamic Bayesian Networks. In: 9th European Conference of Symbolic and Quantitative Approaches to Reasoning with Uncertainty (ECSQARU-2007), Pages: 560-571, Springer, 2007.

Dynamic Bayesian networks are a special type of Bayesian network that explicitly incorporate the dimension of time. They can be distinguished into repetitive and non-repetitive networks. Repetitiveness implies that the set of random variables of the network and their independence relations are the same at each time step. Due to their structural symmetry, repetitive networks are easier to use and are, therefore, often taken as the standard. However, repetitiveness is a very strong assumption, which normally does not hold, as particular dependences and independences may only hold at certain time steps. In this paper, we propose a new framework for independence modularisation in dynamic Bayesian networks. Our theory provides a method for separating atemporal and temporal independence relations, and offers a practical approach to building dynamic Bayesian networks that are possibly non-repetitive. A composition operator for temporal and atemporal independence relations is proposed and its properties are studied. Experimental results obtained by learning dynamic Bayesian networks from real data show that this framework offers a more accurate way for knowledge representation in dynamic Bayesian networks.

[ PDF ] [ Bibtex ]

Ildiko Flesch, Peter Lucas, and Th.P. van der Weide. Probabilistic properties of model-based diagnostic reasoning in Bayesian networks. In: 19th Belgian-Dutch Conference on Artificial Intelligence(BNAIC-2007), Pages: 119-126, 2007.

Much research has been carried out into general properties of Bayesian networks, whereas limited attention has been given to special types of Bayesian networks for specific applications. An example of such an application is model-based diagnosis, i.e. the diagnosis of malfunctioning of devices or systems, based on an explicit model of the structure and behaviour of these devices and systems. Basically, two types of model-based diagnosis are being distinguished: (emph{i}) consistency-based diagnosis, and (emph{ii}) abductive diagnosis. In this paper, we investigate the relationship between consistency-based and abductive reasoning in Bayesian networks. It will appear that abductive diagnoses can be determined using special properties from consistency-based diagnosis, yielding related computationally simplified forms for both probabilistic consistency-based and abductive diagnosis using Bayesian networks. Furthermore, the conceptual relationships between probabilistic diagnostic reasoning and logical diagnostic reasoning are studied.

[ PDF ] [ Bibtex ]

P. Groot. The Role of Model Checking in Critiquing based on Clinical Guidelines. In: 19th Belgian-Dutch Conference on Artificial Intelligence(BNAIC-2007), Pages: 353-354, 2007.

[ PDF ] [ Bibtex ]

P. Groot. Experiences in Quality Checking Medical Guidelines using Formal Methods. In: Proceedings Verification and Validation of Software Systems (VVSS 2007), Pages: 164-178, 2007.

In health care, the trend of evidence-based medicine, has led medical specialists to develop medical guidelines, which are large nontrivial documents suggesting the detailed steps that should be taken by health-care professionals in managing the disease in a patient. In the Protocure project the objective has been to assess the improvement of medical guidelines using formal methods. This paper reports on some of our findings and experiences in quality checking medical guidelines. In particular the formalisation of meta-level quality criteria for good practice medicine, which is used in conjunction with medical background knowledge to verify the quality of a guideline dealing with the management of diabetes mellitus type 2 using the interactive theorem prover KIV. For comparison, analogous investigations have been performed with other techniques including automatic theorem proving and model checking.

[ PDF ] [ Bibtex ]

C. Hesse, R. Oostenveld, T. Heskes, and O. Jensen. On the development of a brain-computer interface system using high-density magnetoencephalogram signals for real-time control of a robot arm. In: Annual Symposium of the IEEE-EMBS Benelux Chapter, 2007.

This work describes a brain-computer interface (BCI) system using multi-channel magnetoencephalogram (MEG) signals for real-time control of a computer game and a robot arm in a motor imagery paradigm. Computationally efficient spatial filtering and time-frequency decomposition facilitate the extraction and classification of neurophysiologically meaningful and task-relevant signal components from all of the 275 channels comprising the high-density sensor array. To our knowledge, this is the first report of an MEG-based BCI system capable of real-time signal processing and control using the whole sensor array. The robust and reliable performance of this system was demonstrated several times in front of a large public audience at an open day celebrating the 5th anniversary of the F.C. Donders Centre for Cognitive Neuroimaging at Radboud University Nijmegen.

[ PDF ] [ Bibtex ]

Koen .V. Hindriks, S.J.B.A. (Stijn) Hoppenbrouwers, Catholijn M. Jonker, and Dmytro Tykhonov. Automatic Issue Extraction from a Focused Dialogue. In: Proceedings of the 12th International Conference on Applications of Natural Language to Information Systems (NLDB`07), Lecture Notes in Computer Science, June, 2007.

Abstract. Various methodologies for structuring the process of domain modeling have been proposed, but there are few software tools that provide automatic support for the process of constructing a domain model. The problem is that it is hard to extract the relevant concepts from natural language texts since these typically include many irrelevant details that are hard to discern from relevant concepts. In this paper, we propose an alternative approach to extract domain models from natural language input. The idea is that more effective, automatic extraction is possible from a natural language text that is produced in a focused dialogue game. We present an application of this idea in the area of pre-negotiation, in combination with sophisticated parsing and transduction techniques for natural language and fairly simple pattern matching rules. Furthermore, a prototype is presented of a conversation-oriented experimentation environment for cooperative conceptualization. Several experiments have been performed to evaluate the approach and environment, and a technique for measuring the quality of extraction has been defined. The experiments indicate that even with a simple implementation of the proposed approach reasonably acceptable results can be obtained

[ PDF ] [ Bibtex ]

The Role of Model Checking in Critiquing based on Clinical Guidelines. , Pages: 411-420, 2007.

Medical critiquing systems criticise clinical actions performed by a physician. In order to provide useful feedback, an important task is to find differences between the actual actions and a set of `ideal` actions as described by a clinical guideline. In case differences exist, insight to which extent they are compatible is provided by the critiquing system. We propose a methodology for such critiquing, where the ideal actions are given by a formal model of a clinical guideline, and where the actual actions are derived from real world patient data. We employ model checking to investigate whether a part of the actual treatment is consistent with the guideline. Furthermore, it is shown how critiquing can be cast in terms of temporal logic, and what can be achieved by using model checking. The methodology has been applied to a clinical guideline of breast cancer in conjunction with breast cancer patient data.

[ PDF ] [ Bibtex ]

R. Jurgelenaite, T. Heskes, and T. Dijkstra. Using symmetric causal independence models to predict gene expression from sequence data. In: ECML-PKDD Workshop “Data Mining in Functional Genomics and Proteomics: Current Trends and Future Directions”, Radboud University Nijmegen, 2007.

We present an approach for inferring transcriptional regulatory modules from genome sequence and gene expression data. Our method, which is based on symmetric causal independence models, is both able to model the logic behind transcriptional regulation and to incorporate uncertainty about the functionality of putative transcription factor binding sites. Applying our approach to the deadliest species of human malaria parasite, Plasmodium falciparum, we obtain several striking results that deserve further (biological) investigation.

[ PDF ] [ Bibtex ] [ External URL ]

M. Kamal, A., J. Davis, J. Nabukenya, T., V. Schoonover, L., R. Pietron, and G.-J. de Vreede. Collaboration Engineering for Incident Response Planning: Process Development and Validation. In: Proceedings of the 40th Hawaii International Conference on System Science (HICSS, Vol: 0, Pages: 1-10, May, Proceedings of the 40th Annual Hawaii Internationa, Radboud University Nijmegen2007.

Many organizations have plans for incident response strategies. Despite Incident Response Planning (IRP) being an essential ingredient in conjuring security planning procedures in organizations, extensive literature reviews have revealed that there are no collaborative processes in place for such a crucial activity. This study proposes a design for a facilitated incident response planning process using technology such as GSS. Three sessions were conducted and an analysis of the sessions revealed that the facilitated IRP process design held up strongly in terms of efficiency, goal attainment, and session participant satisfaction. Future research implications entail devising an all-encompassing integrative general approach that would be applicable to any form of corporate security development planning process.

[ PDF ] [ Bibtex ]

J. Nabukenya, P. van Bommel, and H.A. (Erik) Proper. Collaborative IT Policy-making as a means of achieving Business-IT Alignment. In: Proceedings of the Workshop on Business/IT Alignment and Interoperability (BUSITAL’07), held in conjunction with the 19th Conference on Advanced Information Systems Engineering (CAiSE’07), 978-82-519-2245-6, Vol: 1, Pages: 461-468, June, 2007.

This paper is concerned with the application of collaboration engineering to improve the quality of policy-making processes as they occur in a business-IT alignment context. Policies are needed to guide complex decision-making. The creation of such policies is a collaborative process. The quality of this collaboration has a profound impact on the quality of the resulting policies and the acceptance by its stakeholders. We therefore focus on the use of techniques and methods from the field of collaboration engineering to improve the quality of Business-IT alignment related policy-making processes.

[ PDF ] [ Bibtex ]

M. Op `t Land, and H.A. (Erik) Proper. Impact of Principles on Enterprise Engineering. In: roceedings of the 15th European Conference on Information Systems, Pages: 1965-1976, June, Radboud University Nijmegen2007.

Increasingly, organizations make use of enterprise architectures to direct the development of the en-terprise as a whole and the development of their IT portfolio in particular. This steering and directing is done by means of principles, which are essentially regarded as constraints on the design space for enterprise engineers, thus guiding them in their design efforts. In this paper we study the potential constraining effect of principles on the design of enterprises as well as the guidance designers may receive from these principles. We start by providing a brief dis-cussion on the concepts of enterprise architecture and enterprise engineering. We continue by discuss-ing a strategy to make principles specific and measurable enough to indeed allow them to constrain design space. This is followed by a discussion of a number of examples, taken from real-life practice, illustrating the steering effect of principles. Finally, we also briefly pay attention to the process that may be followed in formulating and formalizing principles.

[ PDF ] [ Bibtex ]

M. Samulski, N. Karssemijer, Peter Lucas, and P. Groot. Classification of mammographic masses using support vector machines and Bayesian networks. In: Proceedings of SPIE, Medical Imaging, Vol: 6514, 2007.

In this paper, we compare two state-of-the-art classification techniques characterizing masses as either benign or malignant, using a dataset consisting of 271 cases (131 benign and 140 malignant), containing both a MLO and CC view. For suspect regions in a digitized mammogram, 12 out of 81 calculated image features have been selected for investigating the classification accuracy of support vector machines (SVMs) and Bayesian networks (BNs). Additional techniques for improving their performance were included in their comparison: the Manly transformation for achieving a normal distribution of image features and principal component analysis (PCA) for reducing our high-dimensional data. The performance of the classifiers were evaluated with Receiver Operating Characteristics (ROC) analysis. The classifiers were trained and tested using a k-fold cross-validation test method (k=10). It was found that the area under the ROC curve (A_z) of the BN increased significantly (p=0.0002) using the Manly transformation, from A_z = 0.767 to A_z = 0.795. The Manly transformation did not result in a significant change for SVMs. Also the difference between SVMs and BNs using the transformed dataset was not statistically significant (p=0.78). Applying PCA resulted in an improvement in classification accuracy of the naive Bayesian classifier, from A_z = 0.767 to A_z = 0.786. The difference in classification performance between BNs and SVMs after applying PCA was small and not statistically significant (p=0.11).

[ PDF ] [ Bibtex ]

Stefan Visscher, Peter Lucas, Ildiko Flesch, and K. Schurink. Using temporal context-dependent independence information in the Exploratory Analysis of Disease Processes. In: Artificial Intelligence in Medicine (AIME) 2007, Lecture Notes in Computer Science, Pages: 87-96, 2007.

Disease processes in patients are temporal in nature and involve uncertainty. It is necessary to gain insight into these processes when aiming at improving the diagnosis, treatment and prognosis of disease in patients. One way to achieve these aims is by explicitly modeling disease processes; several researchers have advocated the use of dynamic Bayesian networks for this purpose because of the versatility and expressiveness of this time-oriented probabilistic formalism. In the research described in this paper, we investigate the role of context-specific independence information in modeling the evolution of disease. The hypothesis tested was that within similar populations of patients differences in the learnt structure of a dynamic Bayesian network may result, depending on whether or not patients have a particular disease. This is an example of temporal context-specific independence information. We have tested and confirmed this hypothesis using a constraint-based Bayesian network structure learning algorithm which supports incorporating background knowledge into the learning process. Clinical data of mechanically-ventilated ICU patients, some of whom developed ventilator-associated pneumonia, were used for that purpose.

[ PDF ] [ Bibtex ]

P. van Bommel, S.J.B.A. (Stijn) Hoppenbrouwers, H.A. (Erik) Proper, and Th.P. van der Weide. QoMo: A Modelling Process Quality Framework based on SEQUAL. In: Workshop proceedings of the 19th International Conference on Advanced Infromarion Systems Engineering: EMMSAD, Vol: 1, Pages: 121-130, June, 2007.

This paper aims to contribute to the area of conceptual model quality assessment and improvement. We present a preliminary modelling process-oriented ‘Quality of Modelling’ framework (QoMo), mainly based on the estab-lished SEQUAL framework for quality of models. QoMo is based on knowl-edge state transitions, cost of the activities bringing such transitions about, and a goal structure for activities-for-modelling. Such goals are directly linked to concepts of SEQUAL. We discuss how goals for modelling can be linked to a rule-based way of describing processes for modelling. Such process descrip-tions hinge on strategy descriptions, which may be used descriptively (for studying/analysing real instances of processes) as well as prescriptively (for the guiding of modelling processes). Descriptive utility of the framework is critical for the quality/evaluation angle on processes-for-modelling, and reflects the main intended contribution of this paper.

[ PDF ] [ Bibtex ]

Reports

A.J.J. van Breemen, J.J. Sarbo and Th.P. van der Weide. Toward a theory of natural conceptualization. Technical report: ICIS-R07002, January, Radboud University Nijmegen, 2007.

The focus of this paper is on the early phases of ER-modeling consisting of the primary conceptualization of the underlying application domain. To this end we introduce a process model for the generation of meaningful concepts for a domain description. In virtue of its close relation with cognitive activity, this process model also enables the modeler as well as the user to comprehend the concepts of the resulting domain in a natural way. Beyond this goal, natural conceptualization opens the possibility for the introduction of a uniform representation enabling the efficient combination of knowledge obtained from different stake holders during a modeling process.

[ PDF ] [ Bibtex ]

P. van Bommel, H.A. (Erik) Proper, and Th.P. van der Weide. Structured Modeling with uncertainty. Technical report: ICIS-R07022, September, Radboud University Nijmegen, 2007.

This paper starts with the description of the modeling process as a dialog, and describes the associated formal functions, including the feedback supporting the growing mutual understanding. The dialog has a procedural and an informational aspect. For this latter a controlled grammar is used, that has a user friendly and a system friendly side. These sides are related via an elementary syntactical transformation. Assuming some elementary requirements on the dialog participants, we prove the main theorem for information modeling effectiveness. We also propose a system of metrics to support the modeling process. In terms of these metrics, modeling heuristics can be described and evaluated. We demonstrate our ideas by a simple sample session.

[ PDF ] [ Bibtex ]

P. van Bommel, S.J.B.A. (Stijn) Hoppenbrouwers, and H.A. (Erik) Proper. QoMo: A Modelling Process Quality Framework based on SEQUAL. Technical report: ICIS-R07007, March, Radboud University Nijmegen, 2007.

This paper aims to contribute to the area of conceptual model quality assessment and improvement. We present a preliminary modelling processoriented quality framework (‘QoMo’), mainly based on the established SEQUAL framework for quality of models. QoMo is based on knowledge state transitions, cost of the activities bringing such transitions about, and a goal structure for activities-for-modelling. Such goals are directly linked to concepts of SEQUAL. We discuss how goals for modelling can be linked to a rule-based way of describing processes for modelling. Such process descriptions hinge on strategy descriptions, which may be used descriptively (for studying/analysing real instances of processes) as well as prescriptively (for the guiding of modelling processes). Descriptive utility of the framework is primary for the quality/evaluation angle on processes-for-modelling, and reflects the main intended contribution of this paper.

[ PDF ] [ Bibtex ]

A.J.J. van Breemen, and J.J. Sarbo. Sign Processes and the Sheets of Semeiosis. Technical report: ICIS-R07003, January, Radboud University Nijmegen, 2007.

After positioning our work in the field of information systems science, we will introduce the basic Peircean semiotic terms pertinent to sign analysis (sign aspects) and those pertinent to interpretation processes (interpretant aspects). Next we will match the sign aspects with the interpretant aspects in order to be able to derive our semiotic process model of cognitive activity. In order to derive the process model we will introduce the concept of a semiotic sheet. This paper is of interest to those engaged in semiotically oriented approaches to information systems and those interested in the Peircean theory of interpretants alike.

[ PDF ] [ Bibtex ]

G.J.N.M. (Guido) Chorus, Y.H.C. (Yves) Janse, S.J.B.A. (Stijn) Hoppenbrouwers, C.J.P. (Chris) Nellen, and H.A. (Erik) Proper. Formalizing Architecture Principles using Object-Role Modelling. Technical report: ICIS-R07006, February, 2007.

This technical report is the result of two experiments conducted as part of an ongoing research effort to formalize architecture principles. The experiment involves a first, and modest, evaluation of the use of ORM and ORC as a means to formalize and ground architecture principles. The experiments involve the evaluation of the use of ORM and ORC to formalize the example principles provided by the TOGAF (The Open Group Architecture Framework) and principles taken from industrial practice.

[ PDF ] [ Bibtex ]

M.A.J. van Gerven. Approximate Inference in Graphical Models using Tensor Decompositions. Technical report: ICIS-R07024, November, Radboud University Nijmegen, 2007.

We demonstrate that tensor decompositions can be used to transform graphical models into structurally simpler graphical models that approximate the same joint probability distribution. In this way, standard inference algorithms such as the junction tree algorithm, can be used in order to use the transformed graphical model for approximate inference. The usefulness of the technique is demonstrated by means of its application to thirty randomly generated small-world Markov networks.

[ PDF ] [ Bibtex ]

M.A.J. van Gerven. Tensor Decompositions for Probabilistic Classification. Technical report: ICIS-R07011, May, Radboud University Nijmegen, 2007, Longer version of a short paper presented at IDAMAP 2007.

Tensor decompositions are introduced as a novel approach to probabilistic classification and can be interpreted as a particular kind of mixture model. Since many problems in medicine and biology can be described as a classification problem, the approach is seen as a useful tool for biomedical data analysis. The approach is validated by means of a clinical database consisting of data about 1002 patients that suffer from hepatic disease. It is shown that the approach performs comparably to state-of-the-art results that have been obtained using a naive Bayes classifier.

[ PDF ] [ Bibtex ]

J. Nabukenya, P. van Bommel, H.A. (Erik) Proper, and G.-J. de Vreede. An Evaluation Instrument for Collaborative Processes: Application to Organizational Policy-Making . Technical report: ICIS-R07017, July, Radboud University Nijmegen, 2007.

Decision-making in organizations is guided by policies. Organizational policy-making is a complex process in which several parties are involved, with multiple backgrounds, incompatible interests, and diverging areas of interest, yet they all have to be brought together to produce an acceptable policy result. Therefore, we propose to use techniques from collaboration engineering (CE) in this context. There is hardly any experience with CE in the field of organizational policy-making. In order to evaluate the effectiveness and efficiency of CE in organizational policy-making, it is important to have a systematic evaluation instrument. We distinguish between general and domain-specific indicators. Moreover, we consider measurement means and operationalization tools, such that organizational policy-making stakeholders can apply our instrument in their own organization.

[ PDF ] [ Bibtex ]

J. Nabukenya, P. van Bommel, and H.A. (Erik) Proper. Repeatable Collaboration Processes for Mature Organizational Policy Making. Technical report: ICIS-R07019, July, Radboud University Nijmegen, 2007.

Organizational policy making processes are complex processes in which many people are involved. Very often the results of these processes are not what the different stakeholders intended. Since policies play a major role in key decision making concerning the future of organizations, our research aims at improving the policies on the basis of cooperation. In order to achieve this goal, we apply the practice of collaboration engineering to the field of organizational policy making. We use the thinklet as a basic building block for facilitating intervention to create a repeatable pattern of collaboration among people working together towards achieving a goal. Our case studies show that policy making processes do need collaboration support indeed and that the resulting policies can be expected to improve.

[ PDF ] [ Bibtex ]

S.J. Overbeek, P. van Bommel, H.A. (Erik) Proper, and D.B.B. Rijsenbrij. Characterizing Knowledge Intensive Tasks indicating Cognitive Requirements - Scenarios in Methods for Specific Tasks. Technical report: ICIS-R07005, February, 2007.

Methods for specific tasks include conceptual modelling in the context of information systems and requirements engineering in software development. Such methods dictate a specific way of working by describing necessary knowledge intensive tasks to fulfil while applying the method. An actor that is applying a method for specific tasks wishes to reach the goals following from the way of working. An actor may experience difficulties when trying to fulfil tasks as part of a method, related to the cognitive abilities required to fulfil a certain task versus the specific cognitive abilities possessed by the actor. This paper focusses on the cognitive abilities required to fulfil a knowledge intensive task while applying a method for specific tasks. This is based on a categorization and characterization of knowledge intensive tasks and scenarios in conceptual modelling of information systems and requirements engineering.

[ PDF ] [ Bibtex ]

E.D. Schabell. Taking a peek inside the cmlFramework. Technical report: ICIS-R07008, March, Radboud University Nijmegen, 2007.

This paper will walk you through a demonstration run of the Collaborative Modeling Lab (CML) Framework prototype. This will start with a conceptual overview of the component structure. The reader will then be walked through a processing run in which a Niam Normal Form (NNF) grammar is used to Filter a Logbook, produce a Contract, Distill this into an Essence, and finally using the Builder to generate an eventual Model in ORM.

[ PDF ] [ Bibtex ]