[All BiBTeX entries for this year]

H.A. (Erik) Proper. * PRONIR proposal. *Technical report, Nijmegen Institute for Information and Computing Sciences, University of Nijmegen, Nijmegen, The Netherlands, EU, 2002, NWO Project Proposal.

The knowledge and information resources we need to conduct our activities in daily life,
be it at work or at home, are increasingly available in some electronic form by way of
the Internet.
Some examples of such resources are: *documents*, *people* (by their e-mail or
chat addresses), *document collections*,
*objects and facts in databases* and even entire *applications*.
People in search of knowledge turn to the Internet with the aim of finding relevant
knowledge or information that will support them in executing their tasks.
This research project aims to develop a theory, and demonstrate its validity by means
of a prototype system, for profile based retrieval of heterogeneous networked resources.

N. Lynch, R. Segala, and F.W. Vaandrager. * Hybrid I/O Automata. *Technical report: NIII-R0201, February, Radboud University Nijmegen, 2002.

Hybrid systems are systems that exhibit a combination ofdiscrete and continuous behavior. Typical hybrid systems include computer components, which operate in discrete program steps, and real-world components, whose behavior over time intervals evolves according to physical constraints.Important examples of hybrid systems include automated transportation systems, robotics systems, process control systems, systems of embedded devices, and mobilecomputing systems.Such systems can be very complex, and very difficult to describe and analyze.This paper presents the Hybrid Input/Output Automaton (HIOA) modeling framework, a basic mathematical framework to support description and analysis of hybrid systems.An important feature of this model is its support for decomposing hybrid system descriptions.In particular, the framework includes a notion of externalbehavior for a hybrid I/O automaton, which captures its discrete and continuous interactions with its environment.The framework also defines what it means for one HIOA toimplement another, based on an inclusion relationship between their external behavior sets, and defines a notion of simulation, which provides a sufficient condition for demonstrating implementation relationships.The framework also includes a composition operation for HIOAs, which respects external behavior, and a notion of receptiveness, which implies that an HIOA does not block the passage of time.The framework is intended to support analysis methods from both computer science and control theory.This work is a simplification of an earlier version of the HIOA model. The main simplification in the new model is a clearer separation between the mechanisms used to model discrete and continuous interaction between components.In particular, the new model removes the dual use of external variables for discrete and continuous interaction.

F.A. Grootjen, and Th.P. van der Weide. * Conceptual Query Reformulation. *Technical report: NIII-R0202, Computing Science Institute, University of Nijmegen, 2002.

One of the biggest challenges of search engines is to capture the user's information need. This paper proposes a way to interpret the results of a user's initial query. This is done by positioning the preliminary result into a semantical structure called concept lattice. The gathered conceptual information can be used for relevance feedback in subsequent searches.

A.T. Arampatzis. * Unbiased S-D Threshold Optimization, Initial Query Degradation, Decay, and Incrementality, for Adaptive Document Filtering. *Technical report: NIII-R0203, February, Radboud University Nijmegen, 2002.

We develop futher the S-D threshold optimization method. Specifically, we deal with the bias problem introduced by receiving relevance judgements only for documents retrieved. The new approach estimates the parameters of the exponential-Gaussian score density model without using any relevance judgements. The standard expectation maximization (EM) method for resolving mixtures of distributions is used. In order to limit the number of documents that need to be buffered, we apply nonunifom document sampling, emphasizing the right tail (high scores) of the total score distribution.For learning filtering profiles, we present a version of Rocchio's method which is suitable and efficient for adaptive filtering. Its main new features are the initial query degradation and decay, while it is fully incremental in query updates and in calculating document score statistics. Initital query degradation eliminates gradually the contribution of the initial query as the number of relevant training documents increases. Decay considers relevant instances (documents and/or initial query) of the near past more heavily than those of the early past. This is achieved by the use of half-life, i.e. the age that a training instance must be before it is half as influential as a fresh one in training/updating a profile. All these new enhancements are consistent with the initial motivation of Rocchio's formula.We, moreover, use a form of term selection for all tasks (which in adaptive tasks is applied repeatedly), and query zoning for batch filtering and routing.

[ Missing PDF ] [ Bibtex ]

H.A. (Erik) Proper, and S.J.B.A. (Stijn) Hoppenbrouwers. * Information System Evolution as Language Evolution. *Technical report: NIII-R0205, Nijmegen Institute for Information and Computing Sciences, University of Nijmegen, 2002.

In this article we look at the evolution of information systems from the
perspective of the evolution of domain languages, and provide a report on
some of our

Many if not all analysis and design approaches for information systems base themselves on techniques involving some sort of natural language analysis. However, the view on language underlying these approaches ignores several issues concerning the nature of language.

This article primarily aims to create awareness for the issues at play, and argues that current approaches ignore most of them. We finish by presenting an overview of our research efforts concerning the tackling of the matters discussed.

P. Achten, and R. Hinze. * Combining Generics and Dynamics. *Technical report: NIII-R0206, July, Radboud University Nijmegen, 2002.

In this paper we discuss the surprisingly elegant interaction between two lines of rather independent research, namely generic (or polytypic) programming and dynamic types. We show how the two disciplines meet and emphasize their characteristics in the aerea of interactive programming as implemented in the functional language Clean. In this language, an extensive library for creating Graphical User Interfaces is provided. We demonstrate how dynamic types can be used to store arbitrary GUI objects as resources and how generic functions can be used to manipulate these values. This results in a flexible system of strongly typed and higher-order resources.

[ Missing PDF ] [ Bibtex ]

M. Hendriks. * Enhancing UPPAAL by Exploiting Symmetry. *Technical report: NIII-R0208, October, Radboud University Nijmegen, 2002.

Efficiency is one of the major concerns in the world of model checking.Consequently, many techniques to optimize the time and space usage of model checking algorithms have been invented. One of these techniques is reduction of the searchable state space through arguments of symmetry. This technique can be very profitable and has been implemented in various model checkers, but not yet in UPPAAL, a model checker for timed systems. This paper proposes an enhancement of UPPAAL with symmetry reduction. We adopt the theory of symmetry of Ip and Dill and their scalarset data type, as implemented in the model checker Murphi. The main result of this paper is a soundness proof of our symmetry enhancement, which does not follow trivially from the work of Ip and Dill since the description languages of UPPAAL and Murphi, which are used to detect the symmetries, are quite different. A secondary result is a proof of the computational difficulty of space-optimal exploitation of full symmetry in a setting with DBM technology.

K. Brown, P.J.F. Lucas, and D. Fowler. * Decision Network Semantics of Branching Constraint Satisfaction Problems. *Technical report: NIII-R0209, December, Radboud University Nijmegen, 2002.

Branching Constraint Satisfaction Problems (BCSPs) have been introduced to model dynamic resource allocation subject to constraints and uncertainty. We give BCSPs a formal probability semantics by showing how they can be mapped to a certain class of Bayesian decision networks.

A.L. de Groot, and H. Wupper. * A Taxonomy of Reactive Systems. *Technical report: NIII-R0210, December, Radboud University Nijmegen, 2002.

This paper proposes a general terminology for talking about real-time systems and how we observe their behavior. While none of the notions presented here are new, bringing them all together gives us a better overview.

[ Missing PDF ] [ Bibtex ]