Automatically annotating texts with background information has recently received much attention. We conduct a case study in automatically generating links from narrative radiology reports to Wikipedia. Such links help users understand the medical terminology and thereby increase the value of the reports. Direct applications of existing automatic link generation systems trained on Wikipedia to our radiology data do not yield satisfactory results. Our analysis reveals that medical phrases are often syntactically regular but semantically complicated, e.g., containing multiple concepts or concepts with multiple modifiers. The latter property is the main reason for the failure of existing systems. Based on this observation, we propose an automatic link generation approach that takes into account these properties. We use a sequential labeling approach with syntactic features for anchor text identification in order to exploit syntactic regularities in medical terminology. We combine this with a sub-anchor based approach to target finding, which is aimed at coping with the complex semantic structure of medical phrases. Empirical results show that the proposed system effectively improves the performance over existing systems.

Consumer products are becoming increasingly software intensive. The software complexity of individual products grows, while the diversity of products increases and the lead time must decrease. Software reuse is the answer to this, not only within a family but also between families of consumer products. We have devised an approach based upon a software component technology to enable reuse. This paper describes that approach, and it zooms in on two important aspects of component-based development. One aspect concerns the prediction of system properties from properties of components, which we illustrate using thread synchronization as example. The other aspect concerns branching of our software in our configuration management systems, where our analysis leads to the discovery that we may be constantly rewriting our own code and to the definition of the turn-over factor to quantify this. We end this paper with a brief validation of our approach.

Abstract Consumer products become more complex and diverse, integrating functions that were previously available only in separate products. We believe that to build such products efficiently, a compositional approach is required. While this is quite feasible in hardware, we would like to achieve the same in software, especially in the low‐level software that drives the hardware. We found this to be possible, but only if we let software components communicate horizontally, exchanging information along software channels that mirror the hardware signal topology. In this paper a concrete protocol implementing this style of control is described and many examples are given of its use. Copyright © 2003 John Wiley & Sons, Ltd.

Two trends have made reuse of embedded software for consumer electronics an urgent issue: the software of individual products becomes more and more complex, and the market demands a larger variety of products at an increasing rate. For that reason, various business groups within Philips organize their products as product families. A third trend is the integration of functions that until now were only found in separate products (e.g. a TV with Dolby Digital sound and a built-in DVD player). This requires software reuse between product families, which- when organized systematically- leads to a product population approach. We have set up such a product population approach, and applied it in various business groups within our organization. We use a component technology that stimulates context independence, and allows the composition of new products out of existing parts. We use an architectural description language to explicitly describe the architecture, and also to generate efficient bindings. We have aligned our development process and organization with the new 'compositional ' way of working. This paper outlines our approach and reports on our experiences with it.

When building small product families, software should be shared between different members of the family, but the software can still be created as one system (with variation points) with a single architecture. For large and diverse product families (product populations), the software can no longer be developed in one context and at one moment in time. Instead, one must combine software components of which the development is separated in space and in time, each with their own evolution path. In other words, we need independent deployment of components. We discuss four aspects of independent deployment. Two of these aspects: upward and downward compatibility, deal with variation in time. The other two: reusability and portability, deal with variation in space. For each aspect, we indicate the relevance, provide some examples, and list several techniques to deal with it. The paper can thus be seen as a guide for product population development.

This paper reports on our experience with a relational approach to support the analysis of existing software architectures. The analysis options provide for visualization and view calculation. The approach has been applied for reverse engineering. It is also possible to check concrete designs against architecture-related rules. The paper surveys the theory, the tools and some of the applications developed so far. © 1998 John Wiley & Sons, Ltd.