Catálogo de publicaciones - libros

Users can better understand complex data sets by combining insights from multiple coordinated visual displays that include relevant domain knowledge. When dealing with multidimensional data and clustering results, the most familiar displays and comprehensible are 1- and 2-dimensional projections (histograms, and scatterplots). Other easily understood displays of domain knowledge are tabular and hierarchical information for the same or related data sets. The novel parallel coordinates view [6] powered by a direct-manipulation search, offers strong advantages, but requires some training for most users. We provide a review of related work in the area of information visualization, and introduce new tools and interaction examples on how to incorporate users’ domain knowledge for understanding clustering results. Our examples present hierarchical clustering of gene expression data, coordinated with a parallel coordinates view and with the gene annotation and gene ontology.

The abundance of data available nowadays fosters the need of developing tools and methodologies to help users in extracting significant information. Visual data mining is going in this direction, exploiting data mining algorithms and methodologies together with information visualization techniques. The demand for visual and interactive analysis tools is particularly pressing in the Association Rules context where often the user has to analyze hundreds of rules in order to grasp valuable knowledge. In this paper, we present a visual strategy that exploits a graph-based technique and parallel coordinates to visualize the results of association rule mining algorithms. This helps data miners to get an overview of the rule set they are interacting with and enables them to deeper investigate inside a specific set of rules. The tools developed are embedded in a framework for Visual Data Mining that is briefly described.

The introduction of computer technology caused a shift away from real objects as sources of information towards desktop computers as interfaces to information now (re)presented in a digital format. In this paper, I will argue for returning to the real world as the starting point for designing information and communication environments. Our approach is to design environments that exploit the affordances of real world objects and at the same time use the potential of computer-based support. Thus, we move from human- interaction to human- interaction. Combining the best of both worlds requires an integration of real and virtual worlds resulting in hybrid worlds. The approach will be demonstrated by sample prototypes we have built as, e.g., the Roomware components and smart artefacts that were developed in the project “Ambient Agoras: Dynamic Information Clouds in a Hybrid World” which was part of the EU-funded proactive initiative “The Disappearing Computer” (DC).

Many ubiquitous computing scenarios deal with cooperative work situations. To successfully support these situations, computer-supported co-operative work (CSCW) concepts and technologies face new challenges. One of the most fundamental concepts for cooperation is sharing. By analyzing applications of sharing in the context of ubiquitous computing it can be shown that ubiquitous computing enables an extended view on sharing. In this paper, we show that this extended view seamlessly integrates the view of “traditional” CSCW and additionally incorporates ubiquitous, heterogeneous, and mobile devices used in a common context.

This paper presents an intuitive metaphor and user interface for managing access permissions of a shared workspace. It is based on a combination of the room and key metaphor, which represent shared workspaces and the users’ access permissions. Keys and rooms can be used for different modes of group formation. We report on experiences with the metaphors in the context of the collaborative learning environment CURE.

The vision of Ambient Intelligence is based on the ubiquity of information technology, the presence of computation, communication, and sensorial capabilities in an unlimited abundance of everyday appliances and environments. Today’s experimental smart environments are carefully designed by hand, but future ambient intelligent infrastructures must be able to configure themselves from the available components in order to be effective in the real world.

We argue that enabling an ensemble of devices to spontaneously act and cooperate coherently requires software technologies that support self-organization. We discuss the central issues pertaining to the self-organization of interactive appliance ensembles and outline potential solution paradigms: Goal-based interaction and distributed event processing pipelines.

Nowadays, we witness the advent of a new era in information technology as an enabler for an often proclaimed “knowledged society”. Many academicians – computer scientists, computational linguists and others – but also researchers in corporate settings investigate the use of “knowledge” as a means for intelligent information access in enterprises. Ontologies are commonly considered a formal means for modelling and instantiating conceptual structures upon which implicit knowledge can be inferred. In an industrial context, however, domain-specific conceptual information is only one corporate information resource. Relevant business information comprises facts on goods, projects, experts, customers, competitors and many other information bits as they are stored in enterprise information systems. A great wish of many executives is, however, to gain an integrated view on these assets and their relationships, thus deriving new insights relevant for their business. To achieve this goal, ontologies can help with respect to semantic data integration. A problem is, however, that the construction and maintenance of ontologies is expensive. Another problem is that the business data are usually stored redundantly in different heterogeneous data repositories and the connectivity – a prerequisite for intelligence – is not explicit. The approach discussed in the remainder will show how a mix of information extraction and classification methods can be used to automatically set-up and update a network of business objects serving as a corporate memory index. The latter represents a rich semantic access structure for filtering and individualizing the retrieval of relevant business information.

Shortening time-to-market for new product innovations is now and in future one of the critical success factors for market competitiveness and ability. The ability to faster and better arrive at innovative products is based on the knowledge, what constellations are excessively time-consuming and how this time barrier can be broken down in order to achieve acceleration. The ability to develop new things fast and effectively and to introduce them on the market does nowadays substantially depend on the increasingly complex knowledge and creative performance of all employees during the entire innovation project. At the moment we do, however, lack a computer-supported solution, aiming to provide the employee working on the innovation process with an appropriate support for an effective acceleration of innovations. A system for the acceleration of innovation, on the on hand, has to be backed up with a model allowing to represent innovation projects – and the occurring time-consuming constellations with all their complex interrelations and their semantics – in an information system. On the other hand, the system must be equipped with all knowledge concerning undesired time consumers and the respective avoidance strategies, the so-called innovation-acceleration knowledge. Ontology-based models and methods enable computer-supported representation of innovation projects, computer aided identification of time-consuming constellations and serve to infer and provide innovation-acceleration knowledge to ensure faster and better innovations.

R&D efforts in the area of e-Science aim at building an infrastructure that supports researchers’ work processes in a flexible way relying on Grid-based, on-demand resource assignment. We believe, that within the developing e-Science environments an improved awareness of researchers for their scientific context as well as a more flexible support for the interaction of the researcher with this context is desirable. This paper discusses approaches for context exploitation in support of an increased awareness for the scientific context as well as approaches for more active context construction by the researcher for specific research projects.

In this paper, we present a recommendation system providing users with personalized TV schedules. The TV Scout architecture addresses the “cold-start” problem of information filtering systems, i.e. that filtering systems have to gather information about the user’s interests before they can compute personalized recommendations. Traditionally, gathering this information involves upfront user effort, resulting in a substantial entry barrier. TV Scout is designed to avoid this problem by presenting itself to new users not as a filtering system, but as a retrieval system where all user effort leads to an immediate result. While users are dealing with this retrieval functionality, the system continuously and unobtrusively gathers information about the user’s interests from implicit feedback and gradually evolves into a filtering system. An analysis of log file data gathered with over 10,000 registered online users shows that over 85% of all first-time users logged in again, suggesting that the described architecture is successful in lowering the entry barrier.