Dominic Heutelbeck

Distributed Hybrid Genetic Programming for Learning Boolean Functions

When genetic programming (GP) is used to find programs witli Boolean inputs and outputs, ordered binary decision diagrams (OBDDs) are often used successfully. In all known OBDD-based GP-systems the variable ordering, a crucial factor for the size of OBDDs, is preset to an optimal ordering of the known test function. Certainly this cannot be done in practical applications, where the function to learn and hence its optimal variable ordering are unknown. Here, the first GP-system is presented that evolves the variable ordering of the OBDDs and the OBDDs itself by using a distributed hybrid approach. For the experiments presented the unavoidable size increase compared to the optimal variable ordering is quite small. Hence, this approach is a big step towards learning well-generalizing Boolean functions.

Distributed Leader Election in P2P Systems for Dynamic Sets

The collection of and search for location information is a core component in many pervasive and mobile computing applications. In distributed collaboration scenarios this location data is collected by different entities, e.g., users with GPS enabled mobile phones. Instead of using a centralized service for managing this distributed dynamic location data, we use a peer-to-peer data structure, the so-called distributed space partitioning tree (DSPT). A DSPT is a general use peer-to-peer data structure, similar to distributed hash tables (DHTs), that allows publishing, updating of, and searching for dynamic sets. In this paper we present an efficient distributed leader election algorithm that can be used in DSPTs to eliminate redundant network traffic.

Towards Support for Long-Term Digital Preservation in Product Life Cycle Management

Important legal and economic motivations exist for the design and engineering industry to address and integrate digital long-term preservation into product life cycle management (PLM). Investigations revealed that it is not sufficient to archive only the product design data which is created in early PLM phases, but preservation is needed for data that is produced during the entire product lifecycle including early and late phases. Data that is relevant for preservation consists of requirements analysis documents, design rationale, data that reflects experiences during product operation and also metadata like social collaboration context. In addition, also the engineering environment itself that contains specific versions of all tools and services is a candidate for preservation. This paper takes a closer look at engineering preservation use case scenarios as well as PLM characteristics and workflows that are relevant for long-term preservation. Resulting requirements for a long-term preservation system lead to an OAIS (Open Archival Information System) based system architecture and a proposed preservation service interface that respects the needs of the engineering industry.

Context Spaces - Self-Structuring Distributed Networks for Contextual Messaging and Resource Discovery

Geographical addressing and resource discovery are important services in mobile context-aware computing environments. In this paper we present a protocol that maintains a self-organizing routing backbone that supports these services. Every node taking part in the protocol actively participates in the maintenance of the network. While distributing administrative tasks, the protocol takes into account the context and capabilities of the nodes. The network acts robustly with respect to massive geographical movement of the participating nodes and runs without central administration. We also introduce the concept of context spaces that act as a tool for context-awareness, information filtering, and workload distribution. Context spaces can be used to build complex context-aware systems. Based on our protocol, we also present a new approach for structuring file-sharing networks.

A peer-to-peer data structure for dynamic location data

The collection of and search for location information is a core component in many pervasive and mobile computing applications. In distributed collaboration scenarios this location data is collected by different entities, e.g., users with GPS enabled mobile phones. Instead of using a centralized service for managing this distributed dynamic location data, we present a new peer-to-peer data structure, the so-called distributed space partitioning tree (DSPT). A DSPT is a general use peer-to-peer data structure, similar to distributed hash tables (DHTs), that allows publishing, updating of, and searching for geometrical objects

Fault Tolerant Geographical Addressing

Geographical addressing and resource discovery are important building blocks for creating mobile location-aware systems. Mobile systems rely on wireless network technologies, which are prone to frequent failures. In this paper, we present a protocol for maintaining a self-organising routing backbone that supports geographical addressing and resource discovery. The routing backbone detects node failures and automatically repairs its infrastructure, while maintaining the operability of the network at least in partitions an thus maximising service availability. The system operates in a peer-to-peer fashion, i.e. the administrative tasks are distributed between all participating nodes, taking into account their capabilites, location, and mobility. In this paper we explore maintenance aspects of the network that are caused by node failures.

EDISON Data Science Framework: A Foundation for Building Data Science Profession for Research and Industry

Data Science is an emerging field of science, which requires a multi-disciplinary approach and should be built with a strong link to emerging Big Data and data driven technologies, and consequently needs re-thinking and re-design of both traditional educational models and existing courses. The education and training of Data Scientists currently lacks a commonly accepted, harmonized instructional model that reflects by design the whole lifecycle of data handling in modern, data driven research and the digital economy. This paper presents the EDISON Data Science Framework (EDSF) that is intended to create a foundation for the Data Science profession definition. The EDSF includes the following core components: Data Science Competence Framework (CF-DS), Data Science Body of Knowledge (DS-BoK), Data Science Model Curriculum (MC-DS), and Data Science Professional profiles (DSP profiles). The MC-DS is built based on CF-DS and DS-BoK, where Learning Outcomes are defined based on CF-DS competences and Learning Units are mapped to Knowledge Units in DS-BoK. In its own turn, Learning Units are defined based on the ACM Classification of Computer Science (CCS2012) and reflect typical courses naming used by universities in their current programmes. The paper provides example how the proposed EDSF can be used for designing effective Data Science curricula and reports the experience of implementing EDSF by the Champion Universities that cooperate with the EDISON project.

IJDL focussed issue on persistent archives

Digital preservation is a major concern in the digital age. The classical challenges we face in this area are multifaceted and range from data medium conversion to software emulation for documents and multimedia objects. As we generate massive amounts of highly complex data, new challenges arise to keep resources usable, meaningful, and understandable over long time periods. The four papers collected in this issue look at different aspects of persistent archives, i.e., classification of collections, different distributed architectures for archival, as well as the state-of-the-art in the domain of engineering which has been and is still driving a lot of the requirements for persistent archives. In the paper “Using wavelet analysis for text categorization in digital libraries: a first experiment with Strathprints” by Daranyi/etal the focus is on handling large amounts of documents and to categorize them automatically for later access. It combines different research areas to form an innovative approach. In “Extending OAI-PMH over structured P2P networks for digital preservation” Everton/etal explore the possibility to distribute the responsibility of keeping copies and therefore prevent loss of information to a network of OAI-enabled libraries. They also take the reliability into account and calculate the best distribution. To investigate complex objects Brunsmann explores in “State-of-the-art of long-term preservation in product life

Recalot.com: Towards a Reusable, Modular, and RESTFul Social Recommender System

Many different recommender system RS frameworks have been developed by the research community. Most of these RS frameworks are designed only for research purposes and offline evaluation of different algorithms. A reuse of such frameworks in a productive environment is only possible with high effort. In this paper, we present a concept of a generic reusable RESTful recommender web service framework, designed to perform directly offline and online analysis for research and to use the recommender algorithms in production.

Towards an Open Archival Information System Compliant Exchange Format to Ensure Reproducibility of Assays in Cancer Care

Recent studies by Amgen and Bayer showed that only 11% and 25% respectively of reviewed cancer research projects produced scientifically valid results. The awareness of this fact is accompanied by increasingly stringent regulations, which enforce compliance to ensure the ability to validate research results through its proper reproducibility. Furthermore, research undertakings in this field are expensive and the return on investment needs to be secured by research purchasers through proper management of the knowledge that is required for long term result reuse. For example, the provision of personalised medicine is tightly interwoven with the execution of laborious molecular diagnostic assays besides the comprehensive access to all constituents of it to enable full result reproducibility for its later validation. Hence, a key factor for successful utilization of assay findings is a comprehensive preservation support of assay results together with all relevant information that come into existence throughout the phases of an assay lifecycle. The provision of tools and services that provide comprehensive preservation functionalities will dramatically increase the effective support research of personalised medicine in cancer care through improved reproducibility capabilities for later validation.