The Service Oriented Computing (SOC) paradigm aims at building complex systems by composing them from less complex systems, called services. Such a (complex) system is a distributed application often involving several cooperating enterprises. As a system is usually subject to change, individual services will be substituted by other services during the systems life-cycle. Substituting one service by another one should not affect the correctness of the overall system. Verification of correctness is challenging, as the overall system is usually not known to any of the involved enterprises. The focus of the BOSS project is to study service substitution for a set of practical relevant correctness notions. Thereby, we restrict ourselves to changes of the service behavior. We develop algorithms to decide service substitution and provide rules to refine services while preserving correctness of the composition. We investigate service substitution at the level of the service definition as well as migrating running instances from one service definition to another one. In addition, we study service similarity and develop algorithms to measure the similarity of two services. The analysis techniques developed in this project will be implemented in a prototype and evaluated using several real-life case studies.

Our increased life expectancy unfortunately goes hand in hand with an increased number of years living with chronic conditions. These conditions cannot be cured, but both the development of comorbidities and the quality of life are strongly influenced by proper self-management and by deploying a healthy lifestyle. In the EDIC project, we develop a novel data-driven artificial coaching platform that supports chronically ill patients in making optimal lifestyle decisions. We bring in expertise in data mining, telemedicine, e-coaching, behavioral sciences and domain expertise, and particularly focus on the bariatric and diabetes-2 patient population.

There is a revolution happening online, as web browers are no longer the main tool to explore information on the Internet, due to fact that users are moving towards mobile devices. Due to their smaller screen sizes and different interaction paradigm, it is inconvenient to use traditional web browsers on smartphones and tablets. Hence, many mobile applications have been developed to enhance the user experience, and any online information or service provider has it?s own ?app.? Evaluation remains a central component of web search applications because it helps to understand the quality of current system and which direction to take in order to improve it. However, the usual explicit user behavioural signals (e.g. clicks) are not sufficient anymore, as other implicit behavioral signals such as browsing become more prominent. We developed a technology that allow to infer user satisfaction with the search result pages on mobile devices. Our method is using the interaction signals (e.g. touch, swipe or zoom) instead of clicks which are rare and less meaningful. Improving user satisfaction with their services is key for millions of e-commerce businesses, and the goal of feasibility studies is to explore market potentials for the developed technology.

This research has developed algorithms for dynamic data analytics that are based on techniques for automatically constructing machine learning pipelines for the task at hand. The approach has been validated on two complementary practical application tasks: The early detection and treatment optimization for Parkinson’s disease, and the cost- and environmentally optimized management of energy for private households with electric vehicles. The project output includes a new online automated machine learning pipeline, new methods for multi-variate time series prediction, and new approaches for early stage Parkinsons disease diagnostics from videos.