We are pleased to announce the initiation of three new Path-Finding Projects. The projects are intended to provide the opportunity to rapidly meet short-term scientific challenges, serve as a pilot for future research projects, address immediate technological goals, or investigate the potential to initiate larger projects.
The three projects result from the July Open Call for Path-Finding Projects. Path-Finding Project proposals can be submitted at any time. The eScience Center funds projects by the direct provision of cash and the in kind provision of expertise from our eScience Research Engineers.
For more information on oru funding opportunities visit our project calls page.
Case Law Analytics
Dr. Gijs van Dijck
This project aims to develop a technology that assists the legal community in analyzing case law. Legal scholars commonly and understandably study a limited number of court decisions when conducting legal research.
With tens of thousands court decisions (in the Netherlands) that are published yearly, numerous decisions remain unstudied. In this project, a technology will be built that allows analyses of which court decisions are central in a network of decisions. This will be done by focusing on citations: the number of references to a certain court decision in other court decisions (in-degree centrality). The in-degree centrality of court decisions is likely reveal new patterns among decisions on various topics. Moreover, it will shed more light on which decisions are the most important within a certain network. The automated process that will be developed enhances the analysis of relationships among a large number of decisions.
Future projects may expand the application (1) by processing additional information in the decisions and coding it into researchable variables and (2) by expanding the technology to other countries than the Netherlands. This project (and future projects) can fundamentally transform the way the law is studied (researchers, students) and used (practitioners).
Parallelisation of multi point-cloud registration via multi-core and GPU for localization microscopy
Dr. Bernd Rieger
Delft University of Technology
We aim to develop an image processing infrastructure to facilitate image processing and in particular to enable multi point-set registration for super-resolution fluorescence microscopy. The next step in this field after the Nobel prize 2014, will be nanometer resolution by higher (virtual) labelling densities of emitters via computational means.
Image processing infrastructure has been developed by our group over the last 20 years and is packaged into DIPlib, a C-library, and DIPimage, a Matlab front-end (www.diplib.org). These libraries have been downloaded about 1000 times every year since 2000 (of which about 20 loyal academic groups). In order to reach our aim we need to develop a new, accessible open source image processing package supported by our usebase that is suitable for multi-core and GPU parallelization.
To provide for multi-dimensional data with flexible object-types we need e.g. run-time type information. For our application in super-resolution microscopy, efficient algorithm design and parallelization are a must. Here 104 point-clouds must be registered, where each cloud contains 103-104 points, and the computation of the cost-function itself scales quadratic in the number of points. We would like an all-to-all template-free registration, but that is currently not feasible.
Data quality in a distributed learning environment
Prof. Andre Dekker
MAASTRO Clinic, Maastricht University
Rapid learning healthcare (RLHC) based upon routine clinical care data is a rising novel approach for providing an evidence base to support optimized clinical decisions and therefore individualize cancer treatment. However, data quality in RLHC is critical to the amount of confidence that can be placed in the acquired knowledge and therefore the decisions to be taken. This issue of data quality in RLHC must be met in order for its full potential as an evidence base upon which to make decisions to be realized.
At MAASTRO we have been working towards this goal over the past number of years by developing a distributed learning platform. In this platform SPARQL endpoints are created for routine clinical care data stored in networked hospitals. Applications are distributed to these endpoints to learn predictive models. In such an approach data quality is a critical issue as researchers can no longer directly access and quality assure the data.
A number of necessary data quality tools have been identified such as the ability to detect, report, and impute missing/implausible/contradicting data. The state-of-the-art is to apply reasoning at the data store, our proposed solution is to expand the SPARQL query using reasoning centrally and apply it locally.