DataScience@UiO: research and innovation cluster at the Department of Mathematics and Department of Informatics
The University of Oslo is building up a research community in Data Science. Four PhD positions are allocated to this new initiative, producing top research at the crossroad between statistics, machine learning, logic and computer science, with a specific focus on innovative uses of big data in their research projects.
Data Science at the University of Oslo (UiO) is fueled by internationally recognized data producing scientists at our university and anchored in the Department of Mathematics and Department of Informatics. Two centers of excellence in research based innovation have recently been funded (each with about 3 million Euro annually and with more than 100 researchers):
Big Insight, a consortium of academic, industrial and public partners, develops statistical and machine learning methodologies and computational tools to extract knowledge from complex and big data sources, with focus on two central themes, namely novel personalized solutions and sharper predictions of transient behaviors.
Sirius, Centre for Scalable Data Access, is a consortium of academic and industrial partners, and develop new scalable methodology for information systems for accessing data in an efficient and robust computational environment.
Innovation will be central in all four PhD projects. Candidates will develop new ideas and methodology motivated by challenges from our industrial partners. Support will be given by Inven2, the next-generation UiO innovation company, to build bridges between excellent research and the next generation of technology-based industry.
The four PhD students, together with other PhD students, postdocs, researchers and staff in data science, are expected to contribute to building a unique scientific community at UiO, joining and organizing internal and open activities (including industrial days, reading groups, industrial internships, drop-in Data Science advising service for other researchers, junior student supervision, etc.).
Brief description of the four research projects:
Recommendation systems for highly incomplete data
Users’ rating and ranking of items, and their clicking tracks on webpages, reveal their preferences. Such data arise in very many areas in the digital world, including business, entertainment, politics. We will develop new approaches and scalable algorithms to preference learning based on such data, exploiting knowledge on users and relation between items.
Smart grid communication and operation
Electrical power is now increasingly generated by smaller power sources connected in a network, such as small hydroelectric plants, wind turbines, private solar cell installations, so that both production and consumption are highly decentralized as a smart grid. We will work with various formulations of the problem as constraint optimization and scheduling, designing methods and fast approximate algorithms.
Probabilistic Methods for Entity Resolution
In many large organizations, data is often split into multiple, separate, complex databases of different type and quality, often assembled in different periods and by different actors, using different identifiers for the same entities. We will tackle the problem of name resolution across databases by augmenting ontology and database reasoning techniques with predictive statistical models, in order to quantify object similarity probabilistically, and hence produce mappings and rules that resolve name conflicts.
Ontology-based data exploration
One of the challenges of big data is understanding, navigating and exploring large and complex datasets, in order to extract useful information. We will work on ontology-based data access systems for data visualization and exploration in the context of data integration.