Deadline: 15 November 2017
Postdoctoral researcher in recommender systems, information retrieval or data mining
Delft University of Technology – Faculty of Electrical Engineering, Mathematics and Computer Science
- Publication date
- 17 October 2017
- Level of education
- Salary indication
- €3,475 to €4,757 gross per month
- Closing date
- 15 November 2017
- 38 hours per week
- Vacancy number
The FairNews project is part of the Dutch National Science Agenda and aims at delivering interdisciplinary answers to questions of algorithmic biases in the context of news consumption and how those biases affect fundamental rights (freedom of expression, etc.). The project’s use case is provided by De Volkskrant, one of the largest Dutch daily morning newspapers.
The postdoctoral researchers will work as part of a multidisciplinary team of 1 legal scientist, 1 communication scientist and 1 data scientist, led by Dr Joris van Hoboken (IViR/UvA), Prof. Claes de Vreese (ASCoR/UvA) and Dr Claudia Hauff (TU Delft).
The Software Technology department is part of the Faculty of Electrical Engineering, Mathematics and Computer Science at the Delft University of Technology (TU Delft). The department offers an internationally competitive interdisciplinary setting for its 25 faculty as well as its 100 postdoctoral and PhD researchers. The department consists of six groups which research the design, engineering and analysis of complex, distributed and cooperative software and computer systems. The Web Information Systems Group has six full-time faculty members and more than 15 engineers, PhD candidates and postdoctoral researchers. The group concentrates in its research on engineering and science of the Web and aims to improve the understanding of people’s actions, interests, motivations, and behaviors on the Web, and subsequently leverage that knowledge to build Web applications that are semantic, personalized and adaptive. Within this group, research is carried along different research lines including human computation, recommender systems, information retrieval, data mining and database systems.
The postdoctoral researcher will be based within TU Delft’s Web Information Systems Group and supervised by Dr Hauff. The post doc will be part of the larger interdisciplinary team.
Equal opportunities of access to news is a necessary precondition for a functioning democracy. Data analytics, machine learning and personalised recommendations make it possible to filter news, based on individual user profiles and ‘social sorting’. We are only beginning to understand the implications for political knowledge and participation in the public debate, but also the realisation of fundamental freedoms, such as freedom to expression and non-discrimination.
The objective of this project is to develop solutions that lead to a) more transparency and effective ways of informing users about algorithmic profiling and targeting in the news media and b) guidance for offering personalised recommendations in the news media in a way that is fair, non-discriminatory and accessible. As a practical use case, the project will work together with one of the leading Dutch newspapers.
The project is part of the Dutch National Science Agenda, and here in particular the Big Data Route: VWData.
The postdoctoral researcher will be charged with:
- investigating algorithmic biases in a data-driven manner (data provided by the use case);
- proposing algorithmic remedies for the observed algorithmic biases;
- engineering user-facing prototypes that contain explainability components and enable users to learn why they receive certain information recommendations;
- evaluating the proposed solutions in simulated and real-world studies;
- disseminating the conducted research through conference papers and journal articles.
The postdoc has:
- a PhD degree in Computer Science. Candidates who are due to submit their dissertation before 1 December 2017 are welcome to apply;
- a research background in recommender systems, user modeling, information retrieval, data mining or a closely related area;
- a clear research interest in the topic of the project (documented by relevant publications);
- the willingness, and commitment to do autonomous work in a multidisciplinary team, that includes political scientists and communication scholars;
- the ability to engineer prototypes (using a modern software stack);
- the ability to lead and contribute to academic publications, contribute to grant proposals, and interact with stakeholders outside of. Strong verbal and written communication skills in English are therefore also required. Dutch is not required, but considered a plus.
Questions about the project content should be addressed to:
All general questions about the application procedure should be addressed to the:
The position ideally starts on 1 January 2018 (or soon thereafter) and lasts for 2 years. Depending on your knowledge and experience, your monthly gross salary will range between €3,475 and €4,757 according to the Collective Labour Agreement (cao) for Dutch Universities, based on a full-time appointment. The final conditions are pending budget approval.
We offer a pension scheme, a holiday allowance of 8% per year and flexible employment conditions.
The deadline for applications is 15 November 2017 at 24:00 PM CET. Please state vacancy number 17-509 and the letter A in the subject line of your application. Without stating the vacancy number, the application cannot be processed. To apply for this position, please send at least the following documents in English by e-mail to firstname.lastname@example.org:
- motivation letter;
- curriculum vitae, names and contact details of 3 references;
- proof of your PhD degree. If you have not completed your degree at the time of application, please provide a statement from your supervisor confirming the submission date of your PhD before 1 December 2017;
- two writing samples in English, at least one of which academic (e.g., journal article, PhD dissertation).
Please send all your documents in PDF format. Only complete applications will be considered. #LI-DNP
No agencies please