CBS and VU Amsterdam: progress through cooperation

/ Author: Miriam van der Sangen
Vrije Universiteit Amsterdam (VU) and Statistics Netherlands (CBS) will explore new opportunities to intensify the existing cooperation between the two institutions. To this end, an expert meeting between researchers from both institutions was held at VU on 29 May 2017. Both existing and new opportunities for collaboration were discussed, as well as new techniques such as ‘distributed analytics’ for collective analysis of multiple data sources which are stored in different locations without having to move them. At the end of the expert meeting, CBS Director General Tjark Tjin-A-Tsoi and VU Rector Magnificus Dr Vinod Subramaniam signed a framework agreement ratifying the collaboration.

Framework agreement

CBS and VU have worked together for years in several areas. With the aim to intensify this collaboration, the decision was made to sign a framework agreement. Professor Dorret Boomsma, biologisch psychologist at VU and known for her research in the field of behavioural genetics and Twin Register research: ‘The framework agreement simplifies everything that is already in place and enhances direct cooperation between CBS and VU researchers, for example. Now that we have met and talked with each other in person, we can follow that up with a number of projects.’ Boomsma sees the expert meeting as very valuable: ‘Five of our researchers gave a presentation, after which the CBS researchers provided direct feedback.’

Netherlands Twin Register

Boomsma started to work on a Twin Register in 1987. Thirty years on, the register contains data on over 100,000 ‘multiples’ and is one of the most important international databanks for human genetics. In her research group, Boomsma is searching answers to questions such as: to which extent are human characteristics and diseases genetically programmed, and to which extent are they determined by the environment and the way a person was raised as a child? Boomsma explains what she anticipates for the Twin Research programme as a result of closer cooperation with CBS:
‘We would like to find the answers to some basic questions. For instance, CBS has interesting information which can be used to map people’s environment characteristics. The Twin Register contains information about genetic relationships and on individual characteristics, which can only be collected by approaching the individuals themselves. Think of attitudes and personality traits, for example. Combining those various types of information could yield new insights.’

Myriad possibilities

While working with CBS over the past two years, Boomsma has found that nowadays, myriad forms of research are possible by unlocking database information. ‘We often look at Scandinavia to learn about population registers, but in that respect we also have huge possibilities here in the Netherlands. However, much remains to be done in order to unlock this information. An important next step is the National Road Map of The Netherlands Organisation for Scientific Research (NOW), which offers a research infrastructure on a large scale.’ Aside from the possibilities, Boomsma also sees a number of potential hurdles for cooperation: 'In the framework agreement, CBS stipulated the condition that we must conduct statistical analyses within the CBS environment (due to privacy regulations, ed.). But this means we will encounter obstacles, for example because we cannot simply link functional MRI data to CBS data on education, due to the sheer size of the datasets.’

‘CBS considers it vitally important to make progress on text mining in the field of big data’

Textdata mining

Bart Bakker is Head of Team Methodology at CBS in The Hague and professor by special appointment at VU Amsterdam, where he teaches Register Methodology for social scientific research. Like Boomsma, he considers the expert meeting on 29 May as extremely informative. ‘We organised a number of workshops, covering such topics as data and text mining (using various ICT techniques to retrieve valuable information from large amounts of text, ed.), exchanging data sources and data linkage. CBS considers it vitally important to make progress on text mining in the field of big data, e.g. for classification of retail products in order to determine the consumer price index (CPI). As of July 2017 we will hold discussions with VU to assess whether we could collaborate in this area.’ Another research area of interest to CBS, according to Bakker, is ‘distributed analytics’. This is a technique that involves collective analysis of multiple data sources stored at different institutions without the need to move them. VU is already conducting trial research in this field, while this topic is also being studied by other Dutch universities (Eindhoven, Maastricht).

Pioneering

Looking back on the expert meeting, Bakker sees many opportunities for collaboration. ‘Three types of researchers were present at this meeting: first, those already working with CBS; second, those who work with CBS data but do not communicate on the data with anyone at CBS; and third, those not working with CBS data and not familiar with our researchers. The latter two groups in particular are interesting for expansion of our network and mutual exchange of knowledge.’ The expert meeting concluded with Director General Tjark Tjin-A-Tsoi of CBS and rector magnificus Vinod Subramaniam of VU signing the framework agreement. Both stressed the importance of joint investment for further progress. Tjin-A-Tsoi pointed out that the use of big data such as sensor data offers significant opportunities for academic research as well as official statistics. ‘There is a world to win if such data can be used effectively. Since we are just pioneering, there is a real chance that projects do not immediately yield the desired results. By investing jointly in this type of research in particular, the risk may become smaller on both sides.’