Linked data in the spotlight – with Platform Linked Data Netherlands

Interview with Pieter van Everdingen, Community Manager at Platform Linked Data Netherlands (PLDN)

Semantic interoperability is strongly connected to the concept of linked data, being one of its core components and allowing heterogenous data sources to connect via the Internet. In this interview, we discussed with Pieter van Everdingen, the Community Manager from Platform Linked Data Netherlands (PLDN), which is an open network community of experts aiming to stimulate further adoption of linked data. Pieter van Everdingen has been involved with the PLDN since 2012 and has background in technology management, AI and business rules. Together we looked at PLDN’s activities and objectives, but also at topics such as linked data in the Dutch public sector, the relationship with SEMIC vocabularies and the current obstacles for a wider use of linked data. 

What is linked data and what are the benefits of using such an approach?
In a nutshell, linked data is a web-based technology, which allows you to describe, model, store, publish and interlink data from different sources in a uniform way via the Internet. At its core, linked data has two elements: uniform resource identifiers (URIs) and the use of controlled vocabularies. URIs make data accessible, connectible and reusable for a global audience on the Internet and the controlled vocabularies help to add the right semantics, metadata and provenance information to this data. Individuals and organisations can then have a better understanding of what your data means, how it can be reused and what meaningful new connections can be made with other datasets. With linked data, different datasets become more compatible, more interoperable and easily extensible between different environments and domains. Working this way, data silo issues can be solved, by removing the semantic and technical barriers for easier data sharing and exchange. 

Could you give us an overview of Platform Linked Data Netherlands (creation, objectives, activities, participants)?
PLDN is an open network community of linked data experts in the Netherlands who would like to share their knowledge and experiences in linked data projects with everyone who has an interest in linked data and who would like to get started with their own activities in this field. It has developed into a vibrant community with several hundred active members and a few thousand interested parties. 

Our PLDN activities concentrate around organising networking events, facilitating activities within a number of working groups and releasing publications on linked data topics, like the Gems of Linked Data Applications book (available in Dutch), which contains successful examples of linked data initiatives from our PLDN network that we would like to share with a larger audience. We also stimulate our PLDN community members to experiment with linked data via our new user-friendly and free to use PLDN lab environment. And we work with organizations in our network to start up new promising linked data initiatives and projects that can possibly be financed or co-financed from innovation budgets and grants within the EU and the Netherlands. PLDN is completely open for any work collaboration that would lead to further adoption of linked data and more best practices. 

Speaking of your activities, Play-a-LOD is one of PLDN’s initiatives which educates on linked data in a more playful way. Can you tell us more about this?
The first discussions of Play-a-LOD started in 2017, when we were looking for a way to explain linked data in an easy to understand and playful way using a game format. A small PLDN working group started developing a card board game using resources from an animal ontology for the cards of the game. During the game, 2-4 players try to create semantically sound triples (Subject-Predicate-Object constructs) with the cards they have. Each triple construct corresponds to the sum of the points on the respective cards, resulting in a score during each turn for each player. Every card also has a QR-code, which can be used to access additional digital information about a card (as linked data) from e.g. DBpedia and Wikidata in the triple store of our PLDN lab environment. The player with the highest total score for all the created triples wins the game. Play-a-LOD was successfully launched in June 2019 and was an immediate success. To everyone's surprise, the game format works really well to explain linked data in an easy to understand and playful way. And we are now thinking of possible follow-up activities to the first Play-a-LOD game (e.g. games using other ontologies, games in other languages, a digital version of the game, etc.).

Do you collaborate with public sector organisations? How do the Dutch public authorities integrate linked data practices?
In our PLDN activities, we have a good mix of public and private partners we collaborate with. This allows to have strong synergies and contributes to expand our network. Regarding public authorities, they are interested and involved in our activities at all levels within the public sector. Representatives from the public sector share with us their project results at our events, they participate in our working groups, they write articles for our publications and help us identify relevant case studies and successful projects. We have a very good relationship with Forum Standaardisatie (Dutch Standardisation Forum), as our work is directly dependent of open internet standards, where we also stimulate further adoption of these standards via our activities. It is also worth mentioning that there is a growing interest in linked data in the Dutch public sector, notably via the increased popularity of knowledge graphs worldwide, which attracted more attention to our topic in the past year.

What do you see as the main obstacles in the near future for linked data initiatives in the public sector?
First, the concept of linked data is a real paradigm shift in the area of data integration when you compare it to more traditional data integration methods. So, you have to think differently, you are not thinking in terms of tables, columns and rows anymore, but in terms of overlapping graphs of interlinked data, which look like data clouds. And these graphs can contain cross-organization, cross-domain and cross-border data sources, where you can connect any linked data source via the Internet in a uniform way without making unnecessary copies of that data. This is very different from more traditional data integration methods, where you work mostly with a variety of very rigid data formats, while you copy sizeable amounts of data from one environment to another. We try to tackle this issue via initiatives such as Play-a-LOD. With the current Covid19 crisis, we also see an increasing interest in linked data. The Covid19 context creates the need to combine data from multiple sources in order to analyse very complex global and local situations, which led to a more widespread use of knowledge graphs. I see this as a positive trend for linked data and hopefully this change will continue after the crisis is over.

A second obstacle is the limited availability of easy to use and affordable linked data tooling. A lot of open source linked data tooling are unfortunately not sustainable to last over a longer time period. They are often difficult to install and use when you are just initiated to linked data. Moreover, they usually have a small group of adopters and the maintenance and support activities are often too limited to attract a larger crowd. Commercial linked data products on the other hand, are too expensive and therefore not a good fit for organizations that work with smaller budgets for linked data projects. Thus, there is a gap in linked data tooling which needs to be filled. 

Semantic interoperability plays an important role for linked data. How do you see SEMIC and ISA² vocabularies contributing to linked data initiatives such PLDN?
ISA² and SEMIC provide very good references in terms of what is important regarding data interoperability and what vocabularies can be re-used. We generally see the use of such vocabularies as a good practice and we have a sizeable number of organisations participating in our PLDN network, which are in direct contact with the SEMIC community. In this regard, we have a dynamic best-practices and experience sharing activities with organisations implementing ISA² core vocabularies, amongst which are Belgian institutions with whom we have strong cooperation relationship on semantic interoperability topics. I think these examples of positive experience sharing could pave the way for future collaboration opportunities.