Linked Data in Sweden
Linked Data in Sweden is an annual, one day, conference that has been arranged for at least 6 years in different places in Sweden. This year the conference was arranged by Linköping University and the location was the university area Valla. The topic is linked data and there are presenters and participants from a wide range of areas, from health care to trucks.
The day was introduced by the hostess of the day, Eva Blomqvist from Linköping University. She welcomed everyone, presented the agenda and introduced the first speaker – her colleague Olaf Hartig, presenting the topic “Linked Data Fragments”.
Using the web as a database requires queries to be run on the web reaching out to the data where it belongs. This is the core of Linked Data but due to performance reasons data is nevertheless often pooled in big local data stores. One reason for this is that SPARQL endpoints, that are often used to query data over the net, are vulnerable to badly written SPQRQL requests and quite often non-responsive. A study from 2013 show that half of the investigated SPARQL end points only has an availability of 95% which is far from ok.
Using Linked Data Fragments (LDF), which are much more simple queries than SPARQL, improves the performance on the server by moving the complexity to the client. LDF only return triples matching one single triple pattern (which are easy for the server to respond to). For complex requests many queries might need to be run by the client. The performance of LDF is much lower than with SPARQL for single clients but when scaled up to many clients the performance is not degraded which is the case with SPARQL endpoints.
The second speaker of the day was Alexander Alvsilver from Vinnova. He presented data driven labs which is something that Vinnova has funding to support. The goal is to increase the usage of data. We should not only Collect, Store and Publish data – we must also learn from data, use it as decisions support and build services around it!
The third speaker of the day was me, Magnus Wallberg, representing Uppsala Monitoring Centre. The topic I was there to present was WHODrug as linked data and the proof of concept work that is ongoing. Is this something worth continuing with and if so – how should we proceed. Comments from the audience suggested that we must produce good use cases and make the new technologies compatible with old legacy systems thus making a step by step transition possible. It is important to show a good ROI plan.
Next to present was Damir Nesic from SAAB SCANIA and KTH. He showed how linked data is used as a middle layer in SCANIA to achieve multi system integration and querying. The linked data layer will link together many old legacy systems that cannot be easily replaced and that speak “different languages”. Damir also mentioned that SCANIA use SHACL and SPARQL queries to validate consistency in the Linked Data layer.
Last presentation before lunch was done by Dag Rende from FindOut Technologies. He showed how LDVis could be used to graphically present results from SPARQL queries. Interesting - but it needs lots of more work to be user friendly and appealing to a bigger the crowd. Generally, I think that the RDF society should look more at Neo4j and the graphical representation in that tool.
After lunch Ulrika Dommellöf Mattsson presented the work Naturvårdsverket do with indicators. The EU Environmental Agency is using RDF and linked data since long and now Naturvårdsverket is increasing their usage. So far it has mostly been used to publish data but from now it will be used also for collection and storing (this was at least my interpretation of what she presented). If linked data is used from the start Naturvårdsverket will be able to replace much of the Excel based data handling which is the reality of today. The goal is to make the process of data collection from local level and further on to European level fully automatic. In the work Naturvårdsverket is developing a Master Data concept where the following key concepts are important: rules for management, common data owners, a common view among different stakeholders and uniquely identifiable information objects.
Next presenter was Erik Wallin from Klipsk. He presented the work that Klipsk is doing on “Smart Building Ontologies” with Vasakronan. Erik claims that with better interoperability between IoT systems and sensor utilities we can reduce the energy consumption in buildings with as much as 50% and increase the utilization with 85%. This interoperability can be achieved using Linked Data and common Ontologies. The Ontology being developed by Klipsk and Vasakronan is called VKCore and it is available as open source on GitHub.
In the end of his presentation Erik mentioned the group “Swedish Semantic Practitioners”. This is a group where those that use (and commercializes) linked data can meet and discuss.
Next up was Fredric Landqvist from Chalmers (and Findwise). He presented work being done together with the municipality of Gothenburg to define master data and to find existing Ontologies to re-use, alternatively create own Ontologies. The aim is to build a master data repository for services, persons, organizations and places (as the identified key concepts) in Gothenburg.
The last presentation of the day was by Mattias Palmér from MetaSolutions. He demonstrated how linked data sources could be gracefully integrated in web sites alongside traditional data sources using EntryScape Blocks in a Publishing Environment.
The day was finished with group discussions and tutorials. One group had a tutorial introducing Ontologies and they also discussed Ontology Design. The other group was about project ideas and match making. I participated in the second group which had a lively discussion. There were lots of ideas coming up, problems being raised and solutions suggested. Overall it was a good discussion with topics ranging from version control of Linked Data to real time data streaming using RDF.
After the conference a number of participants went to the Klomp Bar in central Linköping where the discussions continued over some food and drinks. The discussions did however not stay on the Linked Data subject but covered everything between cycling and hyperloops!
One important thing that was discussed was where the next annual meeting will be held - it seems as if the estafet-baton might be handled over to Uppsala. MetaSolutions has already done the first scouting of places and costs. Possible partners in arranging the one day conference could be MetaSolutions, Uppsala University, the municipality of Uppsala and perhaps Uppsala Monitoring Centre. We will see what happens. However - the first thing now should be to thank the current hosts for a well managed and interesting meeting in a very friendly atmosphere!
Linköping as a city was also a very nice location for the conference and even the weather gods made their best to welcome us. The city is nicely located along the Stångå river (also called Kinda canal) with two beautiful sluice areas. There are many beautiful building, nice squares and churche in the central part of the city. I also had the opportunity to do my morning runs in the dawn alongside the river on the nice walk and cycle ways stretching out on both sides.