= DCAT Application Profile for Data Portals in Europe - Brecht Wyns Deidre(intro): DCAT-AP: Try to come together on challenges. Is deversity good, or should we bring the APs together? Basic usecase is to improve discoverability across borders and sectors Shows timeline - [Slide self explanatory] Stat DCAT-AP is on the way. Guidelines for DCAT-AP DCAT-AP has been successfull - proved by uptake. E.g. national extentions Geo-DCAT-AP: ensures compliance with Inspire and Iso19115 Issues with DCAT identified: 1) Issues that can be solved in guidelines 2) Issues for 3rd parties 3) Issues with DCAT itself Guidelines is an open process. Invitation to contribute == Issues with DCAT: - Need #1: To express relations between datasets: Evolution, time series, parts, collections - Need #2: Rights and licenses: No licence on dataset (only distributions) Need #3: Service based data access: Distributions in DCAT is agnostic to this. Designed for file-distribution? Need #4: Relatonships between distributions: Distributions can contain different data and slices of data. There is already a variety of practices. Issue: Many datasets are often ZIP-ed. How to express this? What is in the package? Need #: Data citations: Lineage, authors, how to use, publications, link to input All DCAT-AP work is under ISA2 programme. Stay connected! Thank you. Q: Why distributions as slices? A: Not time to get into this. Talks afterwards and bar at camp session Q: How is the research community involved? A: Only research centre, in order to solve it we will involve more Nicos (comment): The discussions are not closed. This is a working group. In the first fase CERIF was in the loop, we don't have direct input from research right now. The situation is parallell rooms/discussions ----- = DCAT For a Long Term Global Future - Andreas Kuckartz Good news: DCAT-AP.DE will come in Germany. This is enforced by the government. Public consultations revealed critical feedback to the OGD approach.OGD 1.0 was not compatable with anything else. Proposal was dropped and new standard will be based on DCAT-AP. Probably ready March next year. Issue: National extentions. Have to read a lot and several tools necessary. More has to be put in the core to avoid extentions. DCT:formats: both URLs and strings are used. Messy! Validation needed Issues with long term view: PDF used for archiving purposes. (No questions) ------ Using DCAT for Development Data Hub - Beata Lisowska The Development Data Hub is a visualisation tools. Aims to make data more accessible, easy to disaggregate and comparable in an intuitive way. Behind the scenes: a powerful data warehouse Challenge: The growing complexity of interconnected data. Looked at using DCAT Headache: using dcat on the data warehouse. Yes DCAT is flexible... How to merge forces using DCAT on data warehouse and DCAT on front end? Hard! Is there a better way? Conclusions: [Slide self explanatory] Tools needs to provide comprenehcive machine readable metadata Seeking comments on paper (public) Q: Do you think there is overhead to track dataset-journeys? A: Yes but it is worth it. One statement in DCAT is not eneough. The dataset journey should be described. There are ways! Q: Have you used semantic data link? (hard to hear question) A: No Q: Metadata for every data point: how to distinct? A: No answer here yet ----- = Metadata for business and open data - Jeroen Baltussen [paper (PDF)] [slides] RVO.nl wants to use their data internally (and externally) in a more integrated way. We are data oriented and looking for solutions. Scope of problem: 4 areas: 1) product, 2) data element, 3) application, 4) datasets These are different areas. How to choose the right model? There is a lot af standards. Our area is broad. Have looked at several core vocs. Two of them relevant: Core Location Voc and Core Public Service Voc. How to make these different vocs work together? Difficult but we tried: RVO first draft ready. Comment 1: Develop an abtract model for your business and serialize and use API. Comment 2: Models are intended for interoperability inside EU. It does not necessary fit in an organisation. You can't mesh vocabularies together in one model. A: Does it matter what is formalized vocabularies what is not? ADMS is not btw A: NO Comment: Start by establishing a basic set of concepts. ----- = Discussion: Deidre. Thanks to all. Q: What belong in core and what in APs and extentions? A: Lets analyze the national extentions and identify common solutions. My impression is many of the same issues sorted in different ways. A2: I agree. Interresting to look into these extentions. E.g. the Netherlands AP and how it try to link to legal issues. Q: (Deirde): The different suggestions and shortcomings - where to go with intepretations? A: (Beata): It is a limitation that it (DCAT) is so flexible, but also good. Balance is the key. One feedback: A need for non-tech readings/docs on DCAT Brect: How to use DCAT needs to be dealth with by W3C/DCAT too Phil: This is a dilemma. Doing this is a lot of text. A full intepretations is even not possible. There is a limit and depends how you choose to use it. The line is blurred. Q: Are there specific issues with DCAT? A: In germany 20 issues were raised with DCAT-AP. Only a few in reality: Boiled down to Agents and Roles Max: There are several (and ovelapping) standards on related issue. Data cubes, files, endpoint, changing datasets, snapshots. Deidre: DCAT fundamentaly describes datasets. How to evolve? Andrea: New issue: When organisations have/make their own metadata catalog: Think of DCAT as layer on top?, - or replace excisting metadata solutions? This is crucial for adoption. A: we are using DCAT as an inspiration. The idea of an integrating language is attractive. But how? Incredible different formats, at some point we must bring it together. We have to simplify. There will always be tension between depth and with. Comment: Most open data platforms is open source. Making DCAT available in open source tools will make it easy to migrate. To that it will help to have one common format - not one for each country. Phil: CKAN does not brake down to dataset and distributions, Comment: Advice: MAP your silos to DCAT - not try to solve everything with DCAT Deirdre: The importance of DCAT as a interopeabile tool. The role of DCAT is a general model that can bring things togehter. Q: How does this fit with schema.org? A: Google: simplicity is the key. What is the basics? Mathias: There are other tools than CKAN. E.g. Entryscape Catalog. Tools are there as long as we agree on RDF Q: The next step in discoverability is contextualizing. Schould we adress this in a better way? A: Yes. You can go too simple. Q: Is it possible to strech DCAT to solve versioning? A: We think it should. It is a key issue for DCAT to be relevant. We are moving beyond filebased distributions. A2: APIs and files is two differen worlds. It is dangerous to try to do too much A3: To describe what the distributions are, like Andrea suggest is a good approach. That will already help alot A4: When looking for data files or apis doesn't matter. Peter: Basics: Is the datset likely to be updated or not? Bottom line: DCAT is doing something right when you look at adoption = Wrapup - Tension between simplicity and domain complexity. Getting the levels right in documentations is important. - Encouraging that with DCAT uptake (?) - DCAT is a standard for datasets. We are looking for a broader approach - The scope is the most challenging. There is always a risk of overthinking. balancing thjis with the reallife problems peopleru ninto Deidre: Thanx. We got ideas of shortcomings and frustrations. Next thing is solutions.