I have a specific question concerning open licenses.
Is it considered a international best practice to have specific licenses published for each dataset?
There are some cases in which a institution doesn’t worry in publishing a open data license even though the data is public and regulated by a general law of publishing information.
I considered, in those cases, that data is in public domain (because it actually is), but I didn’t find no specific license. Is that right?
For the most practical thing, I think you should only link to the law where you can find that specification. And add this information in the comments, so reviewers will know what was the decision process for adding the law and not the license specifically.
It’s same in Poland. Nobody bothers to explicitly say “This data is in public domain”. With some exceptions all data/information created by public institution is in public domain.
It would be clearer for data recipients if such clear message was included, so I guess that this metric is there to suggest such behaviour.
In Poland I’d say we have following levels:
- explicit message about licensing (super rare)
- data on data portal or BIP (FOIA portal), which is law-regulated and implicates public domain
- data published elsewhere - then let’s look at FOIA practice (see below) or assume it’s in public domain and don’t care about consequnces
- data not published; easy to obtain via FOIA
- troublesome to obtain via FOIA, but most courts are ruling in favour
- controversial in courts
So in the end, I count explicit message about licensing and publishing data on data portal as OK, the rest as NOT. I’m not getting into practical aspects of reuse.
Thank you for bringing this up. Our new “open licensing/public domain” question will be accepted if you can provide evidence on one of the following two points:
- Every submission must check whether an open license is provided. The license must inform you about the legal openness of a specific dataset. This means that you have to find a license that applies to the dataset. The license does not have to refer explicitly to a dataset. It is also possible that a license refers more generally to data provided on a website / catalogue. Key is that you can somehow see that a license applies to the dataset. You should check if the license applies, send a link to the license and document your rationale in the comment box (explaining why you think the license applies). It is also important to verify whether the open license complies with the open definition.
- Even if no open license can be found there it is possible that the dataset is in the public domain (for a broader discussion see here, and here). In this case we accept submissions if the submitters is able to provide a link to the law / disclaimer stating that the data is in the public domain.
So either of the two options is accepted for submission. Key is that you can somehow verify that the license / public domain status applies to the dataset. If this is not possible, please use the comment function to explain why and also send us the link to any possible disclaimer/license, etc.
It would be very helpful to get feedback from others whether it is hard to figure out if an open license /public domain status applies to a specific dataset.
Please do not hesitate to send any other question you might have.
Hope this helps
Thanks for the answers!
I actually followed exactly what you said. When I didn’t find an explicit license, I put the link where I found a statement of the public domain and used comments to explain why I used that.
excellent, I’m happy that the answer was useful. Please do not hesitate and post in this thread any further issues you might have with the open licensing question - for instance whether it is hard to understand if a disclaimer of public domain status applies to a dataset.
This is very useful information for us to write better guidance material, as well as for open data advocates and government to improve the understandability of disclaimers and open licenses.
In case of Russia:
Nowadays, we have our own “licence” which is obliged for each open data dataset providing the following single URL - http://data.gov.ru/information-usage
Entry for National Laws / Czech Republic
It is indeed difficult to answer the question re. open license / public domain status (for Latvia in my case).
There are datasets where it is clear that it has been published in order to inform everyone about the facts that it covers (e.g. state budget details or election results).
They are not covered by copyright and almost certainly are in the public domain (or assumed to be by whoever you ask). Similar as in Poland, almost no datasets have a specific clause saying it is public domain (perhaps because people responsible can’t even imagine someone thinking otherwise).
What to do in this situation in cases when the law does not specify the data is public domain?
For future surveys it would be good to have a 3rd option for this question that would fit situations like this - to distinguish b/w datasets that clearly are not open (e.g. have usage restrictions) and datasets that are most likely public domain.
In the case of Brazil, I think the rationale should like be the following:
- If there is information about licensing accompanying either the dataset, or the website where it is located, use that.
- Otherwise, if it was published on or before 1983, it is in the public domain.
- If the dataset is made available by the executive branch of the federal government, it is considered licensed by the federal Decree 8.777/2016, Chapter II.
- Otherwise, apply the SOCD test. If it fails the test (i.e. is not copyrightable), it is in the public domain.
- Apply the OAR test. If it passes the test (i.e. is not copyrightable), it is in the public domain.
- If the data passes the SOCD test and fails the OAR test (i.e. may be copyrightable), it is not open data.
The SOCD test
What is it? The Brazilian Law 9.610/1998 states in Article 7, XIII, that databases are copyrightable only if, by their “selection, organization or content disposition, they constitute an intellectual creation”. So, factual data is clearly not copyrightable.
The OAR test
What is it? The Brazilian Law 9.610/1998 states in Article 8, IV and V, that official acts and registries (among other cases) are exempt from copyright. That may apply to some government data.
The Access to Information Law (Law 12.527/2011) states in Article 3, I, that publicity is the general rule and secrecy is an exception - which can be made only in cases justifiable by one of the situations defined in Law. So, if a public body has already published the data on the internet, it means it does not have secrecy requirements.
Other relevant legislation
For government datasets on the federal level, Chapter II of the Decree 8.777/2016 states that “any data made available by the executive branch, federal level government, as well as any information of active transparency, are free to be used by the government and by society”.
In some very specific cases there might be other laws that restrict the usage of data (e.g. taxation secrecy, banking secrecy, trade secrecy, etc.), so one should be on the lookout for those.
For a more extensive discussion on the matter (in Portuguese), see also this page.
Open Data Index - Brasil 2017 Cities FAQ
Entry for National Laws / Brazil
Entry for Weather Forecast / Brazil (openly licensed?)