This is a discussion about the submission for Government Budget / Brazil.
Even though the submission has been quite thorough, as the federal Brazilian budget is available from so many official sources, one of them went unreported:
which, in turn, points to, among other ways to access the same data:
In contrast, the link which the submitter appears to have taken into consideration, does not seem to work at this moment.
Data is available in bulk in NTriples format. This is one of the standard formats for RDF and Linked Open Data, something which is considered to be a good practice and 4-star open data. As such, data literate users are expected to be able to use it. The submitter, however, has answered “no” to the question “B5. Is the dataset downloadable at once?”, using as a justification his or her lack of familiarity with the format. It is already pretty difficult today to convince governments to release linked open data, and it will become even more so if they start to be penalized for taking the extra effort of doing so.
Under question “B8. In which formats are the data?”, RDF should have been marked as well, alongside CSV which is already checked.
In response to question B9 (effort), the submitter mentioned there is no API, which is kind of true - there used to be a linked open data API using the open source Elda framework, but it has since been regrettably deactivated. However, a SPARQL Endpoint is still provided. By using it alongside the documentation of the Brazilian budget RDF vocabulary/ontology, users can build quite powerful queries.
Notwithstanding the aforementioned data literacy issue, the data is inherently rather difficult to use because the budget subject matter is itself dry and difficult to learn. It is important to note that the documentation is there - the technical budget manual, one for each year. In order to efficiently use the dataset, users have to know not only the intricacies of the budget described there, but also how to manipulate data on the RDF ecosystem. Not many people know or are willing to learn both, which may explain the low usage this dataset gets.