|Review Comment: |
The revised version of the paper has clearly improved the earlier version submitted to the special issue for Linked Data sets, and hence I am happy to consider in my review an accept, as it is following now much better the guidelines for dataset descriptions, and I also consider that the requests that were done in the initial reviews have been addressed adequately.
Particularly, the division into different levels of quality in the dataset is very good IMO. In fact, I am thinking of replicating this same scheme in some of the datasets that my organisation is helping publishing as Linked Data.
I also acknowledge that the lack of provisioning of a SPARQL endpoint is not a problem, as the dataset is provided for download, and we all understand the difficulty of running an openly available SPARQL endpoint given that it requires human resources to maintain and fine-tune them.
There are only a set of minor comments that may be addressed by the authors if considered useful by them to improve the quality of the manuscript for its camera-ready version, and which are stated here:
- The abstract is probably too long and considering the special issue where this paper will appear, some of the sentences are unnecessary. this applies, for instance, to the first three sentences, which provide general comments on Linked Data and its benefits. I would prefer it authors can focus more quickly on the development that has been done. Probably I would suggest starting with the "We have developed the neXtProt...", and only rescue some of the earlier sentences that talk about the nanopublications model.
- The same comment also applies to the second paragraph of the introduction, which is not at all needed for this special issue. I would suggest removing that paragraph as well.
- In section 2, the authors make a comment on "the term schema is understood in the Linked Data context as the mixture of distinct terms from different RDF vocabularies that are used by a data source to publish data on the Web". This is not necessarily true. All terms may come from a single vocabulary, and furthermore, the vocabularies may be encoded also in OWL. I would suggest removing this sentence.
"a initial" --> "an initial"