As a spinn off of the autoevaluation Tom and I are writing a chapter for a book with the "Enterprise Linked Data" that will be published by Springer january next year.
Comments are welcome
first paragraph: I have to verify if we can refer to the autoevaluation as long it is not officially concluded
4th paragraph: ..:"the team is paying a price for being a bit too far ahead of the curve" should this read "the team is paying a price for having been a bit too far ahead of the curve" ?
"The simple but rigid metadata record format it defined", please change to "the concept of "application profiles" it used for different datasets has allowed....:"
Please insert after the 4th paragraph of Chapter 2 "The AGRIS/DLIO application profile had been meant from the beginning as a means to gather RDF data from partners and had been published in RDF. It was always aimed to generate datasets that could be expressed in triples. The problem was that most AGRIS partners were/are far away from the capacity to express their data in RDF. The AGRIS DTD served as crutch to convert these data into an RDFizable repository."
"In practice, it has fallen to the staff in Rome to correct and validate much of the data semi-manually". Please substitute this with " The introduction of the AGRIS AP as a common exchange format reduced also drastically the necessity for editing and cleaning up data, which had been done before 2000 with a team of more than 10 persons at the AGRIS processing unit in Vienna"
Page 8, 3rd paragraph
"In the AGRIS context, the interoperability of descriptions is based on their syntactic conformance with the AGRIS DTD and ingest procedures are designed to transform local formats into data conformant with the DTD. In the current AGRIS database, in other words, interoperability results from sharing record formats. This has been the dominant paradigm for the interoperability of digital information for many years"
This, in this formulation is not true. The interoperability in the AGRIS database results by the sharing of the meaning of the data elements, this is only mediated through the AGRIS DTD, because there was no other means for doing so.
I propose the following language:
"The dominant paradigm for the interoperability of digital information for many years has been through their syntactic conformance with specific data schemes encoded as XML DTDs or XML Schemas. AIMS application profiles have been developed to define the meaning of specific data elements to make them interoperable. As the concept of Linked Data was not developed at that time and most AGRIS partners had and have no opportunity to publish their data as linked data the AGRIS DTD has served as a crutch to achieve interoperability. Transforming AGRIS partner data into AGRIS DT format wash nothing else then establish an "OWL:same as" relation between data elements of different sets". Therefore also only 4 of the many elements of the AGRIS AP were mandatory. All the others were only possibilities to express "same as relations"."
This might need editing, but is slightly different view. If you agree, this should also be inserted into the autoevaluation.
Page 13, paragraph 2
"The AGROVOC thesaurus is a loose"..... should be changed to "The AGROVOC thesaurus was a loose"..... In the re-engineering process of the last years, more important than the "ontological over commitment" that has been done in the introduction of class/subclass relations, was the cleaning up of AGROVOC, with it's 100s of Orphans, 100s of top terms and 1000s of other inconsistencies.
Page 21, paragraph 4
I have to assure that the reference to the 2004 auto evaluation can be made
Page 23, paragraph2
"CIARD nicely complements the role of AIMS". Please change to: "CIARD is a broader context in in which AIMS can become effective.