basisOfRecord from literature

[pzermoglio]

(These questions were posed during the second chapter of the DwC Hour, 7 Mar 2017).

What do we put in basisOfRecord if we are recording something from the literature (usually types), which we have not seen? Do we presume these are PreservedSpecimen?

Some answers were:

• In the case of type info lifted from a publication, we'd consider this a HumanObservation and the dcterms:type would be "Text".

[debpaul]

Here's at least where I see some of the confusion arising. The type exists, or existed somewhere. Someone mentions it in literature. This is "Specimen" data (from text, sure). It may also include references to the collector's recordNumber for the type specimen, or the collection catalogNumber (or other similar identifier). People want to share it as knowledge a specimen exists. At iDigBio, we don't take data where BasisOfRecord=HumanObservation because our scope is vouchered specimens. So, how do we suggest a researcher find this type specimen data? Senario: this would then be dcterms:type=Text dwc:basisOfRecord=HumanObservation dwc:typeStatus=[nomenclatural type status of the specimen as stated in the text]. These records would be submitted to GBIF, and a researcher would then limit search by typeStatus. Is that how you see it? I wonder if / how the community would then ever manage to connect this HumanObservation record, to a future PhysicalSpecimen record coming from a collection (assuming the type specimen turns up).

[dagendresen]

Furthermore, following from an argument that a type specimen found in literature should get basisOfRecord = "HumanObservation" because it was "observed in literature by a human" - then ALL occurrences observed by an aggregator from an IPT should actually get basisOfRecord = "MachineObservation" because the data records were observed from the IPT by a machine...?! I believe that the basisOfRecord should describe the nature of the original evidence entity. If the original basis of record is difficult to infer from the evidence source, then simply put basisOfRecord = "Occurrence". dwc:Occurrence = "An existence of an organism (sensu dwc:Organism) at a particular place at a particular time". dwc:HumanObservation and dwc:PreservedSpecimen are both subtypes of dwc:Occurrence.

[debpaul]

+1 @dagendresen. Seems to me @dagendresen that our challenge is twofold. How to convey the "source" of information we are providing. We are saying "what is your evidence source?" We're trying to use dcterms:type at the highest level first, then have sub-categories. Then, it also confuses people that Occurrence has the possibility to be a observation or refer to a physical specimen of some sort. So, i like your idea that basisOfRecord would have subtypes. Hm.

[dagendresen]

See also Steve Baskaufs blog post with comments to this Darwin Core Webinar: http://baskauf.blogspot.no/2017/03/controlled-values-again.html

[tucotuco]

It seems we have a consensus. For any type specimen record, regardless of where it comes from, dc:type should be PhysicalObject, dwc:basisOfRecord should be PreservedSpecimen (because there had to be one once, right?), and dwc:typeStatus should be as usual. The term dc:references could be the citation pointing to the literature from which the records was derived. If the actual specimen is ever found and published, it would be the responsibility of the data set manager for the literature record to make the relationship with the actual specimen record.