[Esip-semanticharmonization] [EXT]Fwd: I added a section to the ESIP Semantic Harmonization experience paper

[Kai Blumberg]

Thanks Nancy we can take a look at our next meeting!

On Tue, May 3, 2022 at 9:33 PM Nancy Wiegand <wiegand at cs.wisc.edu> wrote:

> *External Email*
>
> I used an old address, I guess.
>
>
> -------- Forwarded Message --------
> Subject: I added a section to the ESIP Semantic Harmonization experience
> paper
> Date: Tue, 3 May 2022 14:30:19 -0500
> From: Nancy Wiegand <wiegand at cs.wisc.edu> <wiegand at cs.wisc.edu>
> To: Gary Berg-Cross <gbergcross at gmail.com> <gbergcross at gmail.com>
> CC: Ruth Duerr <ruth.duerr3 at gmail.com> <ruth.duerr3 at gmail.com>, Kai
> Blumberg <kblumber at mpi-bremen.de> <kblumber at mpi-bremen.de>, Pier
> Buttigieg <pier.buttigieg at awi.de> <pier.buttigieg at awi.de>, Kate Rose
> <Kate.rose at noaa.gov> <Kate.rose at noaa.gov>, Pier Buttigieg
> <pbuttigi at mpi-bremen.de> <pbuttigi at mpi-bremen.de>, Nancy Wiegand
> <wiegand at cs.wisc.edu> <wiegand at cs.wisc.edu>
>
> To describe our latest matching process, I added a section. *You can use
> it or not, or edit, whatever. *
>
> *Aligning ENVO Terms with SWEET*
>
>
> *As part of our harmonization project, after adding cryo terms to ENVO, we
> then mapped those ENVO terms to SWEET for the purpose of aligning the two
> ontologies. To do this, we first populated a spreadsheet with our newly
> entered ENVO terms alongside potential matching terms in SWEET and, for
> each term, using a human curated approach versus machine learning or
> automated string matching, determined a potential relationship expressed
> using SKOS mappings. As a group endeavor (e.g., 4-5 people), we looked up
> definitions using BioPortal (https://bioportal.bioontology.org/search
> <https://bioportal.bioontology.org/search>) for SWEET and a lookup service
> for ENVO (e.g., https://www.ebi.ac.uk/ols/ontologies/envo
> <https://www.ebi.ac.uk/ols/ontologies/envo>), and had a discussion, also
> noting where the term occurred in each class hierarchy. *
>
>
>
> * SWEET, however, does not have definitions for many of its terms. So,
> even if the ENVO and SWEET terms matched string-wise, we refrained from
> calling it an 'exact match'. An example is the ENVO term 'ice sheet' and
> SWEET's 'IceSheet' for which we chose 'close match' (skos:closeMatch)
> because of a lack of a SWEET definition. We noted this by recording a
> comment that ice sheet is a candidate for an exact match if SWEET were to
> add ENVO's definition to SWEET's term at some future time. Our intent now
> is not to presume exact definitions without an existing definition in
> SWEET. As mentioned, we also looked at class hierarchies to determine
> closeness of match. It was found that some very similar terms, even nearly
> exact strings, had a different basic upper parent class in each ontology,
> such as 'freezing rainfall' in ENVO and 'FreezingRain' in SWEET. We
> recorded this match as close instead of exact and commented that ENVO
> classifies the concept as a type of 'process', but SWEET has it as a type
> of atmospheric precipitation which is under 'phenomena'.  An example of a
> related match (skos:relatedMatch) is ENVO's 'snow layer' and SWEET's
> 'SnowCover'. Further, SWEET has two instances of SnowCover, one under
> 'phenHydro' (hydrospheric phenomenon) and the other under
> 'propSpaceThickness' (a thickness). To handle this, we made two copies of
> ENVO's 'snow layer' and mapped each as related to SWEET's terms. An example
> of a broader match (skos:broadMatch) is ENVO's 'planetary cryosphere'
> mapped to SWEET's 'cryosphere'. *
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.esipfed.org/pipermail/esip-semanticharmonization/attachments/20220504/172e36b2/attachment.htm>