search

tetherless-world / nanomine-graph

the visualization web app for nanomine project
MIT License
1 stars 4 forks source link

ChemProps entity resolution vs. user-input sample label string #54

Open mdeagen opened 3 years ago

mdeagen commented 3 years ago

ChemProps provides a "standard name" for material compounds, but the conversion is not perfect. We should preserve the string literal name of the compound as input by the curator just in case, as a quality check, similar to how the rdfs:label for a sample is created.

Such instances may need a mechanism to override incorrectly resolved entities (as well as providing the training feedback to ChemProps).

Example: A block co-polymer comprising polystyrene and poly(4-vinylpyridine) is being tagged as Poly(methyl acrylate) image

mdeagen commented 3 years ago

Added a note to the SETLr review notebook in "Chemical name" section.

Proposed modeling scheme using rdfs:label on the component part node (same scheme is currently used for Surface Treatment materials, which are not entity-resolved by ChemProps and do not have rdf:type):

?sample a mm:PolymerNanocomposite ;
        sio:hasComponentPart [ a mm:Polystyrene ; # the "standard name" determined by ChemProps
                               rdfs:label "polystyrene (40 kDa)"] # the string literal from the XML
mdeagen commented 3 years ago

NOTE: Do not change this section (which adds rdfs:label to the entire Class); only add rdfs:label to the component node image