Collection from Union Catalogue missed in lobid

[hagbeck]

In the union catalogue exists collection i.e. for e-books which are not starting with "ZDB". These are missed in lobid.

In http://lobid.org/resources/HT020022241.json there is the collection name "cuvillier". Other collection names can be found in https://service-wiki.hbz-nrw.de/display/VDBE/Produktsigel+und+interne+Selektionskennzeichen

[acka47]

Thanks for the request. Collection information is in 078 in the source data, from the example http://lobid.org/hbz01/HT020022241:

<datafield tag="078" ind1="e" ind2="1">
  <subfield code="a">cuvillier</subfield>
</datafield>

Currently, we are only transforming some dedicated collection information (NWBib, Edoweb, FRL, ZDB) to RDF. We will have to think about a way for adding generic collection information. Maybe with a bnode, so that the resulting JSON could look like this:

{
    "inCollection": [
        {
            "label": "cuvillier"
        }
    ]
}

However, before approaching this it would be nice to know how many different collections are named in 078 and what their tags are.

[dr0i]

However, before approaching this it would be nice to know how many different collections are named in 078 and what their tags are.

We could do this with ES aggs but this would mean to explicitly allow aggregations over labels (meaning: new index config, and performance impact). Doing this on the 36k smallTest reveals just:

{
"key" : "Zeitschriftendatenbank (ZDB)",
"doc_count" : 163
},
{
"key" : "eResource package",
"doc_count" : 129
},
{
"key" : "Nordrhein-Westfälische Bibliographie (NWBib)",
"doc_count" : 123
},
{
"key" : "Edoweb Rheinland-Pfalz",
"doc_count" : 16
},
{
"key" : "Elektronische Zeitschriftenbibliothek (EZB)",
"doc_count" : 10
},
{
"key" : "Fachrepositorium Lebenswissenschaften",
"doc_count" : 10
},
{
"key" : "Rheinland-Pfälzische Bibliographie",
"doc_count" : 8
}
so you may still want to have a complete list on the whole index?

[acka47]

However, before approaching this it would be nice to know how many different collections are named in 078 and what their tags are.

The list with all the collections is linked in the original issue comment:
https://service-wiki.hbz-nrw.de/display/VDBE/Produktsigel+und+interne+Selektionskennzeichen

Here are the 36 IDs that are not an ISIL:

asmi
budri
caso
chiso
Cont
cuvillier
dawsonera
edso 
editlib
elgar
lyell
hade
hirzel
huguenots
iorm
kenso
learntechlib
Logos
mansi
misso
MPSO
mpig
NNg
obp
oso
pearson
philon
luther
smalib
minnso
uncso
vkal
vkv
vogel
wageningen
woodhead
wtm

Here is another proposal to model this in JSON-LD with a newly added identifier key that is mapped to dct:identifier. The advantage would be that we'd also have the label, a disadvantage is that we'd have to maintain a id->label map:

{
    "@context":{
        "@import":"http://lobid.org/resources/context.jsonld",
        "identifier":"http://purl.org/dc/terms/identifier"
    },
    "inCollection":[
        {
            "identifier":"cuvillier",
            "label":"Cuvillier-E-Books"
        }
    ]
}

[dr0i]

Hm, would it make sense to use something like this:

"inCollection":[
    {
        "id":"https://lobid.org/vocabs/cuvillier",
        "label":"Cuvillier-E-Books"
    }

With this approach we could utilize Etikett to set the label as we normaly do. Also, we could provide some more data about the collections if we want. Disadvantage: we have to enhance the vocabs.

[acka47]

+1 But I think we should then use lobid-resources URIs that – in this case – would not resolve:

"inCollection":[
    {
        "id":"https://lobid.org/resources/cuvillier",
        "label":"Cuvillier-E-Books"
    }

[acka47]

Decision after offline discussion:

"inCollection":[
    {
        "id":"https://lobid.org/collections#cuvillier",
        "label":"Cuvillier-E-Books"
    }

For now, these will not resolve but we could add a file at https://lobid.org/collections in the future if needed.