Custom Query (900 matches)

For long time the resourceClass facet contained around |346| entries.

It was identified that the two main collections contributing to this list where Oxford Text Archive |154| and COllections de COrpus Oraux Numeriques (CoCoON ex-CRDO) |64|. Moreover these values very coming from a generic <dc:type> field and were rather subject or genre categorisations, next to a qualified <dc:type dcterms-type="DCMIType"> field featured in all records with type-classification adhering to DCMIType vocabulary. Also, in case of OTA the field-values delivered through OAI were partly mingled - concatenating multiple value into one string (see #806). Therefore the post-harvesting OLAC2CMDI script was adapted to only consider <dc:type dcterms-type="DCMIType"> field (if there is one) and ignore the unqualified <dc:type> field. see: ​https://github.com/clarin-eric/metadata-conversion/commit/56426933dc1bfee9a9ca36801986ecb5244f2d80 [Menzo]

By this measure, we were able to reduce the list of values in resourceClass facet to |138|

The last time information was sent to stats.clarin.eu for the FCS aggregator was March 2017. It should be reactivated again.

There are numerous records without any resource proxy.

The VLO query for that: ​https://vlo.clarin.eu/search?q=resources:0+-_searchPageRef:*+-_hasPartCount:*+-_contentSearchRef:* yields around 20.000 records (|21171| as of 2018-02-20)

out of these from CLARIN centres: ​https://vlo.clarin.eu/search?q=resources:0+-_searchPageRef:*+-_hasPartCount:*+-_contentSearchRef:*&fqType=dataProvider:or&fq=dataProvider:CLARIN+Centres (|8200| as of 2018-02-20)

Following is the list of offending collections: ​http://alpha-vlo.clarin.eu/solr/vlo-index/select?facet.field=collection&facet=on&q=resources:0%20-_searchPageRef:*%20-_hasPartCount:*%20-_contentSearchRef:*&rows=0

On the one hand, we agree that records without resourceproxies have little value for the user, but there seems to have been a decision (and a good reason) to still keep those in.

This could be resolved (or at least made explicit to the users) by a dedicated availability facet.

For now we at least want to keep track of this phenomenon, here.