How do we know what people are doing with our stuff?

[tomcrane]

(raised after discussion with @jennpb)

Summary

An institution publishes IIIF resources. On its own site, with its own user interface / viewer(s) / etc, it can capture as much visitor interaction with that content as it wants, through a combination of tools such as Google Analytics and server access log analysis.

How does the publisher know what people are doing with those IIIF resources - collections, manifests, images, text content, other annotations - outside of this controlled context?

Long version

https://docs.google.com/document/d/1OrKQlj4tn4-Uw7IdY2k4abtnLHEpT_2GC1O92n2seDg/edit#

(but please comment here rather than the document!)

[glenrobson]

What is the current support for referrer? Back in the day when I worked at NLW we used to get useful information in the access logs on the site that usually hosted Mirador or the UV. With the switch to https we got less useful results but looking around on google it seems that as long as the manifest is published over https it should receive the referrer header unless the user has decided to hide it.

[jbhoward-dublin]

There are concerns about recording provenance, resource discovery and user experience in the full document. I'd like to pick up on a specific reference related to discovery and provenance--use of LDN notifications, using the Rerum Inbox as an exemplar service. I find this a compelling approach, but success is proportionate to community buy-in both for use of this kind of solution, and for assurance of redundancy of these kinds of data across different jurisdictions.

Another observation, that is tangential to what I read as the major concerns in Tom's document: With re-use, resources can become separated from important value-added services provided by the owner of a resource (so also an interoperability issue). The owner of a resource may also want to know about value-added services applied to their content by third parties.

For example: A user of a IIIF client in Dublin annotates a manifest made available by a library in Berlin, and those annotations are stored in the local annotation store in Dublin; this annotation store also provides a IIIF content search endpoint. This annotation store is searchable, but the manifest of the Berlin item references only its own local IIIF content search service; the IIIF client enables the annotation just created to be viewed, but does not enable its content to be searched, potentially creating confusion and frustration.

In this same scenario, if a notification is also sent to the LDN Inbox, then the annotation can be potentially discovered by third parties anywhere, including the owner of the Berlin resource. However, the issue of a lack of awareness of the IIIF content search endpoint where the content is indexed (in Dublin) remains.

Should then, the LDN notifications not only reference the URI of the annotations/supplementary content associated with the manifest, but also the associated content search endpoint?

Finally, if more than one content search endpoint holds data pertinent to a manifest, and different levels of support for the OA- or W3C annotations specifications have been implemented by the two search services, how does the IIIF client negotiate this, and what does the user experience look like?

[tomcrane]

(posted by Jason Evans in Wikimedia channel):

Jason [7:34 PM] http://openmlol.it/Media/Search?Publisher=1456 Looks like we have our first example of reuse

And this is within a couple of days of IIIF manifests on Wikidata!

[aisaac]

@tomcrane @azaroth42 @mattmcgrattan I'm catching up with older msgs... I see the 'discovery' label you've added. How should we handle the discussion on this? Should it be an issue in the specific Discovery project, instead?

[jpstroop]

I could see a higher-level document (not really a spec; maybe recipe?) regarding IIIF and analytics best-practices? (Eds. discussed on 15 Apr. 2019)

[azaroth42]

Could define a service that gets called with some structured data for analysis.