Jul 10, 2015
Metadata query logic and configuration changes resulted in a
much more stable operation.
• internal response times well under 1 second
• timeouts reduced from 100s per day to a few dozen
• outages (now very rare) are mostly self induced (scheduled or not)
Metadata distribution capability expanded
• unleashed our own infrastructure
• channeled slower traffic creating multi-lanes, fast-lane for DOI queries
Stability
Less time spent fighting fires
• get ahead of the curve
• undertake larger changes at a measured pace
• take a closer look at what is in place, take nothing for granted
Stability
things to know about
can now deposit abstracts to crossref
• a way to distribute metadata with abstracts to third parties (via crossref metadata services)
interfaces to crossref:
(a rest compliant API)
oai.crossref.org
doi.crossref.org
api.crossref.org
apps.crossref.org
(core system deposit & query)
(bulk metadata retrieval)
(applications)
Changing the handling of conflict
<query enable-multiple-hits="false" key="key"><article_title match="fuzzy">
Imaging Proteolysis by Living Human Glioma Cells</article_title><author search-all-authors="false">Sameni</author><issn>1431-6730</issn><volume>382</volume><issue>5</issue><year>2001</year><journal_title>Biological Chemistry</journal_title>
</query>
query on
now returns (use to return nothing)
<doi type="journal_article">10.1515/bchm.2001.382.5.785</doi>
(the more recently deposited doi)
so what have we changed?
but that’s not the end of the story
Newer !
query on
returns
(the older doi)
<query enable-multiple-hits="false" key="key"><article_title match="fuzzy">
Imaging Proteolysis by Living Human Glioma Cells</article_title><author search-all-authors="false">Sameni</author>
</query>
<doi type="journal_article">10.1515/BC.2001.094</doi>
That’s not right (discovered when making this presentation)
So what to do
automatically process all unresolved conflicts
• merge family of conflicts into one
• pick a prime and resolve the conflict
473,954 DOIs currently involved in unresolved conflicts
(1,356,650 DOIs at one time in a conflict)
Web callbacks for deposit completion
Deposit XML
HTTP
POSTQ Email
SMTP In Box
The way almost everyone does it now.
• error prone, mailboxes fill up
• attachments or in-line
• mime-types and content format
doi.crossref.org
Is it done yet?http://doi.crossref.org/servlet/submissionDownload?usr=?&pwd=?&type=result&file_name=file.xml
working on it
Done
Web callbacks for deposit completion
An option that some currently use (larger members).
• polling, not bad, but yuckkk
1. You implement an end-point (the receiver of a callback)
2. You give crossref the http(s) URL of the end-point
3. Upon deposit completion we call the end-point
Web callbacks for deposit completion
HTTP headers contain the information
4. You call the retrieval URL to get the log file
Web callbacks
not just for deposit completion logs
• can be used for forward link alerts
• can be used for uploaded batch query jobs
http://help.crossref.org/csv-upload
web deposit
looking forward
relations
Crossref has always supported relations between DOIs and
between DOIs and things. Components and cited-by link DOIs
in very specific ways. CrossMark ‘updates’ define specific
relations (ex. withdraws) between DOIs
In the past year+ we have supported relations between Crossref
DOIs and other DOIs. You can currently deposit a Datacite DOI
as a component of an article OR deposit a Datacite DOI as a
<citation> in an article’s bibliography
relations: will allow a crossref DOI to be linked to any other
item that has an identifier (DOI or other) via a defined set of
relationship types
a DOI can claim a relationship of type X to some other item
what to expect
relations
we will continue to have application specific treatment of relations
• <citations> = isCitedBy/cites relations
• <contributors> = hasAuthor/isAuthoredBy
• <crossmark><updates> = hasErratum, isRetractedBy … etc
new specific treatments will emerge
• see Linked Clinical Trials
relations.xsd (deposit sub-schema) a generalized treatment
• relations between dissimilar thing (hasReview/isReviewOf)
• relations between items of the same work (hasTranslation)
Questions