Author: Hisham Shahtout - atypon systems
Introduction
CrossRef allowes records that have identical metadata to be inserted into
the CrossRef database. However, these records are marked as records in conflict.
Submitters of the metadata are notified of these conflicts so a further
action can be taken to resolve them. This document outlines the different scenarios
of when conflicts are created or automatically resolved as well as
the different actions that can be taken by a CrossRef administrator
to resolve them.
Definitions
- DOI conflicts: DOI conflicts occur when two or more DOIs
have the same query-able metadata and no article title and/or sequence OR
when they have the same query-able metadata and same article title and/or sequence.
Conflicts can be resolved in one of three ways:
- Automatically resolved through a subsequent record submission.
- Aliased Making one DOI an alias of a primary DOI. (See DOI alias and Primary DOI below)
- Resolved Simply resolving the conflict without marking a primary or
alias DOI (i.e. records point to two different citations indeed). In this case,
Metadata queries will not resolve to either DOIs
- DOI ambiguity: DOI ambiguity occurs when two or more DOIs
have identical query-able metadata but different article title and/or sequence.
Whenever a metadata query resolves to two or more ambigious DOIs
no results are returned.
- DOI Alias: DOI aliases act as pointers. An administrator
can resolve a conflict between two DOIs by aliasing them.
As a result, the DOIs are aliased in the handle system
and the conflict is marked as resolved. As part of the
aliasing of two DOIs the administrator must designate
one DOI as primary. Once a DOI is aliased the metadata
pointed to by that DOI cant be updated. Also, if an aliased
DOI is queried through a DOI query the metadata contained
by the primary DOI is returned.
- Primary DOI: In a set of aliased DOIs there can be only one primary
DOI that will be returned when the DOI metadata is queried.
Use case scenarios
Use case I
Publisher submits metadata (MD) in the following sequence:
MD - DOI1 - Full_Text
MD - DOI2 - Abastract_Only
MD - DOI3 - Bibliographic_record
In this case even though the metadata is identical for all three records. The type
of the content distinguishes the records and therefore no conflicts are flagged
Use case II
Publisher submits metadata (MD) in the following sequence:
MD - DOI1
MD - DOI2
The system will allow DOI2 in (regardless of its timestamp) but will flag a conflict X
A metadata query on MD returns nothing.
The publisher resolves conflict X by marking DOI1 as a primary thus making DOI2 and alias
A metadata query on MD returns DOI1
Publisher submits: MD2-DOI1 (MD2!=MD1)
The system updates MD to MD2 for DOI1 but not for DOI2
A metadata query on MD2 return DOI1
A metadata query on MD returns DOI1
A DOI query on DOI1 returns MD
A DOI query on DOI2 return MD
Use case III - Auto resolution
Publisher submits metadata in the following sequence:
MD - DOI1
MD - DOI2
The sytem creates a conflict X between (DOI1,DOI2)
Publisher submits an update for DOI1
MD2 - DOI1 (where MD2!=MD)
In this case X is automatically resolved.
Use case IV - PAP (Publish Ahead of Print)
There are cases where publishers send DOI records for articles that
have not been printed yet and therefore dont have page numbers.
Publisher submits metadata (MD) in the following sequence:
MD1 - DOI1 (No page number)
MD2 - DOI2 (No page number)
MD1P - DOI1
MD2P - DOI2
In this case, MD1 and MD2 will be in conflict that will be automatically resolved
when MD1P-DOI1 is submitted. Because of the potentially large number of
these cases, it is recommended that the publisher put the DOI in the publisher_item/sequence
element for DOI1 to distinguish it from the metadata for DOI2.
Use case V - Multiple conflicts
Publisher submits metadata in the following sequence:
MD - DOI1
MD - DOI2
System flags conflict X1 between (DOI1,DOI2)
Publisher submits same metadata for a different DOI
MD - DOI3
The system flags a conflict X2 between (DOI1,DOI2,DOI3)
Use case VI
Once a citation in conflict is resolved. It is never flagged
again as a conflict
Publisher submits metadata in the following sequence:
MD - DOI1
MD - DOI2
System flags conflict X1 between (DOI1,DOI2)
admin resolves X1 by making DOI1 primary.
MD - DOI3
A new conflict: X2 is generated between (DOI3,DOI1) only.
Note that (DOI3,DOI2) are not flagged as in conflict
because DOI2 is an alias of DOI1.
Use case VII
DOI updates can result in new conflicts as well as resolving
existing ones. For example, publisher submits same metadata
in the following sequence:
MD - DOI1
MD - DOI2
MD2 - DOI3
System flags conflict X1 between (DOI1,DOI2)
Publisher submits: MD2 - DOI2
The system in this case automtically resolves X1 since
DOI2 is updated. However, a new conflict X2 is created
between (DOI2,DOI3)
Use case VIII
Note that as a result of updating DOIs. A conflict would
be considered completely resolved only if there is one
citation left in the conflict or if that citation
was the source of the conflict Here is an exmaple:
MD - DOI1
MD - DOI2
System flags conflict X1 between (DOI2,DOI1)
Publisher submits: MD - DOI3
System flags conflict X2 between (DOI3,DOI2,DOI1)
Publisher submits: MD2 - DOI1
In this case, DOI1 will be marked as auto resolved
and X1 will be fully resolved. X2 however is not
fully resolved since (DOI3,DOI2) are still in conflict.
Use case IX
Updating a DOI can result in auto resolving
a conflict in full if the DOI was the source of
the conflict. For example:
MD - DOI1
MD - DOI2
System flags conflict X1 between (DOI2,DOI1)
Publisher submits: MD - DOI3
System flags conflict X2 between (DOI3,DOI2,DOI1)
Publisher sumits: MD2 - DOI3 (MD2!=MD)
In this case X2 is fully resolved because DOI3 was
the source of the conflict.
Conflict notifications
Publisher are notified of conflicts through the submission result XML
that is emailed back. For example:
<![CDATA[
<record_diagnostic status="Warning">
<doi>10.5555/DOI1</doi>
<msg>Added with conflict</msg>
<conflict_id>34</conflict_id>
<dois_in_conflict>
<doi>10.5555/DOI_IN_CONFLICT</doi>
</dois_in_conflict>
</record_diagnostic>
]]>
Indicates that the DOI 10.5555/DOI1
conflicted with an existing CrossRef metadata
when added to the system.
If a DOI record resolved an existing conflict, the conflict ID(s) are sent back
in the XML as well. For example the following XML result portion indicates that
10.5555/DOI2 created a new conflict 134 and auto resolved two existing ones:
17 and 21
<![CDATA[
<record_diagnostic status="Warning">
<doi>10.5555/DOI2</doi>
<msg>Added with conflict</msg>
<conflict_id>134</conflict_id>
<dois_in_conflict>
<doi>10.5555/DOI2_IN_CONFLICT</doi>
</dois_in_conflict>
<resolved_conflict_ids>17,21</resolved_conflict_ids>
</record_diagnostic>
]]>
