The current issue and full text archive of this journal is available at

[Link]/[Link]

Purdue
The Purdue University Research University
Repository
HUBzero customization for dataset publication
and digital preservation 15
Carly C. Dearborn, Amy J. Barton and Neal A. Harmeyer Received 1 July 2013
Purdue University Libraries, Purdue University, West Lafayette, Indiana, USA Revised 18 July 2013
Accepted 18 July 2013

Abstract
Purpose – The purpose of this case study is to discuss the creation of robust preservation
functionality within PURR. The study seeks to discuss the customization of the HUBzero platform,
composition of digital preservation policies, and the creation of a novel, machine-actionable metadata
model for PURR’s unique digital content. Additionally, the study will trace the implementation of the
Open Archival Information System (OAIS) model and track PURR’s progress towards Trustworthy
Digital Repository certification.
Design/methodology/approach – This case study discusses the use of the Center for Research
Libraries Trusted Repository Audit Checklist (TRAC) certification process and ISO 16363 as a rubric
to build an OAIS institutional repository for the publication, preservation, and description of unique
datasets.
Findings – ISO 16363 continues to serve as a rubric, barometer and set of goals for PURR as
development continues. To become a trustworthy repository, the PURR project team has consistently
worked to build a robust, secure, and long-term home for collaborative research. In order to fulfill its
mandate, the project team constructed policies, strategies, and activities designed to guide a
systematic digital preservation environment. PURR expects to undertake the full ISO 16363 audit
process at a future date in expectation of being certified as a Trustworthy Digital Repository. Through
its efforts in digital preservation, the Purdue University Research Repository expects to better serve
Purdue researchers, their collaborators, and move scholarly research efforts forward world-wide.
Originality/value – PURR is a customized instance of HUBzerow, an open source software platform
that supports scientific discovery, learning, and collaboration. HUBzero was a research project funded
by the United States National Science Foundation (NSF) and is a product of the Network for
Computation Nanotechnology (NCN), a multi-university initiative of eight member institutions. PURR
is only one instance of a HUBzero’s customization; versions have been implemented in many
disciplines nation-wide. PURR maintains the core functionality of HUBzero, but has been modified to
publish datasets and to support their preservation. Long-term access to published data are an essential
component of PURR services and Purdue University Libraries’ mission. Preservation in PURR is not
only vital to the Purdue University research community, but to the larger digital preservation issues
surrounding dynamic datasets and their long-term usability.
Keywords Institutional repositories, Data sharing, Digital preservation, Metadata, Data curation
Paper type Case study

Data sharing and open access are no longer simply buzz phrases in the scientific and
publishing communities. Benefits to data sharing and reuse include increased OCLC Systems & Services
Vol. 30 No. 1, 2014
collaboration, interdisciplinary innovation, and new solutions to pervasive social pp. 15-27
problems. The life and use of scientific data should extend beyond its original purpose. q Emerald Group Publishing Limited
1065-075X
The new and sometimes competing demands placed on data have created what many DOI 10.1108/OCLC-07-2013-0022
OCLC call the “data deluge.” Coupled with the sheer bulk of data created in modern research,
30,1 the rapid advances in technology and tools, and the interdisciplinary research
objectives can make open access a challenging objective (Faniel and Zimmerman,
2011).
While the benefits of open access seem clear, the logistics surrounding
sustainability are less so. A key factor in developing and sustaining open access to
16 data are addressing issues surrounding preservation and data management. The
emerging field of data curation combines the scalability of data management with the
commitment to long-term preservation. An effective data management plan will factor
in issues such as use of open standards for file formats, well-formed metadata, and
information management literacy with the goal of viable future access (Lee and Tibbo,
2007). The strategies of data management are currently being addressed more and
more by university libraries and institutional repositories. These bodies are
increasingly providing assistance with data creation, management, curation, and
ultimately, preservation (Cragin et al., 2010). Purdue University Libraries specifically
sought to operationalize this narrative by providing a platform on which Purdue
researchers can receive data management support from subject librarians, fulfill the
data management requirements of most funding agencies, take steps toward long-term
data preservation, and provide immediate access to their research data.
Incentives and mandates for data sharing are changing the landscape of university
libraries, especially at Purdue – a leader in science, technology, and engineering
research. More and more federal funding agencies require data management plans as
part of their grant awarding process. In January of 2011, the National Science
Foundation (NSF) began requiring a two page data management plan while other
agencies have been requiring them for much longer. The National Institutes of Health
required its grant applicants to take measures towards data management as early as
2003. The National Institution of Justice requires awardees to submit a data-archiving
policy 90 days prior to the end of a funded project (Witt, 2012).
The increased focus on data management and data sharing by these major funding
agencies has necessitated a sea change in university library core functions. Librarians
have the unique training to help researchers handle their data and prepare sustainable
approaches to its management. It was this understanding that prompted the Purdue
University Dean of Libraries, the Purdue University Vice President of Information
Technology, and Purdue University Vice President for Research to plan the
development of a campus-wide data management platform using HUBzero software.
This group of Purdue administrators created the Purdue University Research
Repository Working Group in March of 2011. This group represented the major
stakeholders from the University community and included experts from Purdue
Libraries, Sponsored Program Services, and Information Technology at Purdue (ITaP).
The Working Group began planning and developing the data management platform,
now realized as the Purdue University Research Repository (PURR). Invested parties
from the Libraries gradually formed the PURR Project Development Team (PURR
Team). This team includes librarians, archivists, software engineers, and graduate
students and is largely responsible for the continued development and maintenance of
the repository.
This case study will discuss the progress of the Working Group and the PURR
Team as they continue to develop PURR’s platform and preservation infrastructure.
This discussion will include the creation of guiding policies and procedures, plans to Purdue
place those policies into action, and the unique metadata which describes and informs University
these actions. While PURR is operational, many of the components discussed within
this case study are still in development and have not been fully implemented within
PURR’s environment.

Customization of hub platform 17

PURR is a customized instance of HUBzerow, an open source software platform
developed at Purdue University which supports scientific discovery, learning, and
collaboration. HUBzero’s concept originated from the US National Science
Foundation’s Network for Computational Nanotechnology (NCN). NCN’s unique
system, [Link], allowed researchers, educators, and professionals to share
resources and collaborate on issues surrounding nanotechnology. Built on the open
source LAMP (Linux Apache, MySQL, and PHP) platform and the Joomla! content
management system, nanoHUB allows for seamless integration of nanoHUB’s many
learning tools (Klimeck et al., 2008). HUBzero is a generic version of nanoHUB’s cyber
infrastructure and is available for customization in other scientific disciplines. PURR is
just one instance of a HUBzero customization.
The HUBzero platform seemed an obvious choice for the PURR Working Group. Its
home-grown development allowed for easy code integration and provided a local
technical support base. Also, its collaborative and publishing services matched the
Working Group’s strategic vision for PURR. PURR allows for collaboration within a
project, or a dedicated working space. Projects provide space for data, wikis for project
tracking and collaboration, and to-do lists. Once a producer creates a project, he or she
can invite other collaborators to the project. PURR’s platform allows producers to
publish data for public access and discovery. A dataset publication is the term used to
describe the published files and their associated metadata.

PURR digital preservation policy

While work on the technical infrastructure was underway, work also began an
over-arching mandate for preservation of Purdue-affiliated datasets. A sub-group of
the PURR Working Group, comprised of experts in digital initiatives, data curation,
and digital preservation, co-authored the PURR Digital Preservation Policy[1]. The
Preservation Policy provides a basic framework for the preservation direction and
operations within PURR. At its core, the policy states that PURR, as part of Purdue
Libraries, “is responsible for identifying, securing, and providing the means to
preserve and ensure ongoing access to selected digital assets.” A preservation priority
structure was developed based on dataset-to-publication relationship, ongoing
teaching value, and long-term research value. The PURR Preservation Policy serves
as an internal guide in conjunction with recognized digital preservation standards for
preservation strategies and actions taken on PURR research datasets.
Part of the underlying foundation of the PURR service model is the promise of
dependable, long-term storage of data publications. In order to confidently assure both
the designated user community (Purdue’s“faculty, staff researchers, and graduate
students at Purdue University campuses and their immediate collaborators”),[2] and
the larger academic community that PURR is a desirable, secure data repository, the
PURR team charted a course to become a trustworthy repository via the Center for
OCLC Research Libraries Trusted Repository Audit Checklist (TRAC) certification process.
30,1 The TRAC assessment offered a roadmap to digital preservation infrastructure
through its rigorous rubric. Within TRAC, considerations for metadata, software and
hardware migration, digital object reliability (fixity), storage, disaster preparation, and
much more are specified as integral to any well-formed digital preservation
environment. The PURR team turned to ISO 16363 following its formal recognition –
18 ISO 16363 is a direct descendant of TRAC and the international standard for
Trustworthy Digital Repositories – as the guide for holistic preservation practices.[3]
However, before any datasets could be ingested, code written, or preservation actions
could take place, increasingly-granular policies had to be written, vetted, and added
into the functional environment. Armed with ISO 16363 and internationally-recognized
archival best practices, the PURR team began to build a digital preservation
environment from the ground-up.

Developing policies and strategies

In early 2012, a digital archivist and metadata specialist joined the PURR Team to
begin development of the policies that would inform the digital preservation strategies
and actions of a fully-functional PURR. Taking into account the PURR Preservation
Policy as well as ISO 16363, the digital archivist created two foundational preservation
documents, the Preservation Strategic Plan and the Preservation Strategies.[4] The
Preservation Strategic Plan outlines the overall objectives sought by the repository in
order to provide long-term digital preservation. For example, one objective is designed
to “ensure digital content is stored securely, redundantly, and remotely.” However,
preservation is useless without accessibility, so the objectives also reiterate that
long-term accessibility is part of any fully-functional digital repository. As such,
another objective is to “develop and implement strategies for planning for new formats
and technologies.”
Objectives are useful, but to accomplish anything, action is necessary. For each
published dataset, the preservation activities documented in the Preservation Strategic
plan must take place in order to ensure present-day and long-term use of the data.
These activities are in accordance with archival best practice and relate to the
provenance, intellectual content, metadata, archival information package (AIP), fixity
information, and rights. The Preservation Strategic Plan also states that PURR “will be
continually monitored and updated.” Trustworthy repositories consistently make
efforts to improve, and PURR is no different. In order to align itself with the most
modern and comprehensive preservation techniques, PURR has and will continue to
evolve its means and methods by seeking input from a variety of sources.
The second foundational preservation document, Preservation Strategies, details
the strategies and actions each published dataset undergoes. For each dataset, PURR
will undertake one of three preservation levels – bit-level preservation, full
preservation, or no preservation. Nearly all datasets will undergo bit-level activities.
Bit-level preservation will occurs if the object format is unrecognized or unsupported,
or if full preservation is not possible at the time of ingest. In PURR the object file type is
checked at ingest against the PRONOM file format registry using DROID (Digital
Record Object Identification) to determine its level of support. (The DROID tool is
discussed in greater detail later in this article.) Full preservation allows for the
preservation of the original intellectual and structural content of the object. In full
preservation, objects will receive bit-level treatment but also undergo transformation, Purdue
normalization, or migration actions at ingest and whenever appropriate during the University
preservation life cycle. In extreme cases, datasets found to be corrupt or of potential
danger to the repository will receive no preservation. Such objects will not remain
within PURR.
More specific policies and preservation actions developed from the Preservation
Strategic Plan describe categories of preservation support, preferred file formats, and 19
the possible preservation actions for each. To illustrate, consider that a dataset using
an open source file format which is supported by multiple software platforms is a
better candidate for long-term preservation than a proprietary format with low
adoption rates. The three levels of preservation within PURR will allow for flexibility
at the point of ingest and during the life cycle of datasets. For example, a dataset
ineligible for full preservation at its time of ingest may later become supported. New
file formats, file support, and other new developments in the digital preservation field
will continue. With the advent of new standards, PURR will adapt to strategically and
programmatically implement new policies, strategies, and actions.
PURR’s Preservation Strategies also lists the specific preservation actions which
take place during the life cycle of a dataset in order to maintain its integrity and the
renderability. These actions include: file format recognition, normalization, fixity
checks, migration, and file backup.
File format recognition will take place once a dataset has been successfully
published in PURR. At ingest each dataset is analyzed to determine its file format.
PURR uses DROID, developed by The National Archives of the UK, to perform batch
file analysis and identification. DROID, an open source Java tool, is able to provide
detailed information such as file age, size, last modification date, or duplicate
information. DROID then links that identification to a central registry, such as
PRONOM, for technical information about each format.[5] PRONOM is also a product
of the National Archives of the UK. Once the initial file format of the dataset is checked,
this information will be documented for use in potential transformation, migration, and
fixity checks during the entire life cycle of the dataset in PURR. Formats recognized by
the DROID instance receive full preservation, while unrecognized formats undergo
bit-level preservation until such time that the dataset may be further analyzed.
Normalization may take place to align the dataset with the current preservation
standards. As each dataset enters PURR, items not structured within an established
preservation file format may undergo a process to transform and normalize the object
into an analogous long-term preservation format. For example, a text document will be
normalized into a PDF. This process is undertaken to ensure the continued
representation of the dataset throughout its life cycle. Normalization events are then
recorded in preservation metadata (PREMIS) associated with the dataset throughout
its life cycle.
Fixity checks are completed for all data within PURR on a regularly-scheduled basis
to ensure no loss of data has occurred. Fixity checking is done by comparison of hashes
generated at the time of ingest or after another preservation event. In the case a fixity
check uncovers file degradation; the corrupted data will be removed and replaced with
its uncorrupted counterpart at mirror sites. Hashes and fixity checks will be recorded
in metadata associated with the dataset.
OCLC Migration to other preservation formats will take place over the life of many data
30,1 objects. As such, PURR will continue to monitor content for potential file format
obsolescence. If circumstances dictate data within PURR is at risk of obsolescence, the
content will undergo migration to a new file format more conducive to its preservation.
This will be done to bring PURR in line with rapidly evolving archival best practices
and to ensure long-term preservation and access. This migration may include
20 “upgrading” datasets to a newer version of the same format or transforming datasets
into a completely new file structure. Migration events are also recorded in metadata
associated with the object.
File backup is done for all content within PURR. All data within PURR is fully
duplicated on a regular basis to prevent catastrophic loss of information. Purdue
recently became part of the MEtaArchive Cooperative and in time, information will be
backed up and mirrored at other geographically-dispersed sites to provide a means of
recovery in case of disaster. This is addressed later in this article.) File duplication also
prevents data loss in cases of data corruption detected through regular fixity checks. In
recovery events, actions taken will be recorded in metadata associated with the object.
Periodic checks of the datasets within PURR will determine what processes need to
be enacted. Actions may be taken as many times as necessary to preserve the objects
and datasets for the designated community. These preservation actions are the
building blocks for the PURR preservation service model.

OAIS model in hub environment

Much like the customization of HUBzero into the PURR platform, PURR customized
the Open Archival Information System (OAIS) Reference Model to fit its specific
service model. An ISO international standard, OAIS provides a conceptual framework
which defines the major functions involved in establishing an archival repository.
OAIS was designed as a versatile and customizable model and does not suggest any
particular implementation.
Figure 1[6] shows PURR’s designed workflow mapped to the OAIS Reference
Model. The producer interaction with PURR begins with the creation of a project space.
Prompted by a need for data management, preservation, or a wish to disseminate
research, a producer can establish a Project Space on the PURR web site
([Link] invite colleagues to collaborate, and eventually, choose to
publish a dataset through PURR’s online publishing platform. Once the producer is
ready to publish a dataset, online forms prompt the producer to provide descriptive
information for the dataset and to select a license for terms of use. This information in
addition to the actual dataset comprises PURR’s Submission Information Package
(SIP). The producer-supplied information is included in the descriptive metadata
section which will be discussed later.
The final step of the publication process allows the producer to review and save the
item or review and submit for publication. Selecting “submit” will mint a Digital Object
Identifier (DOI), a unique and persistent identifier, which will facilitate data citation
and access.[7] Selecting “submit” will also prompt the Gatekeeper function. This
function represents the review process that PURR’s repository specialist, or the subject
librarian associated with the producer’s discipline, conduct to verify that the dataset is
an appropriate addition to PURR. The Gatekeeper reviews the SIP to verify that the
producer provided adequate metadata, the content information is complete, and the
 Purdue
University

21

Figure 1.
PURR’s OAIS workflow
OCLC entire submission meets with PURR’s collection policy. If not, the Gatekeeper will send
30,1 the submission back to the producer with comments for review or suggestions to place
in another repository.
Once the Gatekeeper process approves the SIP, the process to create an Archival
Information Package (AIP) begins. While the AIP creation tool has been written and
tested, it is still in development and is not fully integrated with the live PURR site.
22 PURR uses the Library of Congress BagIt specifications to package the dataset and its
associated metadata – the descriptive information and the preservation description
information. BagIt is a hierarchical file packaging format used primarily for storage
and transfer of preservation-quality digital content. BagIt “bags” consist of a
“payload,” or the dataset encapsulated in the bag, and “tags,” the metadata used to
record bag transfer and storage (Boyko et al., 2009). The “bags” are read-only and
cannot be altered once serialized.

Customized metadata in PURR

The HUBzero platform was not developed with metadata or preservation in mind.
Therefore, a custom metadata implementation was necessary. after joining the PURR
Team, the metadata specialist reviewed metadata standards to identify a “best fit”
standard given the purpose of metadata in the PURR environment. The purpose is
threefold: describe the dataset, identify dataset ownership and access conditions, and
generate robust preservation metadata for long term preservation. At the conclusion of
the standards review it was determined there was no single standard that met PURR’s
metadata requirements. However, by weaving together several standard schemes, all
the requirements could be met. The metadata specialist developed PURR’s metadata
scheme using the following standards:
.
Metadata Encoding and Transmission Standard (METS)[8].
.
Dublin Core Metadata Initiative (DCMI) Metadata Terms (dcterms)[9].
.
Metadata Object Description Schema (MODS)[10].
.
Preservation Metadata: Implementation Strategies (PREMIS)[11].

These standards, chosen due to their acceptance by the information management and
academic communities, will continue to undergo support for the foreseeable future.
More specifically, METS was chosen because it was designed to be extended by
incorporating other defined metadata standards within the descriptive metadata and
the administrative metadata container. METS acts as the wrapper into which the other
standards are embedded. DCMI dcterms was selected for the descriptive metadata in
anticipation of Open Archives Initiative Protocol for Metadata Harvesting (OAI-PMH)
implementation within PURR. The MODS standard was selected to identify ownership
and access conditions of the published dataset. Lastly, PREMIS was selected to
support PURR’s requirement for long term preservation of published datasets.
In order to achieve a high-level of dataset discoverability, the descriptive metadata
must be complete and indexed. Datasets are described using the DCMI Metadata
Terms (dcterms) standard. Through the project creation and dataset publication
processes, the producer fills out online forms that capture descriptive metadata. The
fields on the forms include Project Name, Project Alias, Title, Synopsis, Abstract,
Authors, Tags, License, and Release Notes. The Project Name and Project Alias fields
are populated by the producer at the time a project is created. During the publication Purdue
process, the Synopsis field captures a succinct description of the dataset or the research University
that produced the dataset. The Abstract field provides more space for the producer to
describe the dataset and/or the research project. The Authors field is repeatable and
can capture a single author or the primary author and additional authors and/or
contributors. The Tags field captures keywords that are indexed for dataset searching.
The producer typically provides natural language terms in the Tags field. 23
Once submitted, the publication is queued until a subject librarian reviews the
publication to ensure appropriateness, checks grammar and spelling, and adds
controlled vocabulary subject terms in the Tags field. The natural language and
controlled vocabulary terms in the Tags field, along with the other bibliographic fields
such as Title, Author, Abstract, etc. enrich the description and are indexed for
searching and discoverability. The License field is a dropdown list of Creative
Commons[12] licenses reviewed and approved by the PURR Steering Committee.
Finally, the Release Notes field is a place for the producer to include any notes with
regards to file(s) descriptions, and any other pertinent information about the dataset or
research methods.
Once the dataset is approved and published, the descriptive metadata values
provided by the producer and subject librarian are stored in tables in PURR’s database.
The approval and subsequent publication of a dataset will eventually trigger an AIP
Creation Tool to run. The Tool first process creates the AIP BagIt bag and then begins
the process to dynamically generate serialize the metadata in a well-formed, validated
Extensible Markup Language (XML) file to be included in the completed AIP. First, the
METS metadata wrapper is written to the file. Next, the AIP Creation Tool maps the
producer, librarian, project and system generated descriptive metadata values to
dcterms elements and inserts the elements in the METS descriptive metadata
container. The mapping is shown in Table I.

PURR field dcterm field Value generated by

Authors (primary) , dcterms:creator . Producer

Authors (secondary/contributor) , dcterms:contributor . Producer
, dcterms:date . System (ISO 8601 format)
Project name , dcterms:description . Producer (project creation)
Project alias , dcterms:description . Producer (project creation)
Publication state , dcterms:description . System (draft/review/published)
Publication version , dcterms:description . System
Abstract , dcterms:description . Producer
Notes , dcterms:description . Producer
Synopsis , dcterms:description . Producer
, dcterms:format . BagIt System (AIP)
, dcterms:identifier . doi System (DataCite)
, dcterms:publisher . System
License , dcterms:rights . Producer
Tags , dcterms:subject . Producer/subject librarian
Title , dcterms:title . Producer Table I.
, dcterms:type . dataset System Field mapping
OCLC The AIP Creation Tool will then generates the PREMIS preservation metadata for each
30,1 file comprising the dataset. In the METS administrative metadata container a technical
metadata section is generated for each file. Every file is a PREMIS object with a unique
identifier. The dataset files are rendered read-only in the AIP so they may not be
altered in preservation status. In the PREMIS metadata, “fixity-checked” and
“read-only” behaviors are recorded as well as the content is a “primary” file or
24 “secondary” file. Composition level, which indicates compression and encryption is
determined and recorded. File size and format are identified and recorded in the size
and format elements. As proscribed in the preservation documentation preservation
level is determined based on file type recognition via DROID. If the file format is a
supported format in PURRs preservation policy, the preservation level recorded in the
metadata is “full.” Otherwise, a file will have “bit-level” preservation recorded.
The PREMIS rights section follows the technical metadata section. PURR associates
two licenses with a published dataset, the Terms of Deposit license and a Creative
Commons license. The Terms of Deposit license grants permission for PURR “to use,
duplicate and distribute the Work and to transfer the Work to any format or medium
now known or later developed for archiving, preservation and access.”[13]. The
Creative Commons license is applicable to the dataset inclusive of all files therein. The
AIP Creation Tool accesses the name of the license selected by the producer on
submission (e.g. CC0 – Creative Commons) and the location of the license terms
(e.g. [Link] from the PURR database. In
addition to the license information, the Tool assigns the first PREMIS Agent.
According to the PREMIS Data Dictionary,[14] an agent is a person, organization, or
software associated with rights management and preservation events such as file
validation, repository ingest, migration, etc. Each agent has a unique identifier and is
then recorded in the rights section. The license information is then written to the
metadata XML file.
The next AIP Creation Tool process will generate the digital provenance metadata.
The MODS standard is included in the provenance metadata to record the name of the
primary contact of the dataset, a contact email, organizational affiliation, and, most
importantly, the access conditions for the dataset. On publication, if the dataset is
immediately accessible to the public, the value “publicly accessible” is recorded.
However, PURR allows for a publisher to set an embargo with a specific release date. In
the case of an embargo, the value will be, for example, “embargoed until 1/1/2014”. It is
important to capture this information for disaster recovery so that embargoed datasets
remain inaccessible until the recorded release date.
The digital provenance section also includes all the agents and events that interact
with the dataset and files. There are three agents recorded in the metadata. There is a
Creative Commons agent that is the rights granting agent for the Creative Commons
license. Another rights agent is the Terms of Deposit agent. The Terms of Deposit
agent grants permission for the repository and PURR preservation staff to manipulate
the dataset and files for purposes of long term preservation and access. The software
agent, HUBzero, is also identified and recorded in this section. All agents have unique
IDs so they can be linked to the events each performs on the dataset and files. Table II
includes the events recorded in the digital provenance section[15].
The final process for the AIP metadata generation is the file section. The file section
simply records the file hierarchy (primary, secondary, etc.), the files’ unique IDs, and
 Purdue
Event
name Event description Event preservation explanation University
Capture Initial capture of the publication data from Preserving capture would help with
the user –the first event in the event stream HUBzero debugging, as it tells us when the
SIP capture/creation process started
In- Generated when a SIP must be revised Preserving in-revision would help with 25
revision before it can be approved (it would occur HUBzero debugging, as in-revision is a
between capture and validation) major change in the SIP status. Note that in-
revision only occurs if the SIP is sent back
to the author(s) for revision before it can be
approved for AIP status
Validation Validation of the SIP to ensure it is ready to Validation is one of the major steps in a SI
become an AIP P’s journey to AIP status, so it should be
preserved
Ingestion Creation of the AIP from the approved SIP Ingestion is the creation of the AIP, so it
should be preserved
Fixity Periodic event where the fixity of the files in Preserving fixity check would help with
check the AIP is re-validated HUBzero debugging, as it tells when there
is a problem with the preservation process
Replicaton Copying the AIP bit-for-bit to another As replication creates another copy of the
location for preservation purposes (as in AIP, it should be preserved for debugging
LOCKSS) purposes
Migration Transforming the AIP and its contents into As migration creates a newer,
a more-contemporary format automatically-generated version of the AIP,
it should be preserved for debugging Table II.
purposes PREMIS events

the files’ storage locations. It also records the names of the files included in the AIP.
After the AIP Creation Tool finishes generating the metadata, the well-formed,
validated metadata are included in the AIP along with all the files and other
preservation data and preservation files. The AIP will then be considered completed
and ready for preservation.
In special occasions, an AIP can serve as a SIP. For example, in the case of transfer
from one preservation system to another, a current AIP would serve as the submission
package to this new platform. A Dissemination Information Package (DIP) is created
from the same source material as the AIP; however, it is not a copy or a derivative of an
AIP. A DIP is generated on demand once a member of the designated community visits
the web interface and downloads the dataset. The designated community also has the
ability to download the DIP’s associated metadata files. This is not a typical process
but one that works best for PURR’s publication model, especially as the AIP workflow
is still in development. Generating the DIP from the submission materials allows PURR
to provide immediate access to published datasets.

Conclusion
While development of PURR’s preservation infrastructure is ongoing, the team is
making progress toward the goal of becoming a trusted digital repository. PURR will
utilize a distributed digital preservation model as a strategy for AIP back-ups. In early
2013, Purdue University Libraries became a member of the MetaArchive Cooperative.
OCLC Developed in partnership between six southeastern US university libraries with
30,1 backing from NDIIPP, MetaArchive utilizes LOCKSS software to create a digital
preservation network which approaches digital preservation through replication and
geographic distribution. While still in the early phases of integration, MetaArchive
promises to provide PURR with robust archival backup, in addition to Purdue’s local
and satellite storage infrastructure. Once Purdue and PURR are fully integrated with
26 the cooperative, PURR will be able to satisfy additional ISO 16363 items.
ISO 16363 continues to serve as a rubric, barometer and set of goals for PURR as
development continues. To become a trustworthy repository, the PURR project team
has consistently worked to build a robust, secure, and long-term home for collaborative
research. In order to fulfill its mandate, the project team constructed policies, strategies,
and activities designed to guide a systematic digital preservation environment. PURR
expects to undertake the full ISO 16363 audit process at a future date in expectation of
being certified as a Trustworthy Digital Repository. Through its efforts in digital
preservation, the Purdue University Research Repository expects to better serve
Purdue researchers, their collaborators, and move scholarly research efforts forward
world-wide.

Notes
1. The PURR Digital Preservation Policy was written by Paul Bracke, Associate Dean for
Assessment and Technology, Jake Carlson, Data Services Specialist and Associate Professor of
Library Science, and Sammie Morris, Head, Archives and Special Collections and Associate
Professor of Library Science. The policy may be accessed on the Purdue University Research
Repository web site, available at: [Link]
2. PURR Designated Community definition, available at: [Link]
PURR/whocancreateanewproject
3. Information regarding both TRAC and ISO 16363 can be found online at the Center for
Research Libraries, Metrics for Repository Assessment, available at: [Link]/archiving-
preservation/digital-archives/metrics-assessing-and-certifying
4. As of July 1, 2013, these PURR documents are not available online.
5. DROID. The National Archives of the United Kingdom, available at: [Link].
[Link]/information-management/our-services/[Link]
6. Figure 1 designed by Brandon Beatty (Purdue University Research Repository).
7. In 2010 Purdue University Libraries became a founding member in DataCite, an
international consortium which promotes the sharing of datasets by issuing DOIs, available
at: [Link]
international-cooperative-to-advance-research/
8. METS. Metadata Encoding and Transmission Standard. Library of Congress, available at:
[Link]/standards/mets/
9. DCMI. Dublin Core Metadata Initiative, available at: [Link]
terms/
10. MODS. Metadata Object Description Schema, Library of Congress, available at: [Link].
gov/standards/mods/
11. PREMIS. Library of Congress. Preservation Metadata Maintenance Activity, available at:
[Link]/standards/premis/
12. Creative Commons licenses, available at: [Link]
13. PURR Terms of Deposit, available at: [Link] Purdue
14. PREMIS. Library of Congress. Preservation Metadata Maintenance Activity, available at: University
[Link]/standards/premis/
15. Data Dictionary for Preservation Metadata, available at: [Link]/content/dam/
research/activities/pmwg/[Link]

References
27
Boyko, A., Kunze, J., Littman, J., Madden, L. and Vargas, B. (2009), NDIIPP Content Transfer
Project: The BagIt File Packaging Format, available at: [Link]
display/Curation/BagIt
Cragin, M., Palmer, C., Carlson, J. and Witt, M. (2010), “Data sharing, small science and
institutional repositories”, Philosophical Transactions of the Royal Society A, Vol. 368
No. 1926, p. 4023.
Faniel, I.M. and Zimmerman, A. (2011), “Beyond the data deluge: a research agenda for
large-scale data sharing and reuse”, The International Journal of Digital Curation, Vol. 6
No. 1, p. 59.
Klimeck, G., McLennan, M., Brophy, S.P., Adams, G.B. and Lundstrom, M.S. (2008),
“[Link]: advancing education and research in nanotechnology”, Computing in
Science and Engineering, Vol. 10 No. 5, pp. 17, 19, 22.
Lee, C. and Tibbo, H. (2007), “Digital curation and trusted repositories: steps toward success”,
Journal of Digital Information, Vol. 8 No. 2, available at: [Link]
php/jodi/article/view/229/183
Witt, M. (2012), “Co-designing, co-developing, and co-implementing an institutional data
repository service”, Journal of Library Administration, Vol. 52 No. 2, p. 3.

Corresponding author
Carly C. Dearborn can be contacted at: cdearbor@[Link]

To purchase reprints of this article please e-mail: reprints@[Link]

Or visit our web site for further details: [Link]/reprints
Copyright of OCLC Systems & Services is the property of Emerald Group Publishing
Limited and its content may not be copied or emailed to multiple sites or posted to a listserv
without the copyright holder's express written permission. However, users may print,
download, or email articles for individual use.