A Metatadata Application Profile for Data Repository Resources

This work was my final project for IST 681 Metadata, a course offered jointly by the Universities of British Columbia (Canada) and Syracus (United States). My project focused on creating appropriate metadata standards for open source data repository resources in order to improve their discovery, access and retrieval. Data repositories offer an ideal way for researches to publish and share data they create with others. This is essential for ensuring that such data can be preserved and that others can easily find them. Data repositories often do this by archiving submitted data, assigning them a digital object identifier (DOI), and providing a web content that carefully describe each data object including what they are about, how to cite them, or the number of times they are cited or downloaded.

Despite the proliferation of open source repositories for facilitating data exchange, discovery and reuse, not enough attempts have been made to create core metadata standards to aid in their description. Creating such standard could be important for many reasons, the primary being that it would greatly enhance the finding, selection and retrieval of datasets by users based on their specific topics, needs or interests. A related reason could also be that it would enhanced transparency and trust in scientific knowledge production as other scholars would be able to easily discover and work on datasets submitted by their peers, and by so doing, validate and reproduce their findings and thus extend the pace of scientific discovery by building on top of their colleagues' efforts. One likely use case for such a metadata standard could be that a library - most probably academic or a similar research institution - could include metadata records of such repositories in their catalogs for members - researchers, students and similar users - to find, retrieve and resuse.

Thus, given the level of interest in research data exchange and resuse, and the emergence of data repositories as important tools for facilitating these activities, implementing core standards for describing the resources of such repositories has become crucial. This project therefore looked at uncovering the most salient metadata elements for data repository resources that can best enhance their discoverability. I begun with a domain analysis of users and use case scenarios that may be found in the next section.

User Domain Analysis

usecase1

Although open source repositories for data exchange attract many users, the main ones may be grouped into three categories: users who contribute data to such repositories, other users who access the published or contributed datasets, and universities, libraries and similar institutions that hosts and or maintains such repositories.

The creators or contributors of datasets decide on the title, date, astracts or descriptions, format and similar attributes of the dataset whereas user communities often encompass researchers from academia and sectors such as private business, health, government, or non-profits. These entities perform several activities during interaction with datasets on repositories such as searching, downloading or harvesting the datasets, visualizing and in some cases, even modifying the datasets to suit their needs.

Host institutions or organizations include academic, research, private or government institutions and libraries that create, host and maintain repositories to house contributed datasets. They are often responsible for granting access both to data contributers to submit their data onto the systems and to users to search for and access these datasets. They also provide unique identifiers to submitted objects and create special webpages for hosting contributed datasets. They make rules on how the datasets is to be accessed and used on the repository, and create facilities for users to cite the data they access on the system

Understanding these various entities and how they interact with open source data repository systems is important for uncovering relevant metadata descriptions or elements that can enhance discoverability of objects.

Element Set Specification

The lack of an appropriate and suitable metadata standard for describing data repository resources create an imperative for an application profile that can enhance discoverability and access of objects. To accomplish this, the project drew elements from existing standards such as Dublin Core (DC) and the Categories for the Description of Works of Arts' new version CDWA Lite. These standards were chosen because they offer important means of describing dataset objects and facilitating their exchange (Zeng & Qin, 2008). In order to identify elements from these standards that are most salient to my project, I evaluated many of the common DC and CDWA Lite elements that are often adopted for data-focused metadata applications. Below is a table providing an overview of selected DC element sets for my application profile, as well as a brief summary of controlled vocabulary used, cardinality and which elements should or should not be mandatory:

Contact
Obligation	Mandatory if Applicable
URI	http://datasets.org/repos/elements/1.1/contact
Repeatable	Yes
Definition	Person, organization or similar entity that may be contacted about the datasets
Purpose	Provides a point of contact information for datasets
Note to confuse with	Publisher,
Examples	North America, Alaska,
Additional Information	May include creator of datasets, a contributing instution or even the publishing authority
crosswalk


Creator
Obligation	Mandatory
URI	http://purl.org/dc/elements/1.1/creator
Repeatable	Yes
Definition	Entity or person responsible for creation of dataset
Purpose	Indentifies authority or entity responsible for accuracy and timeliness of dataset
Not to be confused with	Publisher or contributor
Additional Information	Maybe an organization, individual, institution or name of a position
Examples	Murphy, J.; Health Canada
crosswalk	dc.creator


dataCitation
Obligation	Mandatory
URI	http://datasets.org/repos/elements/1.1/identifier
Repeatable	Yes
Definition	References creator and or authority responsible for datasets creation
Purpose	Allows acknowledgement of datasets creator(s) by users
Not to be confused with	Annotation
Encoding schemes	May include author name, date of publication, title, publisher, verision or URI
crosswalk	cdwalite.citation



dataDescription
Obligation	Mandatory
URI	http://purl.org/dc/elements/1.1/description
Repeatable	Yes
Definition	A brief account of the dataset including its scope, coverage and objectives
Purpose	Provides adequate explanation dataset to help user search decision making
Not to be confused with	Abstract or preface
Additional Information	Maybe an organization, individual, institution or name of a position
crosswalk	dc.description


dateCreated
Obligation	Mandatory
URI	http://purl.org/dc/elements/1.1/date
Repeatable	Yes
Definition	The date or date and time dataset was created
Purpose	Facilitate searching by providing option for filtering search by date
Not to be confused with	Temporal coverage or the time period covered by dataset
Additional Information	Date Exchange Standard - Date, Time, and Date and Time Standard
Examples	2015-05-03; May 3, 2015
crosswalk	dc.date


dateLastModified
Obligation	Mandatory
URI	http://purl.org/dc/elements/1.1/date
Repeatable	Yes
Definition	The date, or date and time the content of the dataset was last changed
Purpose	Facilitate search by focusing on specific dates
Not to be confused with	Temporal coverage or the time period covered by resource
Additional Information	Date Exchange Standard - Date, Time, and Date and Time Standard
crosswalk	dc.date


datePublished
Obligation	Mandatory
URI	http://purl.org/dc/elements/1.1/date
Repeatable	Yes
Definition	Publication date of dataset
Purpose	Provide search filtering option for publication date
Not to be confused with	Date created or date last modified
Additional Information	Date Exchange Standard must be used for recoding publication date attributes
crosswalk	dc.date


Geographic coverage
Obligation	Mandatory if Applicable
URI	http://purl.org/dc/elements/1.1/coverage
Repeatable	Yes
Definition	Information on geographic coverage or spatial extent covered by datasets
Purpose	Provides description on spatial extent or location covered by the dataset
Note to confuse with	Extent i.e information about size of dataset content
Examples	North America, Alaska,
Additional Information	May include single or multiple spatial and temporal coverages, or none
crosswalk	dc.coverage


Identifier
Obligation	Mandatory
URI	http://purl.org/dc/elements/1.1/identifier
Repeatable	Yes
Definition	Unique code assigned to dataset
Purpose	Unique and non-ambiguous identification for datasets
Not to be confused with	Title, or other official names given to title
Encoding schemes	URI; ISSN, ISBN
Examples	Code or ID or unique number for listed dataset items
crosswalk	dc.identifier


Keywords
Obligation	Mandatory if Applicable
URI	http://datasets.org/repos/elements/1.1/keywords
Repeatable	Yes
Definition	Key terms that explain major themes or aspects of the dataset
Purpose	Enables searching for datasets based on major themes or keywords
Additional information	Keywords may span several subject areas and may give an indication of the scope
Examples	phytophthora infestans, no poverty, disease resistance
crosswalk


Language
Obligation	Mandatory if Applicable
URI	http://purl.org/dc/elements/1.1/language
Repeatable	Yes
Definition	Language of dataset content
Purpose	Enables searching of dataset based on specific language
Additional information	Some datasets may have metada in more than 1 language
Examples	English; en-CA,
Encoding schemas	The following may be used RFC 5646, RFC 4647, IANA Language Subtag Registry
crosswalk	dc.language


Notes
Obligation	Mandatory if Applicable
URI	http://datasets.org/repos/elements/1.1/notes
Repeatable	Yes
Definition	Extra pertinent information about dataset
Purpose	Provides additional information about datasets content or usage
Note to confuse with	Abstract; summary
Examples	Subtypes of datasets
Additional Information	May describe some aspects of the datasets or provide recommendations for usage
crosswalk


producerInstitute
Obligation	Mandatory
URI	http://datasets.org/repos/elements/1.1/producerInstitute
Repeatable	Yes
Definition	Entity or organization with financial or administrative responsibility over datasets
Purpose	Identifies entity or organization responsible for production and or maintenance of datasets
Not to be confused with	Publisher, author
Additional information	This entity servmaye as additional reposiory where idatasets may be accessed
Examples	University of British Columbia
crosswalk


Title
Obligation	Mandatory
URI	http://purl.org/dc/elements/1.1/title
Repeatable	Yes
Definition	A name used as access point for dataset
Purpose	Facilitate search and retrieval of dataset; Uniquely identfy dataset
Not to be confused with	File name
Additional Information	Should reflect how users search for the dataset
Examples	Replication datasets for conservation agriculture trials in Kenya in 2011
crosswalk	dc.title

NOTE!!: The above specification table can be accessed as a comma delimited and an Excel spreadsheet on Github

Schema Creation

Having developed appropriate metadata element sets for my project, I created a schema file that would provide a framework for creating .xml documents based on the element specifications I developed. Generally, creating the schema (.xsd) file entailed establishing type properties that would provide the descriptions of what should be contained within the element sets when an .xml document is created based on these element sets. The various types I used included string (xs:string) and dates (xs:date). An important part of creating the schema was deciding on the cardinality of element sets. In other words, the number of times an element could be repeated in the document, and whether they ought to be mandatory or optional. For instance, I reasoned that an element like title may appear only once in a document since dataset objects or resources usually have single titles. However, elements such as keyword> could appear multiple times since there are often more than one keyword for describing datasets. The figure below shows the output of the schema (.xsd) file I created using OxygenXml Editor (Academic trial version).

Example Records.

The generated .xsd schema file was used to create a sample .xml record. As is evident from the the figure below, the document was successfully generated based on the element set properties described in the schema that was created.

The table below shows the output of the .xml record created above

Elements	Values
Title	Replication Dataset for Community-based NTFP Commercialization Survey
Identifier	doi:10.21223/DSRBEV
Creator	Murphy, J. Yankey
dataDescription	A dataset of responses obtained through a community survey to determine impacts of NTFP commercialization activities on rural poverty
dateCreated	2017-02-09
datePublished	2017-04-21
dateLastModified	2019-01-29
Keywords	Poverty alleviation, Forest products commercialization, Conservation, Forest Policy
Language	en-CA
Contact	M.J. Yankey, Faculty of Renewable Forest Resource, University of Texas, US. 1-800-223654
producerInstitute	University of Texas
geographicCoverage	West and Central Africa
dataCitation	Murphy, J.Y. (2017). Replication Dataset for Community-based NTFP Commercialization Survey. https://doi.org/10.21223/DSRBEV.
Notes	Type: Survey; subtype(s): Rural community livelihood survey

Conclusion

The importance of data exchange and resuse for research within and outside academia cannot be overstated. In the absence of this phenomenon, research works may not be easily validated or replicated, which could have important repurcussions on the advancement of knowledge. Data repositories have emerged to make sharing and accessing datasets among researchers both easy and convinient. Efforts to improve description of metadata for enhanced repository objects discovery should therefore be a priority among information and research professionals. Completing this project and the (Metadata) course has taught me an important lesson about best practices for creating and describing information resources for enhanced discoverability and access

Works cited

Zeng, M L. and Qin, Jian (2008) Metadata. Neal-Schuman. New York. 9781555706357 / 9781856046558
Baca, M and Harpring, P. (2016). Categories for the Description of Works of Art - CDWA. The Getty Vocabulary Program. Revised edition. retrieved from: http://www.getty.edu/research/publications/electronic_publications/cdwa/
The Dublin Core Metadata Initiative. retrieved from: https://dublincore.org/
Oxygen XML Editor [Software Application]. Academic Edition. retrieved from: https://www.oxygenxml.com/buy_academic_edition.html