Data Governance Frameworks Presented at IASSIST 2017

IASSIST 2017, held at the University of Kansas in Lawrence from 23–26 May 2017, examined how research data can become a shared language across disciplines, institutions and borders. The archived conference programme brings together work on big data, deep learning, digital agriculture and global food security, all of which depend on trustworthy systems for collecting, describing, sharing and preserving information.

The event’s central concern remains relevant to Australian researchers, universities, government agencies and industry partners. Whether the dataset concerns wheat yields near Wagga Wagga, satellite imagery over the Pilbara or public health records in Melbourne, its value depends on governance. Clear accountability, ethical safeguards, common standards and practical agreements determine whether data can be reused responsibly.

From Data Management To Data Governance

Data management usually focuses on the handling of information: storage, file formats, metadata, backups, access controls and preservation. Data governance is broader. It establishes who may make decisions, which rules apply, how risks are assessed and what happens when different organisations need to work together.

The IASSIST 2017 programme treated data as an infrastructure for research rather than a by-product of research. Sessions involving computational methods and large-scale datasets pointed towards governance questions that sit underneath technical capability. A powerful algorithm cannot correct poorly documented observations, uncertain consent or a dataset whose ownership is unclear.

A workable governance framework therefore connects policy with daily research practice. It should define data ownership or custodianship, permitted uses, quality responsibilities, retention periods, security expectations and procedures for correction. It should also recognise that governance changes over time as a project gains new partners, combines datasets or produces findings with commercial or public consequences.

For Australian organisations, this approach fits a research environment divided among universities, CSIRO, state agencies, councils, private firms and community-controlled organisations. Each body may have different procurement rules, ethics processes and information security standards. Governance provides the common operating language needed to align them.

Shared Standards For Interdisciplinary Research

The phrase “common language” is especially important when researchers from different disciplines contribute to one project. An agricultural scientist may describe a field trial through crop varieties and yield measures, while a data scientist focuses on model inputs, and a social researcher records land-use decisions or community outcomes. Without shared definitions, the resulting dataset can look complete while remaining difficult to interpret.

Metadata is one of the principal tools for resolving this problem. Good metadata records the origin of a dataset, collection methods, units of measurement, geographic scope, time period, processing history and known limitations. Controlled vocabularies and persistent identifiers make it easier to connect records across repositories and cite the people and organisations responsible for them.

The IASSIST discussions around big data and digital agriculture also highlight the importance of provenance. Researchers need to know whether a value came directly from a sensor, was entered by a person, was inferred by an algorithm or was changed during cleaning. A provenance record supports reproducibility and helps future users decide whether a dataset is suitable for a new purpose.

For Australian projects, standards need to accommodate local geography and institutional practice. A dataset may include locations in Australian Eastern Standard Time, measurements using metric units and place names recognised by local communities. It may also involve Indigenous data governance expectations, where authority over knowledge and cultural material cannot be reduced to a generic open-access licence.

Access, Ethics And Responsible Reuse

Open data can accelerate discovery, but openness is not an automatic ethical virtue. The appropriate access level depends on the sensitivity of the information, the expectations of participants and the possibility that separate datasets could be combined to identify individuals or communities. Governance frameworks distinguish between data that can be openly published, data available to accredited users and data that must remain restricted.

This question was particularly significant for global food security collaborations. Agricultural, climate and nutrition datasets often cross national boundaries and involve farmers, communities, public agencies and commercial partners. Researchers preparing such work can benefit from clear data sharing agreements that specify permitted uses, attribution, publication rights, security controls and procedures for ending access.

Australian legislation adds several layers to this environment. The Privacy Act 1988 and the Australian Privacy Principles influence how personal information is collected, used, disclosed and secured, although application can vary according to the organisation and activity. State and territory privacy laws may also apply, especially to public-sector records. The My Health Records Act 2012 creates additional requirements for health information, while research ethics approval may impose conditions beyond the statutory minimum.

Consent should therefore be treated as part of the data lifecycle rather than a single form signed at the beginning. Participants need understandable information about future uses, linkage, storage and withdrawal. Governance should record who approved the use, what limitations apply and how disputes will be handled. For research involving Aboriginal and Torres Strait Islander communities, community authority, cultural safety and Indigenous data sovereignty require meaningful participation in decisions about collection, access and interpretation.

Safeguards That Support Trust

• Assign a named custodian for each important dataset and document their authority.
• Separate identifying information from research files wherever practical.
• Record consent conditions, ethics approvals and restrictions in machine-readable metadata.
• Review access permissions when a project changes scope or adds new collaborators.

Agreements Across Institutions And Borders

Collaborative research rarely operates within a single organisation. A university may host the repository, a government department may provide administrative records, a private company may supply sensor data and an overseas research centre may run modelling. Each participant brings its own expectations about confidentiality, intellectual property, security and publication.

A governance framework turns those expectations into operational agreements. The agreement should identify the data being shared, its format, quality status, lawful basis, approved users and authorised purposes. It should address data security, incident reporting, subcontractors, intellectual property, publication review and the process for amending the arrangement.

The arrangement also needs to distinguish data ownership from data stewardship. One organisation may own a database while another maintains it, and a third may have rights to use selected fields. This distinction is important for long-running projects where staff change, grants end or a repository moves to a new platform.

Australian research teams often work across Sydney, Brisbane, Perth and regional centres with different connectivity, staffing and storage arrangements. A governance model that assumes constant high-speed access may fail in remote fieldwork. Practical provisions should cover offline collection, later synchronisation, device loss, local backups and the handling of data gathered on Country or in areas managed by community organisations.

Questions To Settle Before Data Moves

• Who can approve a new use, user or partner?
• Which records require de-identification, aggregation or restricted access?
• What happens after the grant, contract or collaboration ends?
• How will data quality problems, breaches and participant complaints be reported?

Governance In Digital Agriculture And Food Systems

Digital agriculture illustrates why technical data governance matters. Farms and supply chains now generate information through soil probes, satellite imagery, machinery telemetry, weather stations, yield monitors and farm management platforms. These sources can improve water use, crop planning and food traceability, yet they also raise questions about commercial sensitivity and control.

The IASSIST 2017 focus on digital agriculture and global food security can be read through this governance lens. A dataset used to forecast crop stress may help a grower make better decisions, but it could also reveal production patterns to competitors or affect insurance and lending decisions. Governance must specify who receives derived insights, not just who may download the original files.

The Australian market makes this especially concrete. Large agribusinesses may have sophisticated cloud systems, while smaller farms and family businesses often rely on a mixture of spreadsheets, mobile applications and equipment supplied by different vendors. Data portability, contract transparency and the ability to retrieve records when changing providers are practical governance issues, not abstract policy concerns.

Local conditions also affect data quality. Drought, bushfire, flood and unreliable regional connectivity can create gaps or distort measurements. A responsible framework records missingness and uncertainty rather than presenting a polished dataset that hides uneven collection. It should preserve the context around seasonal conditions, land management decisions and sensor maintenance.

Food service and food security projects have a further operational dimension. When research findings inform a multi-course reception, field workshop or community event, planning must connect guest requirements, dietary information, procurement records and safe handling procedures. Guidance on coordinating food service demonstrates how detailed coordination can reduce errors when many inputs and constraints interact.

Making Governance Usable For Research Teams

A framework succeeds when researchers can use it without becoming policy specialists. Roles should be visible at the start of a project: the principal investigator may be accountable for purpose and ethics, a data manager may oversee documentation, an information security team may manage technical controls, and community or participant representatives may guide culturally appropriate use.

Governance should be proportionate to risk. A public dataset of historical rainfall observations does not need the same approval pathway as linked health and location data. A tiered model can classify information by sensitivity, define standard controls for each category and reserve intensive review for high-risk activities.

Training is equally important. Researchers need practical instruction in naming conventions, secure transfer, version control, access requests, incident response and the limits of de-identification. Governance records should be easy to find inside the project workspace, rather than buried in a document repository that team members rarely use.

The archived IASSIST programme remains useful because it connects policy questions with the realities of research work. Its themes encourage institutions to treat repositories, agreements, metadata and analytical tools as connected parts of a research ecosystem. That perspective supports better decisions when datasets move between disciplines, countries and generations of researchers.

A mature governance model can be assessed through a small set of observable practices:

• Every dataset has a description, custodian, access category and retention decision.
• New uses are checked against consent, ethics approval and contractual conditions.
• Data lineage is recorded from collection through cleaning, analysis and publication.
• Partners can locate the rules that apply to their responsibilities.
• Reviews occur when technology, law, project scope or community expectations change.

Governance area	Practical question	Relevance to Australian research
Accountability	Who is responsible for decisions and incidents?	Clarifies duties across universities, agencies, companies and community partners
Metadata and provenance	Can another researcher understand where the data came from?	Supports reproducibility across metropolitan and regional projects
Access control	Who may use the data, for which purpose and for how long?	Helps manage privacy, commercial confidentiality and culturally sensitive material
Legal and ethical compliance	What laws, approvals and consent conditions apply?	May involve the Privacy Act, state legislation, health rules and ethics requirements
Interoperability	Can systems exchange data without losing meaning?	Reduces friction between Australian institutions and international collaborators
Preservation	What must be retained, migrated or securely destroyed?	Protects long-term research value as platforms and grants change

Data governance frameworks presented through the IASSIST 2017 programme are best understood as practical arrangements for making research trustworthy, interpretable and reusable. They join technical standards with ethical judgement, legal duties and relationships between people. For Australian research communities, that combination is essential when data moves from a field site or hospital to a repository, an international collaboration or a public decision-making process.