What The IASSIST 2017 Opening Session Still Teaches Us

The IASSIST 2017 opening session placed research data at the centre of scholarly communication. Held at the University of Kansas in Lawrence from 23–26 May 2017, the conference examined how data could become a shared language across disciplines, institutions and national borders. Its theme, “Data in the Middle: The Common Language of Research,” was a useful challenge to the idea that data is simply a technical product stored after a project has finished.

For Australian researchers, librarians, archivists and data specialists, the plenary discussion remains relevant. Universities, government agencies and research organisations now work with larger datasets, stricter governance requirements and growing expectations around public value. The questions raised at IASSIST 2017 connect closely with Australian realities, from Australian Bureau of Statistics collections and environmental monitoring to Indigenous data sovereignty and the management of research outputs through national infrastructure.

Data Was Presented As A Shared Research Language

The opening session’s central message was that data links communities that may use different theories, methods and vocabularies. A social scientist, an agricultural researcher, a librarian and a machine-learning specialist may approach evidence differently, yet each depends on reliable description, careful stewardship and a clear account of how information was produced.

This idea shifted attention away from data as a specialist concern. Metadata, file formats, identifiers and preservation policies can sound like back-office matters, but they determine whether a dataset can be found, interpreted and reused. When those foundations are weak, even an impressive collection can remain practically invisible to researchers outside its original project.

The phrase “in the middle” also suggested a role for data professionals as connectors. Their work sits between researchers and institutions, between local knowledge and computational analysis, and between immediate project needs and long-term public access. That role is familiar in Australia, where university libraries and research offices often help academics meet funder expectations while making outputs discoverable through institutional repositories and national services.

A common language does not mean that every discipline must adopt identical methods. It means that researchers need enough shared structure to understand what a dataset represents, what its limitations are and how it can be responsibly used. That is a more practical goal than forcing every field into a single technical model.

Big Data Needs Context Before It Creates Value

The conference programme connected the opening conversation with subjects including big data and deep learning. These fields can process information at a scale that was difficult to imagine in earlier research environments, but processing power does not automatically create knowledge. Algorithms depend on the quality, scope and provenance of the material supplied to them.

Context is particularly important when datasets are assembled from different sources. A variable may have changed definition over time, a survey may exclude a relevant population, or records may have been collected for administrative purposes rather than research. If those details are lost, a model can produce precise results that answer the wrong question.

This lesson applies to Australian policy and research settings. A dataset built from urban Sydney or Melbourne may not represent remote communities, regional towns or island populations. Climate and agricultural records from the Murray–Darling Basin cannot simply be transferred to the tropics of Far North Queensland. The local conditions, collection practices and social circumstances surrounding the data need to remain visible.

Deep learning also raises questions about reproducibility and explanation. A researcher may be able to report a model’s performance without being able to describe every factor that influenced its output. Preserving training data, code versions, parameter settings and evaluation methods gives later users a better chance of assessing the result. Without that record, the research may be difficult to verify or responsibly extend.

The opening plenary therefore remains a useful corrective to technology-first thinking. Large-scale analysis is valuable when it is connected to a well-defined research problem, documented evidence and people who understand the setting from which the data came.

Digital Agriculture Shows Why Place Matters

The conference’s attention to digital agriculture and global food security gave the theme a tangible dimension. Agricultural research brings together satellite imagery, soil measurements, weather observations, farm records, crop models and economic information. Each source describes only part of a complex system, and meaningful results depend on bringing those parts together without erasing their differences.

For Australian audiences, this has immediate relevance. Farmers in Western Australia, South Australia, Victoria and Queensland work in environments shaped by different rainfall patterns, soil types, market conditions and water constraints. A tool developed for broadacre farming near Wagga Wagga may need substantial adjustment before it is useful in the Kimberley or Tasmania.

Data can support better decisions about irrigation, planting windows, disease detection and supply chains. It can also reveal trends that are difficult to identify through individual observations. Yet the value of a digital agriculture project depends on whether producers can understand the output, trust its source and afford the technology required to use it.

The global food-security perspective adds another layer. Research data should not be treated as an unlimited resource extracted from communities and landscapes. Farmers, Indigenous groups, local organisations and public agencies may have legitimate interests in how information is collected, combined and commercialised. Clear agreements about access, attribution and benefit-sharing are part of responsible data practice.

This is where the IASSIST 2017 theme becomes more than a slogan. Data sits between scientific expertise, local experience, public policy and commercial activity. A successful project must create connections across those domains while recognising that not every dataset should be open to everyone.

Stewardship Includes Ethics, Access And Accountability

A major takeaway from the opening session is that data stewardship is an ongoing responsibility. It begins when a project is designed and continues through collection, analysis, publication, preservation and reuse. Decisions made early about file formats, consent and documentation can determine whether data remains useful years later.

Open access is an important part of the research landscape, but openness has limits. Personal information, culturally sensitive material, commercially valuable records and data relating to vulnerable communities may require restricted access. The responsible question is not simply whether a file can be placed online. It is whether access is lawful, ethical, technically secure and consistent with the expectations of the people represented.

Australian institutions increasingly need to consider Indigenous data governance in this context. Aboriginal and Torres Strait Islander communities have distinct rights and interests in data about people, Country, culture and knowledge. Conventional open-data assumptions may be inappropriate where community authority, cultural protocols or collective benefit need to guide access and reuse.

Stewardship also requires accountability for decisions that are easy to overlook. Who selected the variables? Who was excluded from the sample? Who has permission to reuse the data? Which version is authoritative? Can a participant withdraw, or can a community challenge an interpretation? These questions belong in project documentation, not only in an ethics application or final publication.

For Australian universities, the practical work may involve data management plans, secure research environments, persistent identifiers and repository records. Services associated with the Australian Research Data Commons, institutional repositories and national collections can help, but tools cannot replace judgement. Good infrastructure supports responsible practice; it does not decide what responsibility requires.

Turning The Plenary Message Into Research Practice

The lasting value of the IASSIST 2017 opening session lies in its practical implications. Researchers can treat data as a shared research object rather than a private by-product of a single project. That means documenting collections for people outside the original team, using stable formats where possible and recording decisions that affect interpretation.

Librarians and data specialists can make this work more visible by building relationships before a project begins. A conversation during proposal development may prevent major problems with consent, storage or file organisation later. Training should cover the research context as well as software, because a technically correct workflow may still produce misleading or unusable results.

The comparison below summarises how the plenary’s broad principles translate into everyday decisions.

Plenary principle	Research practice	Australian relevance
Data connects disciplines	Use shared metadata, definitions and identifiers	Helps teams across universities, government and industry work with the same evidence
Scale does not guarantee insight	Record provenance, bias, uncertainty and collection limits	Prevents national datasets from being treated as equally representative of every region
Technology needs human context	Combine computational analysis with disciplinary and local knowledge	Supports better decisions in agriculture, health, climate and community research
Access carries responsibility	Apply proportionate controls for sensitive or restricted material	Aligns open research goals with privacy, Indigenous data governance and ethics
Data has a long life	Preserve files, documentation, code and version history	Makes research more reproducible and useful after grants and projects end

Research teams can use the following practices to carry the opening session’s message into current projects:

• Define the intended users, boundaries and potential harms of a dataset before collection begins.
• Record provenance, changes, exclusions and assumptions in language that non-specialists can follow.
• Consult affected communities and knowledge holders about access, interpretation and benefit.
• Use persistent identifiers, version control and reliable repositories for data, code and documentation.
• Match openness to the sensitivity of the material rather than treating public access as automatic.
• Review whether the dataset remains fit for purpose when it is reused in a new location or discipline.

The examples discussed around big data, deep learning, digital agriculture and food security show why these practices matter. A dataset may travel from a laboratory to a policy agency, from a university repository to a commercial tool, or from a local field study into an international comparison. At every stage, its meaning depends on the records and relationships that travel with it.

The opening plenary also offered a broader view of research impact. Impact is not limited to a journal article or a successful model. It can include a dataset that another team can interpret, a community that retains control over its knowledge, or a public agency that can make a better decision because the evidence is transparent. In that sense, data becomes a durable part of the research commons.

For Australian readers looking back at the archived IASSIST 2017 programme, the session remains timely because the underlying issues have grown rather than disappeared. The tools have changed, and the volume of information has increased, but the need for shared meaning, careful stewardship and accountable collaboration is much the same. Data is most powerful when it remains connected to the people, places and purposes that give it significance.