Why Interdisciplinary Data Teams Matter: Lessons from IASSIST 2017

Research questions rarely fit neatly inside one department. A project about food security may involve agricultural science, economics, public policy, computing, environmental studies and community knowledge. Each field brings its own vocabulary, methods and assumptions, yet useful decisions depend on connecting those perspectives.

IASSIST 2017, held at the University of Kansas in Lawrence from 23–26 May, explored that connective work through the theme “Data in the Middle: The Common Language of Research”. Its archived programme brought together discussions of big data, deep learning, digital agriculture and global food security, alongside practical guidance for attendees. Looking back from Australia, the event offers a timely model for building research teams that can move from raw information to shared understanding and responsible action.

Data Becomes Useful When Disciplines Meet

Data is often described as a universal language, but datasets do not interpret themselves. A machine-learning specialist may see a classification problem, while a social scientist sees a question about power, representation or access. An agricultural researcher may focus on soil moisture and yield, whereas a community organisation may ask whether the proposed intervention suits local knowledge and livelihoods.

Interdisciplinary teams matter because they expose these differences early. A statistician can identify sampling bias; a subject expert can explain why an apparently unusual result is perfectly plausible; and a data librarian can make the sources discoverable, documented and reusable. The quality of the final insight depends on the conversation between these roles, not merely on the sophistication of the software.

The IASSIST 2017 theme placed data “in the middle”, a useful image for collaborative research. Data sits between disciplines as a meeting point, rather than belonging exclusively to a technical department. That position encourages teams to ask what was measured, who defined the categories, what is missing and how the results will be used.

Shared Vocabulary Prevents Expensive Misunderstandings

Collaboration can fail when familiar terms carry different meanings. “Big data” may refer to volume for one participant, speed for another and complex social consequences for a third. “Open” can mean freely downloadable, technically interoperable or legally reusable. A team that does not clarify these terms may appear aligned while working towards incompatible outcomes.

The practical answer is to create a shared vocabulary at the beginning of a project. A short glossary can define key measures, acronyms, geographic boundaries, privacy terms and quality standards. Teams can also agree on a data dictionary that records each field’s meaning, unit, source, collection date and known limitations.

This is especially relevant in Australia, where a national project may combine Australian Bureau of Statistics information, state datasets and material collected by universities or councils. A postcode, statistical area and Indigenous community boundary are not interchangeable. Someone working in Brisbane may interpret “local” differently from a researcher in regional Queensland, just as a coastal dataset from Sydney may not transfer neatly to inland New South Wales.

Shared language also supports better public communication. Researchers presenting findings at a university seminar, community meeting or industry forum should be able to explain technical choices without forcing non-specialists to learn an entire discipline’s code. Plain English is not a reduction in rigour; it is a test of whether the reasoning is understood.

Different Roles Create Stronger Questions

An interdisciplinary team should be designed around the research problem, not assembled as a collection of impressive job titles. For a digital agriculture project, this could include a farmer or grower, agronomist, remote-sensing specialist, software engineer, economist, Indigenous knowledge holder and data steward. Each person should have a meaningful role in defining the question, interpreting evidence and reviewing outcomes.

IASSIST 2017’s coverage of digital agriculture and global food security illustrates why that breadth is valuable. Sensors may record rainfall, crop condition or soil chemistry, but they cannot independently explain labour constraints, market access, cultural priorities or the cost of adopting a new system. Technical accuracy is necessary, yet it cannot substitute for knowledge of how agricultural decisions are made.

Australian conditions make this particularly clear. A vineyard near Adelaide, a wheat farm in Western Australia and a horticultural business outside Darwin face different climates, transport systems, labour patterns and commercial risks. A predictive model developed for one region may perform poorly elsewhere if it ignores seasonal variation, farm scale or the realities of water allocation.

Teams also benefit from including people who understand implementation. A policy analyst can identify regulatory barriers, while a communications specialist can translate results into material that a council, farmer cooperative or health service can use. An information professional can preserve the project’s documentation so future researchers can assess how conclusions were reached.

A Collaborative Workflow Turns Evidence Into Action

Interdisciplinary work needs more than goodwill. Teams require a workflow that gives each discipline a clear opportunity to contribute. A useful sequence begins with scoping the problem, mapping available evidence, agreeing on definitions, testing a small prototype, reviewing results collectively and documenting decisions before scaling up.

Early workshops should focus on questions rather than solutions. Participants can map who produces the data, who controls access, who may be affected and what decisions the analysis is expected to support. This prevents a common failure mode: building an impressive dashboard before anyone has established what action it should enable.

The workflow should also include practical constraints. A venue or field project may require information about transport, accessibility, timing, equipment and safety. Even a seemingly minor design choice can cross professional boundaries; facilities staff, event coordinators and technical teams may need to discuss ceiling-height planning when installing displays, sensors or hanging materials. The broader lesson is that specialist knowledge becomes useful when it is translated into decisions other people can act on.

A shared project space can hold the data dictionary, meeting notes, assumptions, code versions, decisions and unresolved questions. Regular review meetings should distinguish between a change in the evidence and a change in the team’s interpretation. That distinction helps prevent disagreements from becoming personal and makes the project easier to audit.

Team function	Contribution to the project	Risk when absent	Useful Australian context
Subject specialist	Defines the real-world problem and tests whether findings make sense	Technically correct but practically irrelevant results	Local farming, health, climate or community knowledge
Data steward or librarian	Documents sources, permissions, formats and preservation needs	Unclear provenance and poor reuse	State, council, university and ABS datasets
Quantitative analyst	Tests relationships, uncertainty and model performance	Overconfident claims or hidden statistical errors	Regional variation across cities and remote areas
Community or industry partner	Explains lived experience and implementation barriers	Low trust and weak adoption	Farmers, councils, Aboriginal organisations and service providers
Software or infrastructure specialist	Builds reliable pipelines, tools and access controls	Fragile systems and inconsistent outputs	Connectivity differences between metropolitan and rural locations
Policy or communications specialist	Connects evidence with decisions and public explanation	Findings remain inside the research team	Government programmes, local markets and public services

Technical Expertise Needs Ethical Direction

Big data and deep learning can reveal patterns at a scale that would be impossible through manual analysis. They can help classify satellite images, forecast demand, identify changes in land use or detect relationships across large collections. Yet a model’s ability to find a pattern does not establish that the pattern is fair, causal or suitable for intervention.

An interdisciplinary team can test those assumptions from several angles. Domain experts can ask whether the variables represent the phenomenon accurately. Social researchers can examine who is made visible or invisible. Legal and governance specialists can review consent, licensing and privacy. Community partners can identify harms that may be invisible in a spreadsheet.

Australian research also needs to take Indigenous data governance seriously. Data about Aboriginal and Torres Strait Islander peoples should not be treated as an ordinary extractive resource. Questions of authority, cultural safety, ownership, access and benefit-sharing belong in the project design, rather than being added after a model has been trained.

Ethical review should continue throughout the work. A project may begin with a benign purpose and acquire new risks when datasets are linked, predictions are used to allocate services or results are published at a fine geographic scale. Teams need explicit rules for access, retention, re-identification, accountability and withdrawal.

Local Context Determines Whether Findings Travel

A model that performs well in a controlled trial may struggle in the field. Data collected during a favourable season may not represent drought, flood or fire conditions. A classification system trained on metropolitan imagery may misread remote landscapes. A health dataset gathered through a major hospital may exclude people who face transport, cost or language barriers.

Australia’s geography magnifies these problems. Researchers working across Melbourne, Perth and remote communities must consider time zones, connectivity, travel costs and different institutional relationships. A team meeting scheduled for the morning in Sydney may fall outside reasonable working hours for collaborators elsewhere. Fieldwork may depend on long road journeys, seasonal access or local approval processes rather than a quick trip between city offices.

The local market matters as well. A digital agriculture tool should account for the scale and margins of Australian producers, the cost of imported equipment, supply-chain distance and the practical value of a recommendation. Advice that assumes dense European transport networks or large American farms may not suit a family business selling through a regional cooperative or a grower supplying Brisbane and Sydney markets.

Local customs and community expectations shape participation. Sharing food at morning tea, allowing time for informal discussion and acknowledging Country can help establish respectful relationships, though these gestures cannot replace genuine decision-making power. Teams should learn how a partner prefers to communicate, who has authority to speak and how findings should be returned to the community.

Conferences Extend Collaboration Beyond One Project

The value of a conference archive lies in the connections it preserves. IASSIST 2017 brought together programme material, plenary information, presentations and attendee guidance in one place. That combination shows that research exchange depends on both intellectual content and the conditions that allow people to participate.

A plenary session can introduce a common challenge, but smaller presentations and informal conversations often reveal how methods operate in practice. Researchers may discover a comparable dataset, a different approach to metadata or a partner who has already solved a technical problem. These exchanges help transform isolated projects into a broader research community.

The practical details matter more than they first appear. Check-in instructions, transportation information and local recommendations reduce friction for attendees, allowing them to concentrate on the programme. For an Australian event, that might mean clear advice about public transport in Melbourne, accessibility at a regional venue, the difference between state and national public holidays, or whether a site is suitable for participants travelling from remote areas.

Digital archives extend that benefit beyond the original dates. A student in Canberra, a data manager in Hobart or a researcher at a university in Sydney can examine the programme years later and identify methods worth adapting. Preserving presentations and context ensures that interdisciplinary learning is not lost when a conference closes.

The Lasting Lesson Is Institutional, Not Merely Technical

Interdisciplinary teams are sometimes treated as temporary arrangements formed around a grant. The stronger lesson from IASSIST 2017 is that collaboration needs durable institutional support. Universities, government agencies and research organisations should recognise data stewardship, partnership building, documentation and public explanation as core scholarly work.

That support includes training. Researchers need opportunities to learn enough about neighbouring disciplines to ask informed questions, while technical specialists need practice explaining systems to people outside computing. Career structures should reward team-based contributions rather than measuring value only through individual authorship or disciplinary publication counts.

Funding models should also allow time for relationship-building and data preparation. Cleaning records, negotiating access, documenting provenance and returning results to communities are essential activities, even when they produce no dramatic headline. A fast project that cannot be reproduced or trusted may cost more than a slower one built on sound foundations.

For Australia, the approach is relevant to climate adaptation, public health, agriculture, biodiversity, urban planning and regional development. The country’s institutions hold valuable data, but value emerges when people can connect it responsibly across boundaries. The central lesson is simple: data may sit in the middle, but people must do the work of building the bridge.