Abstract low-poly geometric background in shades of pale blue and white

Data in the Middle: The common language of research

Building trust in open agricultural data

Open agricultural data promises faster discovery, better models and stronger decisions across food and fibre systems. Weather records, soil observations, crop images, yield estimates and satellite products can be combined in ways no research team could assemble alone. IASSIST 2017, held at the University of Kansas in Lawrence from 23 to 26 May, centred on the common language needed to make research data useful across disciplines.

Yet openness alone does not create confidence. Farmers need to know how information will be used, researchers need reliable provenance and stable terms, and communities need a voice when data describe their land or livelihoods. A trusted approach treats transparency, privacy, commercial fairness and Indigenous authority as connected design choices, rather than obstacles placed after publication.

Dimension Broad public release Controlled research access
Immediate use Direct download or API access Application, approval or secure environment
Suitable content Aggregated records and public-interest datasets Identifiable grower records, confidential commercial data and sensitive cultural information
Governance Open licence, published metadata and clear provenance Data-use agreements, ethics review, de-identification and community governance
Principal benefit Broad reuse, reproducibility and public accountability Protection of rights, managed collaboration and reciprocal benefit
Common weakness Context may be lost, enabling misuse or false certainty Delays, access costs and exclusion can reduce participation
Measure of success Reuse with attribution and documented corrections Meaningful consent, safe use, fair credit and demonstrated benefit

Why trust determines whether open data is useful

Trust is an operating condition for responsible research. It gives growers confidence that shared information will not be misrepresented, gives scientists confidence that records mean what their labels suggest, and gives communities confidence that participation produces fair outcomes. A file can be technically open while remaining practically untrusted if its licence is ambiguous, its quality is unknown, or its social consequences were ignored. Technical transparency is necessary, but it does not by itself create reciprocity or a sense of procedural fairness.

Australian use cases make this concrete. In Melbourne’s agrifood clusters and around Sydney, Brisbane and Adelaide, researchers routinely join climate, produce and supply-chain information. Growers often check Bureau of Meteorology forecasts before spraying or harvest, while farm advisers move between spreadsheets, machinery terminals and cloud dashboards. At Coles and Woolworths, shoppers may see provenance claims on an app, while farmer markets often rely on direct grower stories; much of the operational data behind retail claims remains within commercial platforms.

What the IASSIST 2017 archive contributes

The archived programme, schedule, plenary details and presentations from “Data in the Middle: The Common Language of Research” capture an early snapshot of conversations about big data, deep learning, digital agriculture and global food security. Together, these subjects show why interoperability matters: a rain record, a satellite image and a market forecast become useful only when their units, time scales and assumptions can be compared. Its plenary strand also put research communication alongside technical themes, reflecting the practical work of discovering, appraising and reusing information.

An archive demonstrates the need for stewardship over time. Technology, file formats and institutional links change, so a dataset released with little documentation can lose its meaning even when the original files survive. Persistent identifiers, version histories, preserved metadata and clear maintenance responsibility allow later researchers to judge whether an old result still rests on sound evidence. That lesson is especially relevant to project-funded datasets, whose maintenance often ends when a grant or staff contract ends.

Designing useful shared data

Usable sharing depends on disciplined metadata. Each release should state its variables, units, geographic and temporal coverage, collection method, processing steps, uncertainty, known gaps and quality controls. A machine-readable data dictionary can accompany a human summary, while provenance records connect every published version to its source. These practices apply the FAIR principles—findable, accessible, interoperable and reusable—and turn an attractive download into evidence another team can inspect or reproduce. Shared vocabularies and APIs matter when a rainfall service must align with crop-reporting fields or satellite-derived boundaries.

Access level should follow the sensitivity and public value of each dataset. Public release suits aggregated observations with few direct implications for an individual farm. Registered access may fit data requiring a commitment to responsible use, while mediated or secure access can support research involving identifiable records, confidential machinery data or culturally restricted knowledge. Clear licences should sit alongside access rules, and every restriction needs a stated reason, decision-maker and review date. Decisions should be recorded in a public access register so restricted material can later move to a wider tier when conditions change.

Privacy, contracts and community authority in Australia

Many organisations handling personal information are subject to Australia’s Privacy Act 1988, administered by the Office of the Australian Information Commissioner. Records that can identify growers, workers or landholders can fall within that regime. Since December 2024, organisations must notify eligible data breaches to the regulator and affected individuals. Open-data projects should still use privacy impact assessments, minimisation, aggregation and secure handling of re-identifiable combinations; publishing a record merely because it came from a public dataset does not remove those duties. Land parcels, irrigation schedules and worker histories can become identifying when linked to names, even if no single field appears sensitive.

Commercial boundaries create another. Data generated by tractors, grain handlers, processors, insurers, supermarkets or agricultural technology firms may be governed by contracts, confidentiality terms or platform access policies. A research team can possess a legally usable file yet lack permission to redistribute it, train a commercial model on it or publish supporting details. Reaching an agreement early clarifies ownership, permitted uses, attribution, liability and whether derived outputs remain open. Researchers should not rely on a general statement that data are “open” when platform terms or processor agreements say otherwise.

Community authority can require a different process again. Agricultural and environmental datasets may include Indigenous knowledge about country, seasonal indicators, plants or water. The CARE principles for Indigenous Data Governance stress collective benefit, authority to control, responsibility and ethics. Researchers should work with the relevant Traditional Owner and Traditional Islander groups rather than treat cultural information as another open resource, and should negotiate storage, access, credit and benefit-sharing according to community protocols. Community protocols may permit careful research use without unrestricted release, so transparency and access are not identical goals.

Digital agriculture and responsible machine learning

Deep learning turns uneven data into influential decisions. A crop-disease model trained on images from one trial can perform poorly in another district, while a yield model may reproduce historical gaps in soil sampling or regional representation. Australian wheat, sheep, dairy and horticulture systems add variation in climate, cultivars, farming methods and market conditions. Dataset documentation, subgroup evaluation and external testing are therefore part of data quality, not optional additions presented after deployment. Monitoring data can also change over time, making the date of training as important as the size of the dataset.

Publishing a model does not automatically settle the privacy or commercial questions around its training material. Teams can release model cards, code, evaluation results and the limitations of sensitive training data while keeping protected records in an approved environment. Reproducible workflows, fixed evaluation sets and independently checkable claims give downstream users more confidence than an accuracy score alone. Human review remains important where a model influences spray timing, water allocation, grading or other decisions with direct economic effects. Agencies and companies can also publish aggregate performance or query interfaces when source records or model weights are sensitive.

Connecting Australian evidence to global food security

Global food security depends on evidence that can cross borders without losing its context. Australian research contributes valuable methods for drought monitoring, dryland production, water management and supply-chain forecasting, but national datasets cannot stand in for every farming system. Partners in regions with limited connectivity or different data capacity may need low-bandwidth formats, translated documentation and locally chosen standards. Technical sharing works best when the research question and governance rules are negotiated together. These choices influence early-warning systems, crop insurance design, water planning and public procurement as much as academic publication.

Reciprocity should be visible in the resulting data system. Agricultural partners can help define variables, maintain local records, train analysts and interpret findings, while Australian teams contribute infrastructure, funding and publication access. Credits, governance roles and benefits need to be recorded in metadata and agreements, not left to goodwill. The IASSIST 2017 theme is relevant here: a common language can connect researchers without making one country’s data model, research priorities or definition of success the standard for every partner. Training and documentation should therefore be co-designed with the people who will interpret the outputs after the project ends.

Actions for trustworthy agricultural data programmes

A practical programme can turn these principles into operating rules. The following measures give teams a defensible way to decide what to publish, how to protect people and how to show that reuse is responsible.

  • Set the purpose and audience before release. State the decisions the data can support, intended users, foreseeable harms, exclusions and a retention period, then revisit the scope when a new use is proposed.
  • Complete a sensitivity and access review. Identify personal, commercially confidential, environmentally sensitive and culturally restricted elements, assign an access tier, and publish the reasons and criteria for each restriction.
  • Record provenance as a continuous practice. Capture collection dates, instruments, transformations, version changes, error checks and corrections so users can trace a result from source to output.
  • Build meaningful consent into collection. Explain intended uses in accessible language, separate unrelated secondary uses where feasible, permit withdrawal when appropriate, and give community governance a defined decision-making role.
  • Publish reusable standards. Provide stable identifiers, machine-readable metadata, documented software and workflows, and an open licence such as Creative Commons where conditions allow.
  • Measure trust and participation. Track corrections, reuse, attribution, partner feedback, access requests and safety incidents, then publish periodic transparency reports.

These measures treat trust as something demonstrated through behaviour. They also preserve the productive core of the IASSIST theme: researchers, farmers, data stewards and information professionals translating evidence for one another. The strongest open agricultural data systems make information easier to find and verify while respecting the people and organisations whose knowledge makes it valuable. Openness earns legitimacy when the conditions around reuse are visible and the benefits are shared fairly.

At the Conference

What attendees experienced in Lawrence

p1 reed

Plenary Sessions

Keynotes from Daniel Reed on data, technology, and culture, and Jennifer Clarke on digital agriculture and the Midwest Big Data Hub.

platinum icpsr

Workshops

Full-day technical workshops on Tuesday, May 23. Morning sessions ran 9:00–12:00 and afternoon sessions 13:00–16:00. Laptops were required.

gold rockhurst

Social Events

An opening reception at The Oread, a banquet, and an optional post-conference tour of Kansas City including Crescent Moon Winery.

Venue & Accommodations

Where the conference took place

silver ifdo

Kansas Union

University of Kansas campus, Lawrence. Main conference venue with check-in on the 4th and 5th floor lobbies.

platinum ddi

The Oread

1200 Oread Avenue, Lawrence. Hosted the opening reception and offered a room block for attendees.

silver ciser

The Eldridge

701 Massachusetts Street, Lawrence. A partner hotel with a reserved room block for conference guests.

platinum icpsr

Springhill & TownePlace Suites

Marriott properties in Lawrence with room blocks reserved under the "KU IASSIST Conference" name.

Program Highlights

Sessions and activities

Plenary Sessions Workshops Poster Presentations Committee Meetings Opening Reception Banquet Tour Kansas City Pecha Kucha Check-In Local Favorites

Getting Here

Lawrence, Kansas

Kansas Union · University of Kansas
1301 Jayhawk Blvd, Lawrence, KS 66045

Kansas City International Airport (MCI) is approximately 50 minutes by car. Lawrence Transit Routes 10 and 11 served the area ($1 exact change).

Plan Your Stay

Accommodation options that were available

silver ifdo

The Eldridge

701 Massachusetts Street, Lawrence, KS 66044. Room block now closed.

platinum ddi

The Oread

1200 Oread Avenue, Lawrence, KS 66044. Room block now closed.

silver ciser

Springhill Suites

Marriott property. Room block reserved under "KU IASSIST Conference."

gold rockhurst

TownePlace Suites

Marriott property. Room block reserved under "KU IASSIST Conference."