IASSIST 2017 pre-conference workshops: a detailed summary

The IASSIST 2017 pre-conference workshops prepared participants for the main meeting at the University of Kansas in Lawrence, which ran from 23 to 26 May 2017. Rather than previewing individual conference papers, the workshops addressed practical questions about preparing, describing, communicating and reusing research data. They provided the foundation for the event theme, “Data in the Middle: The Common Language of Research.”

For researchers in Australia, these sessions remain relevant because the same questions arise around national data infrastructure, digital agriculture, machine learning and public transparency. The American examples are useful, but their value lies in the underlying practices: shared definitions, reliable documentation and communication between the people who create data and those expected to use it.

Workshops in the context of the full meeting

The workshop programme occupied a distinct position in the conference schedule. Participants first encountered methods and terminology in practical sessions before applying that knowledge during plenaries, specialist presentations and collaborative discussions.

Feature	Pre-conference workshops	Plenary sessions	Specialist programme
Timing	Before the main meeting	Featured sessions across four days	Breakouts, presentations and posters
Format	Guided, practice-oriented learning	Broad framing by invited speakers	Detailed case studies and research exchange
Central question	How can data be made usable?	Why does data quality matter?	What do particular projects show?
Main contribution	Methods, shared vocabulary and practical habits	Institutional context and strategic perspectives	Evidence, examples and new research questions

The distinction mattered because the conference dealt with data at several levels. A plenary could establish why metadata, stewardship and openness affect research culture, while a workshop could show participants how to create a metadata record, test a chart or document a transformation. Specialist sessions then supplied evidence from projects across disciplinary and organisational boundaries.

The archived programme therefore presents the pre-conference material as an onboarding layer for the wider event. Its emphasis was not mastery of a single product. It was the development of durable habits that participants could carry into their own libraries, laboratories, repositories, policy teams and analytical projects.

Core themes across the workshop sessions

Research data management formed the first major strand. Sessions followed data through its lifecycle, from planning and collection to documentation, preservation and sharing. Participants encountered the need to identify appropriate standards, select suitable repositories, create meaningful metadata and connect datasets with the people or organisations responsible for them. This work transformed data management from an administrative afterthought into part of the research design.

A second strand concerned communication and visualisation. The workshops treated a chart, dashboard or map as an argument rather than a decorative summary. Useful visual communication requires attention to audience, scale, uncertainty, accessibility and the decisions a viewer might make. The central problem was how to preserve technical meaning while making the result understandable to researchers, managers, policymakers and members of the public.

Large-scale computation and deep learning supplied another group of practical questions. Participants were encouraged to consider what changes when data no longer fits comfortably in a spreadsheet or desktop application, and how preprocessing decisions affect later results. Discussion of model training, validation, bias and reproducibility showed why computational scale does not remove ordinary data-quality problems.

Applied datasets connected these methods to areas including digital agriculture and global food security. Environmental readings, satellite imagery, weather observations and production records demonstrated how researchers combine evidence collected at different scales. They also exposed difficulties over access, consent, ownership and the participation of communities whose knowledge informs the interpretation of the data.

Skills participants could apply immediately

The most transferable learning was methodological rather than tool-specific. Participants practised cleaning inconsistent records, identifying missing values, checking units and thinking carefully about data structures. Even when demonstrations used familiar analytical environments, the underlying lessons concerned reproducible transformations: what was changed, why it was changed and how another analyst could follow the same sequence.

Documentation received as much attention as analysis. A usable dataset needs more than a download link. It needs a data dictionary, a clear licence, provenance information and guidance on known limitations. These supporting materials allow users to judge whether the data fit their purpose before they invest time in analysis. They also protect original researchers from being treated as permanent, unpaid support staff for every future question.

Evaluation was another practical skill. Participants had to consider how performance measures relate to the intended use of a model or visualisation. A statistically strong result might still communicate uncertainty poorly, while an accessible chart might simplify too aggressively. The workshops encouraged critical comparison between analytical quality, interpretability and responsible presentation.

Bridging producers, users and decision-makers

That principle is sharpened by the conference retrospective on bridging producer-user gaps, which treats communication as part of the data itself. A dataset becomes more valuable when users can understand its meaning, limitations and appropriate applications without reconstructing the producer’s assumptions from scratch.

The producer-user relationship is rarely a single handoff. Producers may include laboratory teams, field officers, software developers, data stewards and automated systems. Users may include analysts within the same institution, researchers elsewhere, government agencies, industry partners or communities affected by the findings. Each group brings different questions, and “available” data does not necessarily mean data that can be interpreted safely or fairly.

Shared vocabulary helps these groups negotiate those questions. Terms such as quality, metadata, provenance and reusable sound straightforward until two disciplines assign them different meanings. The workshops demonstrated that common language grows through concrete examples: examining a real record, questioning an unclear variable, testing a visualisation and identifying where responsibility for an answer lies.

The “middle” in the conference title was therefore neither a technical layer nor a neutral meeting place. It was the shared environment in which evidence became usable knowledge. Digital agriculture and food-security projects illustrated this especially well because they required scientists, information professionals, data managers and domain specialists to interpret complex evidence together.

Adapting the programme for Australian research

Australia’s geography makes the need for accessible training particularly significant. Research teams are distributed across Sydney, Melbourne, Brisbane, Canberra, Adelaide and regional centres, while many collaborators work outside the country. A workshop model based on short demonstrations, local exercises and repeatable materials can reach more participants than one tied to a single campus. For Australians who attended the original meeting, the journey also involved entering the United States through Chicago or Kansas City and completing customs before travelling to Lawrence.

Local governance expectations shape how those exercises should be designed. The Office of the Australian Information Commissioner, the Privacy Act 1988, sector-specific ethics requirements and Indigenous data-governance protocols all affect what may be shared and how consent should be recorded. Australian customs for acknowledging ownership, respecting community authority and providing a visible route for correction must accompany technical anonymisation.

Australian data infrastructure offers ready-made examples. Trove, the National Library of Australia’s catalogue, the Australian Research Data Commons and data.gov.au demonstrate different layers of discovery and access. Using such services in teaching exercises helps participants encounter persistent identifiers, metadata standards and repository workflows that they can adopt rather than treating them as abstract policy requirements.

The local market is also relevant to the practical conclusions. Universities, CSIRO, government agencies, consultancies and cloud providers including AWS, Microsoft Azure and Google Cloud support much of Australia’s research analytics activity. Open-source tools remain attractive where project budgets are limited, while commercial services can reduce setup time. Digital-agriculture projects involving CSIRO, farmer organisations and regional industries provide a strong setting for testing data-sharing agreements, remote-sensing interpretation and community participation.

A practical checklist for Australian data work

The workshop model can be adapted without copying its American institutional setting. The essential requirement is to give participants a real analytical task, make communication part of the exercise and leave them with documented decisions rather than a folder of temporary files.

• Define the decisions that the data must support before choosing software or visual formats.
• Use a realistic dataset whose consent, ownership and access conditions are visible to participants.
• Budget time for cleaning, documentation and metadata as substantive parts of the workshop.
• Pair technical specialists with domain experts so that neither group controls the interpretation alone.
• End with a named owner, a follow-up activity and a publicly accessible artefact where privacy permits.

Privacy and cultural protocols should be designed into the exercise rather than added after the demonstration. Workshop datasets must not expose real personal information, and facilitators need to explain why an Indigenous or community-governed dataset may require different access conditions. These discussions are especially important in Australia, where “open” cannot be treated as a universal synonym for “responsible”.

The durable lesson from the IASSIST 2017 pre-conference workshops was that reliable data work depends on shared practices as much as technical capacity. Preparation, documentation, visual explanation and evaluation become mutually reinforcing when researchers and information professionals treat them as one workflow. For Australian institutions, that remains the most applicable inheritance of the Lawrence programme: data becomes genuinely common language when its meaning survives the journey from producer to user.