The digitisation drive of recent decades has blessed humanities researchers with a significant body of digital data. Data-driven methodologies are increasingly being applied to such digital humanities data, taking advantage of both the bounty and the fast-paced movement towards data literacy. However, this data is far from inert and objective – it has frequently been collected and/or digitised by human eyes and hands, in activities that are intermediated and guided by (sometimes invisible) human decision-making.

In understanding and explaining results of any data-driven analyses – e.g., statistical, machine-learning, deep-learning, computer vision – an in-depth examination and discussion of data history/ies, biases and gaps, quantified and/or potential errors, and further data quality investigations, along with a comprehension of the concomitant implications on analyses, are required. The need to scrutinize these data characteristics, and how collection methodologies (sometimes accumulated and adapted over many years) shaped them, is critical for choosing appropriate analyses, and for effective interpretation and communication of analytic outcomes. But all too often it appears these important elements of initial data work are under-represented or missing altogether in published literature.

In this paper I share a model for thinking about the impact of data issues on analyses, beginning with the case study of data collected by humans when observing and recording physical objects (eg. a group of related archaeological material) and then extending it. Part critique of the structuring of computational analyses, and part call to action to invest in exploring data and data histories, the goal is uplifting data working methodologies.

The State Library of Western Australia is digitising its at-risk, significant and unique audio-visual collections to ensure long-term preservation and enable online public access. Spanning more than 100 years, the AV collections portray people, places and events unique to Western Australia, including amateur recordings, government and private productions, films, music and interviews. In 2024, the State Library funded two researcher-in-residence opportunities for established researchers to view, listen and reflect on material digitised during 2022-23 to enhance the State Library’s understanding of the collections’ significance, and the potential for future curation and research. Guided by the Library’s Collection Interpretation Strategy, the researchers were encouraged to explore the collection in ways that aligned to their expertise and areas of interest.

The thematic entry point developed by one of the researchers-in-residence, Dr Dean Chan serves as a case study that proffers and reflects on opportunities for making data connections via digital curation and situated storytelling. “Emplacements” encapsulates the 1960s-80s as a formative period in Western Australia’s modern history when placemaking was a key preoccupation in creative practices and the emergent cultural industries, ranging from filmmaking to puppetry. “Emplacements” maps these intersectionalities, reengages the stories via sound and moving image, and makes new connections with other State Library collections, both physical and digitalised, thereby not only preserving but also reanimating the historical record.

The information contained in archives is never neutral and often highly selective. As a result, engaging with state record collections can be a daunting exercise. While it remains important to read a series along its epistemological grain, to gain an understanding of underlying structure and omissions it is often necessary to analyse a run of records in conjunction with others. At the best of times this a daunting challenge. This paper describes an ongoing collaboration between academic researchers, archivists and family historians to digitally excavate the Tasmanian Archives. As a result of this collective effort it has been possible to join-together the digital products of 18 Australian Research Council funded projects dating back to the early 1990s. These have been supplemented with archival indexes, digital images and data transcribed by citizen researchers. The resultant historical research data commons covers more than sixty record series totalling more than 5 million records, all of which have been cleaned, coded to the same standard and (to a greater or lesser extent) linked. In this paper I will outline the principal challenges encountered to date, the advantages of holding digital historical series in common and the future challenges of maintaining public access to the digital traces of the past.