Food, War and Digital collections

Who is the researcher?

Dr Heather Benbow is the convenor of German Studies at the University of Melbourne. Her recent research has focused on food and culture with a particular emphasis on food and intercultural contact and food in war time. Heather is experienced at using online search interfaces to find digital collections material, and her data includes digitised collections content, but she has not previously worked with APIs, OCR and transcription tools, or computational text analysis.

What is the research about?

The supply and consumption of food is a universal human activity that attains sharpened social, cultural, political and strategic significance during times of conflict. Until recently, the centrality of food has been granted little attention in the vast scholarship on war. This research seeks to combine interdisciplinary thematic research and digital technologies to investigate the roles, meanings and experiences of food in war time in the twentieth century, and to reflect on and help shape collections practice. Currently this includes using Museums Victoria (MV) collections as a case study.

What are the barriers to research?

Analysis of MV’s collections documentation reveals that broad topics like food are rarely the focus of classification structures and collections metadata, and the online Collections site only searches a subset of the metadata associated with each item, making content discovery difficult. With regard to textual documents, few are transcribed or machine readable. Date information is captured as text in the museum’s backend database, which means it cannot then be used to filter search results online; and the options for exporting data once potentially relevant items have been identified are limited.

Some of these issues can potentially be addressed by extracting collection data via the MV API, which is supported by online documentation. However, this documentation is primarily targeted at developers, meaning that thematic researchers without coding or API experience will likely need to seek support before querying collections data via the API or downloading datasets for use in other tools

What skills, tools, and support are required to overcome these barriers?

To create a dataset, researchers need entry-level documentation on API use and access, sample code for common queries, and information on the best way to name, manage, and store large numbers of files. Using this data to discover items in the full collection requires access to search infrastructure, and supported OCR and transcription tools to make more content searchable and machine-readable.

Once potentially relevant items have been extracted to create a thematic dataset (in this case of material potentially related to food during war time), text analysis and visualisation tools are needed to draw out key themes, identify and extract named entities, produce concordances, and understand the positive and negative sentiments associated with food in different contexts. Such analysis will also help identify key material requiring close reading.

What sort of research will this help enable?

To date the majority of research which includes mention of food during war time has focused on rationing, and notions of shortage and supply as they relate to soldiers, strategy, and military campaigns. Broadening the scope helps support history ‘from below,’ looking at the cultural and personal significance of food during times of conflict – an approach based in cross-cutting thematic research, which draws on a broad range of resources from numerous collecting areas. Appling advanced search techniques and textual analysis of datasets enhanced by OCR and transcription tools makes this work practically achievable in large collecting institutions.

What is the broader significance?

This case study highlights three elements which are key to effective contemporary humanities research: sourcing and getting access to large datasets; filtering and augmenting those datasets to produce a thematic research collection of relevant material based on key research questions and themes; and analysing the thematic dataset to better understand its content at an aggregate level, and to support close reading and other analysis at item level. Establishing effective tools and support to achieve these aims will produce outcomes applicable to a wide range of HASS disciplines.