Temporal Network Analysis as a Method to Explore Australian World War I Diaries

As the number of transcribed and easily accessible historical documents increases, so does the need for computational methods to assist in the analysis of these documents. Over the past two decades, one such computational method that has been applied is network analysis. This paper proposes that temporal network analysis can be applied to historical diaries to find dates and diaries of interest which should be further explored using a traditional close reading approach. More specifically, this paper focuses on how a network of similar diary entries, as determined by cosine similarity, can be created and explored by looking at key metrics such as network density and node centrality. A case study is performed on a collection of Australian World War I diaries held by the State Library of New South Wales to show that the proposed methodology is viable. Furthermore, this paper discusses future work which could improve the potential of this proposed methodology.