Search results

This study aims to explore the applications of natural language processing (NLP) and data analytics in understanding large-scale digital collections in oral history archives.

NLP and data analytics were used to analyse the oral interview transcripts of 904 survivors of the Japanese American incarceration camps collected from Densho Digital Repository, relying specifically on descriptive analysis, keyword extraction, topic modelling and sentiment analysis (SA).

The researchers found multiple geographic areas of large residential communities of ethnic Japanese people and the place names of the concentration camps. The keywords and topics extracted reflect the deplorable conditions and militaristic nature of the camps and the forced labour of the internees. When remembering history, the main focus for the narrators remains the redress and reparation movement to obtain the restitution of their civil rights. SA further found that the forcible removal and incarceration of Japanese Americans during Second World War negatively impacted and brought deep trauma to the narrators.

This case study demonstrated how NLP and data analytics could be applied to analyse oral history archives and open avenues for discovery. Archival researchers and the general public may benefit from this type of analysis in making connections between temporal, spatial and emotional elements, which will contribute to a holistic understanding of individuals and communities in terms of their collective memory.

This study aims to establish a reliable index to identify interdisciplinary breakthrough innovation effectively. We constructed a new index, the D_Div index, for this purpose.

The D_Div index incorporates the degree of interdisciplinarity in the breakthrough index. To validate the index, a data set combining the publication records and citations of Nobel Prize laureates was divided into experimental and control groups. The validation methods included sensitivity analysis, correlation analysis and effectiveness analysis.

The sensitivity analysis demonstrated the D_Div index’s ability to differentiate interdisciplinary breakthrough papers from various categories of papers. This index not only retains the strengths of the existing index in identifying breakthrough innovation but also captures interdisciplinary characteristics. The correlation analysis revealed a significant correlation (correlation coefficient = 0.555) between the interdisciplinary attributes of scientific research and the occurrence of breakthrough innovation. The effectiveness analysis showed that the D_Div index reached the highest prediction accuracy of 0.8. Furthermore, the D_Div index outperforms the traditional DI index in terms of accuracy when it comes to identifying interdisciplinary breakthrough innovation.

This study proposed a practical and effective index that combines interdisciplinary and disruptive dimensions for detecting interdisciplinary breakthrough innovation. The identification and measurement of interdisciplinary breakthrough innovation play a crucial role in facilitating the integration of multidisciplinary knowledge, thereby accelerating the scientific breakthrough process.