Conclusion and Future Work
OpenRefine and Tableau Public proved to be the most effective digital humanities tools for us when it came to cleaning, normalizing, and standardizing data and creating interactive visualizations. Instead of relying heavily on the more widely used Excel and its analysis and charting tools, we could have taken advantage of them much sooner for increased efficiency. Nonetheless, we learned about the benefits and drawbacks of these tools during the process.
Timemapper appears to be the most problematic digital humanities tool we chose to experiment with for the analysis project. While creating interactive timemaps on the platform is simple, the embedded base maps that users cannot replace are contemporary and thus cannot accurately reflect historical geographic locations and distribution. Furthermore, the platform had been (and continues to be) quite unstable, so we had to create an alternative interactive web mapping application using ArcGIS as a backup.
Social network analysis became the most difficult part of the project because we couldn't find relevant metadata documenting interpersonal relationships in these digital collections. Using the co-occurrence of two names mentioned together in the same document as a criterion for possible undirected relationships, however, helped us gain insights into some of the possible connections between the personal names identified in these digital collections.
Due to time and budget constraints, this project concentrated on metadata (personal and corporate names, subject headings, geographic locations, and dates) rather than the content of all Republican China-related items in these digital collections. Because not all items have adequate and detailed metadata, our findings may not fully represent the collections. As a result, in future work, we will need to examine the content of each item more closely, attempting to identify and analyze additional names, organizations, themes, interpersonal relationships, and other important information that may have been overlooked during the metadata analysis. We will also need to implement new tools and methodologies for close reading and textual analysis to improve efficiency and accuracy.