Comprehending the Digital Humanities
Topology
Topics
Documents
Data
Further Research
While I’d planned this exercise simply to better understand topic modeling and network modeling, as well as to provide an illustration of the two processes, I’ve been asked about analysis of the results beyond noting interesting topologies. Based on my experience, further areas of analysis would involve one or more of the following:
- Rigorous processing of the original texts, perhaps multiple types (Lemmatization and LAT) to see changes in the network based on the differing methods used to prepare the corpus
- Close reading of texts related to particular topic clusters to describe topics as a relationship of texts
- Close readings of texts related to a single topic to describe topics as a relationship of words
- Examination of network concepts such as centrality, betweenness and authority in relation to vernacular concepts–especially performed at multiple levels of aggregation to see if there is a particular scale at which there is the closest relationship between analytical concept and vernacular meaning
- 30-, 40- and 50-Topic sets to see if different topological structures arise
- A comparison of labeling of topics by a variety of scholars based on their interpretation of the topic’s connection to various words, as well as labeling of topics by a variety of scholars based on their interpretation of the topic’s connection to various papers