by Graham Joncas & Nora Li

Written for DADH 2018: 9th International Conference of Digital Archives & Digital Humanities
 

Abstract (100 words):

This paper compares Hugo and Nebula award-winning short stories using text mining and logistic regression. Science fiction is known for its radical singularity: each text is an ‘event’ in the philosophical sense, creating a universe unto itself. In this light, unlike traditional criticism, quantitative methods generalize in the absence of unifying conventions or topoi. Parallel to Meillassoux's concept of extro-science fiction, digital humanities acts as ‘extro-criticism’ within fields of radical contingency (‘hyperchaos’). This asemic forensics not only traces 114 stories’ lexical detritus to each award’s institutional schemata, but presages xenographic re-mappings for conventional literary notions of ‘code’, ‘genre’, and ‘text’.

1. Unzip the data files into your working directory
2. Open sci-fi.R in RStudio (for R v3.5.1 or later)
3. Click ‘Source’ to extract the story metadata
4. To create charts, use the timeseries() function
5. For alternate regressions, redefine the scifi dataframe

• My first DH paper. The philosophy part is exciting, the results fail to live up to the hype.
• The writing was a bit rushed, so I still want to experiment a bit (e.g. SVMs, new variables)
• My code heavily uses the global assignment operator <<-, which feels like bad practice.
• NL thought of comparing the Hugo & Nebula awards, and collected the data.
• GJ thought of the theoretical framework, and wrote the code and prose.

 
This paper and code is licensed under CC BY 4.0
In short: it's fine to use the data, just don't plagiarize our paper.