Evans, James and Pedro Aceves. 2016. “Machine Translation: Mining Text for Social Theory”. Annual Review of Sociology 42:21-50. DOI: 10.1146/annurev-soc-081715-074206
Based on the Evans and Aceves 2016 article, I have a question relating to my project: Given the four semantic representations on p. 31, how do we go about judging what kind of structure is most appropriate for a research question?
Some context for my question:
The Evans and Aceves 2016 article seems like a great primer In my project, I am particularly interested how a domestic discourse on the “the POW/MIA issue” may (or perhaps may not) have affected US foreign policy vis-à-vis Vietnam well into the 1980s and 1990s. I chose this course because I am interested in using computational methods to explore how this discourse changed over time, and I originally intended to begin with a swath of newspaper articles selected for the reference to Vietnam (I already have a test corpus assembled in the vein), then explore it through unsupervised topic modeling to 1) possibly confirm that the POW/MIA issue was a discursive formation in its own right (this seems like it calls for a more coarse structure), and 2) explore my intuition that the ‘live POW myth’ rose and fell in importance at different times (this seems like it calls for representation with a finer structure).
(Also, the approach described above isn't exactly unsupervised topic modeling in the strictest sense, since I've already made a decision on how to select these news articles. I would appreciate knowing how that might be an issue I have to deal with down the road.)
Based on the Evans and Aceves 2016 article, I have a question relating to my project: Given the four semantic representations on p. 31, how do we go about judging what kind of structure is most appropriate for a research question?
Some context for my question: The Evans and Aceves 2016 article seems like a great primer In my project, I am particularly interested how a domestic discourse on the “the POW/MIA issue” may (or perhaps may not) have affected US foreign policy vis-à-vis Vietnam well into the 1980s and 1990s. I chose this course because I am interested in using computational methods to explore how this discourse changed over time, and I originally intended to begin with a swath of newspaper articles selected for the reference to Vietnam (I already have a test corpus assembled in the vein), then explore it through unsupervised topic modeling to 1) possibly confirm that the POW/MIA issue was a discursive formation in its own right (this seems like it calls for a more coarse structure), and 2) explore my intuition that the ‘live POW myth’ rose and fell in importance at different times (this seems like it calls for representation with a finer structure). (Also, the approach described above isn't exactly unsupervised topic modeling in the strictest sense, since I've already made a decision on how to select these news articles. I would appreciate knowing how that might be an issue I have to deal with down the road.)