This needs a better methodology section. It's not a trivial undertaking to analyze the top 200 sci-fi films of each of the past 8 decades. It says it was automa

This needs a better methodology section. It's not a trivial undertaking to analyze the top 200 sci-fi films of each of the past 8 decades. It says it was automated with LLM sentiment analysis, but how? Did you get access to a content library that allowed you to feed the film stream itself to the models? Did you get transcripts? Do all of these films even have transcripts? Did you generate them if not? How do you deal with transcripts losing out on most of the setting you get from the visuals themselves? Do you have access to the screenplays somehow? Does the context window of these models really fit 1600 films at once?