Excellent! I hope there’s more along this line that you can post early in next week’s thread. Late week posts tend to get ignored.
Highlights in the full article:
We experimented with using topic models [3] to find topics that are the most malleable (topic: food, eat, eating, thing, meat and topic: read, book, lot, books, women), and the most resistant (topic: government, state, world, country, countries and topic: sex, women, fat, person, weight). However, topic model based features do not seem to bring predictive power to either of the tasks
Limitations to non-computational application:
The study doesn’t really try to, in the author’s words: ‘Attempt* to capture high-level linguistic properties’
Excellent! I hope there’s more along this line that you can post early in next week’s thread. Late week posts tend to get ignored.
Highlights in the full article:
Limitations to non-computational application:
The study doesn’t really try to, in the author’s words: ‘Attempt* to capture high-level linguistic properties’