University of Massachusetts, Amherst (UMass Amherst) researchers have developed an artificial intelligence (AI) machine that learned how to write political speeches that are very similar to real speeches.
The researchers used a database of nearly 4,000 political speech segments from 53 U.S. Congressional floor debates to train a machine-learning algorithm to produce speeches of its own. The researchers also categorized the speeches by political party and by whether it was in favor of or against a given topic.
The researchers designed the system using an approach based on n-grams, which are sequences of "n" words or phrases. The researchers analyzed the text using a parts-of-speech approach, which tags each word or phrase with its grammatical role before looking at 6-grams and the probability of a word or phrase appearing given the five that appear before it.
UMass Amherst researcher Valentin Kassarnig says the technique "allows us to determine very quickly all words which can occur after the previous five ones and how likely each of them is."
The speech-writing software explores the 6-gram database for a specific category of speech to find the entire set of 5-grams that have been used to start those speeches. The algorithm then chooses one of these 5-grams at random to start its speech.
From Technology Review
View Full Article
Abstracts Copyright © 2016 Information Inc., Bethesda, Maryland, USA
No entries found