Here’s a crossword puzzle clue for you: A tall, long-necked spotted ruminant of Africa.
If you’re someone who knows African wildlife like the back of your hand, it may take you only a moment to come up with: Giraffe.
And congratulations, reader. You’re right.
Researchers from the Univ. of Cambridge, New York Univ., and Université de Montréal have developed a web-based platform that can assist you with your morning crossword puzzle and act as a reverse dictionary. But as fun as solving crossword puzzles might be, the tool may have a deeper role to play when it comes to teaching artificial intelligence (AI) systems human language.
The research behind the tool was published in Transactions of the Association for Computational Linguistics.
“To compile a bank of dictionary definitions for training the model, we started with all words in the target embedding space. For each of these words, we extracted dictionary-style definitions from five electronic sources: Wordnet, The American Heritage Dictionary, The Collaborative International Dictionary of English, Wiktionary and Webster’s,” the researchers wrote in their study.
“To allow models access to more factual knowledge than might be present in a dictionary (for instance, information about specific entities, places or people) we supplemented this training data with information extracted from Simple Wikipedia,” they added.
The researchers released the code for their system online for future research usage.
The research represents an early step towards endowing machines with the ability of human language. Deep learning—which is when scientists feed an artificial neural network with massive amounts of data—is integral to this process.
“Despite recent progress in AI, problems involving language understanding are particularly difficult, and our work suggests many possible applications of deep neural network to language technology,” said co-author Felix Hill, of Cambridge’s Computer Laboratory, in a statement. “One of the biggest challenges in training computers to understand language is recreating the many rich and diverse information sources available to humans when they learn to speak and read.”
Currently, the way computers learn language, according to Hill, is similar to the psychological framework referred to as cognitivism. In order to understand the larger context of human communication though, researchers need to combine cognitivism with behaviorism, in order for a machine to be able to infer meaning.
According to Univ. of Cambridge, the researchers are looking into integrating behaviorist-style models of language learning and linguistic interaction into their system.
R&D 100 AWARD ENTRIES NOW OPEN:
Establish your company as a technology leader! For more than 50 years, the R&D 100 Awards have showcased new products of technological significance. You can join this exclusive community! Learn more.