Struggling to think of a synonym? Desperate for a definition? A new crossword-solving system could be the end of your lexicological lamentations.
The web-based system, developed by researchers from the UK, US and Canada, makes use of artificial neural networks, which are based on the brain’s own learning systems.
The freely-available software was found to be better at solving standard crossword clues than commercial products. It can cope with single words (e.g. ‘snazzy’ – smart), short combinations of words (e.g. ‘Catholic leader’ – Pope), and longer sentences and phrases (e.g. ‘inventor of General Relativity’ – Einstein).
The researchers trained the system to understand words, phrases and sentences by feeding it hundreds of thousands of definitions of English words from six dictionaries, as well as Wikipedia.
“Over the past few years, there’s been a mini-revolution in machine learning,” says Felix Hill at the University of Cambridge, one of the paper’s authors.
This approach could prove useful in other language-related tasks, such as automatic translation and speech recognition.
“Despite recent progress in AI, problems involving language understanding are particularly difficult,” says Hill. “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.”