Word sense disambiguation using a deep logico-neural network

A computer system for word sense disambiguation, comprising:

• one or more processors, one or more computer-readable memories, one or more computer-readable tangible storage devices, and program instructions stored on at least one of the one or more storage devices for execution by at least one of the one or more processors via at least one of the one or more memories, wherein the computer system is capable of performing a method comprising:
• receiving, by one or more processors, input sentences comprising a plurality of words, wherein at least two words in the plurality of words are homonyms;
• converting, by the one or more processors, the plurality of words associated with each input sentence into a first vector comprising possible senses for the at least two words;
• combining, by the one or more processors, the first vector with a second vector comprising a domain-specific contextual vector associated with the at least two words;
• generating, by the one or more processors, a recurrent deep logico-neural network model by receiving a glossary and a labelled training dataset comprising a plurality of sentences, associating each of the sentences with context words obtained from semantic parsing, converting the glossary into a logical neural network using logic of related words, recurrently interleaving logical neural network layers and pruning layers to generate the recurrent deep logico-neural network model, and training the recurrent deep logico-neural network model using back-propagation based on the labelled training dataset;
• feeding, by the one or more processors, the combination of the first vector with the second vector into the recurrent deep logico-neural network model to generate a third vector comprising word senses for the at least two words;
• and
• thresholding, by the one or more processors, the third vector to generate a fourth vector comprising a final word sense vector for the at least two words.

A computer program product for word sense disambiguation, comprising:

• one or more computer readable storage media, and program instructions collectively stored on the one or more computer readable storage media, the program instructions comprising:
• program instructions to receive, by one or more processors, input sentences comprising a plurality of words, wherein at least two words in the plurality of words are homonyms;
• program instructions to convert, by the one or more processors, the plurality of words associated with each input sentence into a first vector comprising possible senses for the at least two words;
• program instructions to combine, by the one or more processors, the first vector with a second vector comprising a domain-specific contextual vector associated with the at least two words;
• program instructions to generate, by the one or more processors, a recurrent deep logico-neural network model by receiving a glossary and a labelled training dataset comprising a plurality of sentences, associating each of the sentences with context words obtained from semantic parsing, converting the glossary into a logical neural network using logic of related words, recurrently interleaving logical neural network layers and pruning layers to generate the recurrent deep logico-neural network model, and training the recurrent deep logico-neural network model using back-propagation based on the labelled training dataset;
• program instructions to feed, by the one or more processors, the combination of the first vector with the second vector into the recurrent deep logico-neural network model to generate a third vector comprising word senses for the at least two words;
• and
• program instructions to threshold, by the one or more processors, the third vector to generate a fourth vector comprising a final word sense vector for the at least two words.