Quantum technology for use with extracting intents from linguistics

A three-stage method for mapping a user utterance onto a plurality of intents, the method comprising:

• at a first stage, said first stage comprising training a system to determine an intent based on an utterace:
• receiving intent training data at an intent training database, the intent training data comprising a plurality of training utterances and a plurality of intents, together with a relationship between each intent, included in the plurality of intents, and the plurality of training utterances;
• canonicalizing the plurality of training utterances into a plurality of tokens, said tokens comprising a plurality of unigrams, bigrams and trigrams;
• forming a token-intent map, said token-intent map comprising a plurality of intents as column identifiers and a plurality of tokens as row identifiers;
• generating a token-row map based on the token-intent map, said token-row map comprising each token and ordinal value of the row in which the token appears in the intent training database;
• generating an intent-column map based on the token-intent map, said intent-column map comprising each intent and the column in which the intent appears;
• mapping, using the token-row map and the intent-column map, the plurality of tokens on a token-intent matrix, said token-intent matrix comprising each token, each intent and a score for each token relative to each intent, said score being based on the data included in the intent training database;
• executing a decomposition on the token-intent matrix;
• generating, from the decomposition, a token-cognitive matrix, a cognitive-comprehension matrix and an intent-cognitive matrix, said cognitive-comprehension matrix being the space of entanglement between the token-cognitive matrix and the intent-cognitive matrix;
• reducing the rank of the cognitive-comprehension matrix;
• computing a plurality of token vectors from a computation of the token-cognitive matrix and the cognitive-comprehension matrix;
• and
• computing a plurality of intent-cognitive vectors from a computation of the cognitive-comprehension matrix and the intent-cognitive matrix;
• at a second stage, said second stage comprising determining the intent of an utterance at runtime:
• receiving an utterance from a user;
• canonicalizing the utterance into a plurality of utterance tokens;
• comparing each of the utterance tokens to the plurality of token vectors;
• retrieving the token vectors that match the plurality of utterance tokens;
• calculating a centroid token vector from the retrieved token vectors;
• executing a distance computation between the centroid token vector and each of the plurality of intent-cognitive vectors;
• generating a scalar value from each distance computation between the centroid token vector and the plurality of intent-cognitive vectors;
• ordering a plurality of intent-cognitive vectors based on each intent-cognitive vector's associated scalar value;
• and
• presenting a predetermined number of intent-cognitive vectors with the highest scalar values to an ensemble processor;
• at a third stage, said third stage comprising processing by the ensemble processor:
• arranging, by the ensemble processor, a plurality of intents associated with the intent-cognitive vectors;
• presenting the plurality of intents to the user;
• retrieving one or more answers associated with the plurality of intents;
• and
• displaying the one or more answers to the user.

three-staged apparatus for mapping a user utterance onto a plurality of intents, the apparatus comprising:

• a first stage comprising:
• an intent training database comprising intent training data, said intent training data comprising a plurality of training utterances and intents;
• a processor configured to:
• canonicalize the plurality of training utterances into a plurality of tokens, the tokens comprising a plurality of unigrams, bigrams and trigrams;
• generate a token-intent map, said token-intent map comprising a plurality of intents as column identifiers and a plurality of tokens as row identifiers;
• generate a token-row map based on the token-intent map, said token-row map comprising each token and the row in which the token appears in the intent training database;
• generate an intent-column map based on the token-intent map, said intent-column map comprising each intent and the column in which the intent appears;
• map, using the token-row map and the intent-column map, the plurality of tokens on a token-intent matrix, said token-intent matrix comprising each token, each intent and a score for each token relative to each intent, said score being based on the data included in the intent training database;
• execute a decomposition on the token-intent matrix;
• generate, from the decomposition, a token-cognitive matrix (U-matrix), a cognitive-comprehension matrix (S-matrix) and an intent-cognitive matrix (V-matrix), said cognitive-comprehension matrix being the space of entanglement between the token-cognitive matrix and the intent-cognitive matrix;
• reduce the rank of the token-cognitive matrix;
• compute a plurality of token vectors from a multiplication of the token-cognitive matrix and the cognitive-comprehension matrix;
• compute a plurality of intent-cognitive vectors from a multiplication of the cognitive-comprehension matrix and the intent-cognitive matrix;
• a second stage comprising:
• a second processor operable to:
• receive an utterance from a user;
• canonicalize the utterance into a plurality of utterance tokens;
• compare each of the utterance tokens to the plurality of token vectors;
• retrieve the token vectors that match the plurality of utterance tokens;
• calculate a centroid token vector from the retrieved token vectors;
• execute a distance computation between the centroid token vector and each of the plurality of intent-cognitive vectors;
• generate a scalar value from each distance computation between the centroid token vector and the plurality of intent-cognitive vectors;
• order a plurality of intent-cognitive vectors based on each intent-cognitive vector's associated scalar value;
• and
• present a predetermined number of intent-cognitive vectors with the highest scalar values to an ensemble processor;
• a third stage comprising:
• the ensemble processor, said ensemble processor operable to:
• arrange a plurality of intents associated with the intent-cognitive vectors;
• present the plurality of intents to the user;
• retrieve one or more answers associated with the plurality of intents;
• and
• display the one or more answers to the user.