Training apparatus and training method

1 . A training apparatus, comprising:

a Central Processing Unit (CPU) configured to:

acquire language data from a storage device, wherein

the language data is associated with a plurality of elements, and

the plurality of elements constitutes a plurality of sentences in a replaceable form;

receive designation of a method from a terminal device, wherein the designation is received for expansion of the language data and generation of the plurality of sentences;

generate expansion coefficients based on a machine learning algorithm and the received designation;

expand the language data based on a Backus-Naur Form (BNF) and the generated expansion coefficients;

generate training data based on the expansion of the language data, wherein the training data is associated with the plurality of sentences; and

train a machine learning model related to a natural language, wherein the machine learning model is trained based on the generated training data.

2 . The training apparatus according to claim 1 , wherein the CPU is further configured to:

receive the designation of the method associated with a number of the plurality of sentences;

expand the language data based on the designation associated with the number of the plurality of sentences, wherein the language data is expanded to generate a sentence of the plurality of sentences; and

train the machine learning model based on the generated sentence.

3 . The training apparatus according to claim 1 , wherein the CPU is further configured to:

receive the designation of the method associated with a weight related to expansion per replacement target element of the language data;

expand the language data based on the designation associated with the weight, wherein the language data is expanded to generate a sentence of the plurality of sentences; and

train the machine learning model based on the generated sentence.

4 . The training apparatus according to claim 3 , wherein the CPU is further configured to:

weight expansion per element of the plurality of elements based on a number of characters of a feature that constitutes the replacement target element; and

expand the language data based on the weighted expansion per element.

5 . The training apparatus according to claim 1 , wherein the CPU is further configured to:

classify each element of the plurality of elements based on similarity of features of a respective element of the plurality of elements;

extract the features in an order from a classified group based on the classified plurality of elements; and

expand the language data based on the extracted features, wherein the language data is expanded to generate a sentence of the plurality of sentences.

6 . The training apparatus according to claim 1 , wherein the CPU is further configured to:

receive the designation of the method associated with a training time to train the machine learning model;

expand the language data based on the designation associated with the training time, wherein the language data is expanded to generate a sentence of the plurality of sentences; and

train the machine learning model based on the sentence.

7 . The training apparatus according to claim 1 , wherein the CPU is further configured to:

accept the designation of the method associated with an accuracy to train the machine learning model;

expand the language data based on the designation associated with the accuracy, wherein the language data is expanded to generate a sentence of the plurality of sentences; and

train the machine learning model based on the generated sentence.

8 . The training apparatus according to claim 7 , wherein the CPU is further configured to: receive a determination request from a user based on the training of the machine learning model, wherein the machine learning model is trained based on the designation associated with the accuracy receive a response from the user based on the determination request, wherein the response includes the designation of the method associated with improvement of the accuracy; expand the language data based on the designation associated with the improvement in the accuracy, wherein the language data is expanded to generate the sentence of the plurality of sentences; train the machine learning model a based on the generated sentence.

9 . The training apparatus according to claim 1 , wherein the CPU is further configured to:

control display of a user interface associated with a user of the machine learning model, wherein the user interface displays a result of the expansion of the language data;

receive the designation from the user based on the display of the user interface; and

train the machine learning model based on a sentence of the plurality of sentences, wherein the sentence includes the result of the expansion of the language data.

10 . A training method, comprising:

acquiring, by a computer, language data from a storage device, wherein

the language data is associated with a plurality of elements, and

the plurality of elements constitutes a plurality of sentences in a replaceable form;

receiving, by the computer, designation of a method from a terminal device, wherein the designation is received for expansion of the language data and generation of the plurality of sentences;

generating, by the computer, expansion coefficients based on a machine learning algorithm and the received designation;

expanding, by the computer, the language data based on a Backus-Naur Form (BNF) and the generated expansion coefficients;

generating, by the computer, training data based on the expansion of the language data, wherein the training data is associated with the plurality of sentences; and

training, by the computer, a machine learning model related to a natural language, wherein the machine learning model is trained based on the generated training data.

11 . A non-transitory computer-readable medium having stored thereon, computer-executable instructions which, when executed by a computer, cause the computer to execute operations, the operations comprising:

acquiring language data from a storage device, wherein

the language data is associated with a plurality of elements, and

the plurality of elements constitutes a plurality of sentences in a replaceable form;

receiving designation of a method from a terminal device, wherein the designation is received for expansion of the language data and generation of the plurality of sentences;

generating expansion coefficients based on a machine learning algorithm and the received designation;

expanding the language data based on a Backus-Naur Form (BNF) and the generated expansion coefficients;

generating training data based on the expansion of the language data, wherein the training data is associated with the plurality of sentences; and

training a machine learning model related to a natural language, wherein the machine learning model is trained based on the generated training data.