Artificial intelligence apparatus for learning natural language understanding models

1. An artificial intelligence apparatus comprising:

a memory; and

one or more processors configured to:

input a corpus of a first language into a machine translation model to obtain a corpus of a second language, and

based on the corpus of the second language being input, train a natural language understanding model to label the corpus of the second language with a command corresponding to the corpus of the first language, the command corresponding to the corpus of the first language having a same meaning as a command corresponding to the corpus of the second language, and to output the labeled command corresponding to the corpus of the first language,

wherein the memory stores a plurality of commands respectively corresponding to a plurality of corpora of the first language,

wherein the machine translation model is trained so as to output the corpus of the second language as a result value if the corpus of the first language is input as an input value,

wherein the corpus of the second language has the same meaning as the corpus of the first language,

wherein the first language and the second language are different languages from each other, and

wherein the one or more processors are configured to:

compare the corpus of the second language with a correct answer group to calculate a Bilingual Evaluation Understudy (BLEU) score,

train the natural language understanding model using the corpus of the second language if the BLEU score is greater than a threshold value,

obtain the command corresponding to the first corpus among the plurality of commands from the memory based on a second corpus of the second language being obtained as a first corpus of the first language among the plurality of corpora input into the machine translation model, and

set the second corpus as an input value and label the command corresponding to the first corpus as a result value to train the natural language understanding model.

2. The artificial intelligence apparatus of claim 1 , further comprising:

an input interface for obtaining second language data,

wherein the one or more processors are configured to:

extract the corpus of the second language from the second language data and input the extracted corpus of the second language into the natural language understanding model to perform an operation according to a command output by the natural language understanding model,

wherein the command output by the natural language understanding model is a command corresponding to the corpus of the first language that is labeled with the extracted corpus of the second language to have the same meaning as the command corresponding to the extracted corpus of the second language.

3. The artificial intelligence apparatus of claim 1 , further comprising:

Wherein one or more processors are configured to:

input the corpus of the first language into the machine translation model to obtain a corpus of a third language, and

based on the corpus of the third language being input, train the natural language understanding model to label the corpus of the third language with a command corresponding to the corpus of the first language, the command corresponding to the corpus of the first language having a same meaning as a command corresponding to the corpus of the third language, and to output the labeled command corresponding to the corpus of the first language, and

wherein the third language is a language different from the second language.

4. The artificial intelligence apparatus of claim 3 ,

wherein the input interface obtains third language data,

wherein the one or more processors are configured to:

extract the corpus of the third language from the third language data and input the extracted corpus of the third language into the natural language understanding model to perform an operation according to a command output by the natural language understanding model,

wherein the command output by the natural language understanding model is a command corresponding to the corpus of the first language that is labeled with the extracted corpus of the third language to have the same meaning as the command corresponding to the extracted corpus of the third language.

5. A method for operating an artificial intelligence apparatus comprising:

inputting a corpus of a first language into a machine translation model to obtain a corpus of a second language; and

based on the corpus of the second language being input, training a natural language understanding model to label the corpus of the second language with a command corresponding to the corpus of the first language, the command corresponding to the corpus of the first language having a same meaning as a command corresponding to the corpus of the second language, and to output the labeled a command corresponding to the corpus of the first language,

wherein a plurality of commands is stored each corresponding to a plurality of corpora of the first language,

wherein the machine translation model is trained so as to output the corpus of the second language as a result value if the corpus of the first language is input as an input value,

wherein the corpus of the second language has the same meaning as the corpus of the first language, and

wherein the first language and the second language are different languages from each other;

comparing the corpus of the second language with a correct answer group to calculate a BLEU score,

wherein the training the natural language understanding model includes training the natural language understanding model using the corpus of the second language based on the BLEU score being greater than a threshold value;

obtaining the command corresponding to the first corpus among the plurality of commands based on the second corpus of the second language being obtained as the first corpus of the first language among the plurality of corpora input into the machine translation model; and

setting the second corpus as an input value and labeling the command corresponding to the first corpus as a result value to train the natural language understanding model.

6. The method for operating an artificial intelligence apparatus of claim 5 , further comprising:

obtaining second language data; and

performing an operation according to a command output by the natural language understanding model by extracting the corpus of the second language from the second language data and inputting the extracted corpus of the second language into the natural language understanding model,

wherein the command output by the natural language understanding model is a command corresponding to the corpus of the first language that is labeled with the extracted corpus of the second language to have the same meaning as the command corresponding to the extracted corpus of the second language.

7. The method for operating an artificial intelligence apparatus of claim 5 , further comprising:

inputting the corpus of the first language into the machine translation model to obtain a corpus of a third language; and

based on the corpus of the third language being input, training the natural language understanding model to label the corpus of the third language with a command corresponding to the corpus of the first language, the command corresponding to the corpus of the first language having a same meaning as a command corresponding to the corpus of the third language, and to output the labeled command corresponding to the corpus of the first language,

wherein the third language is a language different from the second language.

8. The method for operating an artificial intelligence apparatus of claim 7 , further comprising:

obtaining third language data; and

performing an operation according to a command output by the natural language understanding model by extracting the corpus of the third language from the third language data and inputting the extracted corpus of the third language into the natural language understanding model,

wherein the command output by the natural language understanding model is a command corresponding to the corpus of the first language that is labeled with the extracted corpus of the third language to have the same meaning as the command corresponding to the extracted corpus of the third language.