Dialogue system and method of controlling the same

1. A dialogue system comprising a processor including a plurality of cells,

wherein each of the plurality of cells generates a probability distribution for a plurality of pre-stored words, and inputs a word included in an input sentence among the plurality of pre-stored words into a next cell in response to an input, and the plurality of cells output words corresponding to an input meaning representation, and

wherein the processor is configured to:

generate a meaning representation corresponding to an input sentence by performing Natural Language Understanding on the input sentence,

generate an output sentence corresponding to the input meaning representation based on Recurrent Neural network (RNN),

determine whether the input sentence cannot be processed,

calculate a parameter representing a probability of outputting the input sentence when the meaning representation corresponding to the input sentence is input, where the processor calculates the parameter based on an output probability of the word included in the input sentence from the generated probability distribution, and

determine whether the input sentence cannot be processed based on the calculated parameter.

2. The dialogue system of claim 1 , wherein each of the plurality of cells generates a ranking distribution for the plurality of pre-stored words and inputs the word included in the input sentence among the plurality of pre-stored words into the next cell in response to the input.

3. The dialogue system of claim 2 , wherein the processor calculates the parameter based on a ranking of words included in the input sentence in the generated ranking distribution.

4. The dialogue system of claim 2 , wherein the processor determines that the input sentence cannot be processed when the parameter is less than a reference value.

5. The dialogue system of claim 4 , wherein the input sentence that cannot be processed is an Out-of-domain (OOD) sentence.

6. The dialogue system of claim 1 , wherein each of the plurality of cells generates the probability distribution or rank distribution for the plurality of pre-stored words, and inputs the word having a highest output probability or output priority among the plurality of pre-stored words into the next cell in response to input when the input sentence can be processed.

7. The dialogue system of claim 6 , wherein the processor generates the output sentence consisting of words output from each of the plurality of cells.

8. A controlling method of a dialogue system, the method comprising:

generating a meaning representation corresponding to an input sentence by performing Natural Language Understanding on the input sentence; and

determining whether the input sentence cannot be processed using a processor which generates an output sentence corresponding to the input meaning representation based on Recurrent Neural network (RNN),

wherein the processor includes a plurality of cells outputting words corresponding to the input meaning representation, and

wherein the determining whether the input sentence cannot be processed includes:

generating a probability distribution for a plurality of pre-stored words,

inputting a word included in the input sentence among the plurality of pre-stored words into a next cell in response to the inputting of the word,

calculating a parameter representing a probability of outputting the input sentence when the meaning representation corresponding to the input sentence is input, where the parameter is calculated based on an output probability of the word included in the input sentence from the generated probability distribution, and

determining whether the input sentence cannot be processed based on the calculated parameter.

9. The method according to claim 8 , wherein the determining whether the input sentence cannot be processed includes:

generating a ranking distribution for the plurality of pre-stored words in response to the input by the each of the plurality of cells, and

inputting the word included in the input sentence among the plurality of pre-stored words into the next cell in response to the input.

10. The method according to claim 9 , wherein the determining whether the input sentence cannot be processed includes calculating the parameter based on a ranking of words included in the input sentence in the generated ranking distribution.

11. The method according to claim 8 , wherein the determining whether the input sentence cannot be processed includes determining that the input sentence cannot be processed when the parameter is less than a reference value.

12. The method according to claim 11 , wherein the input sentence that cannot be processed is an Out-of-domain (OOD) sentence.

13. The method according to claim 8 , further comprising:

generating the probability distribution or rank distribution for the plurality of pre-stored words by each of the plurality of cells, and

inputting the word having a highest output probability or output priority among the plurality of pre-stored words into the next cell in response to input upon determining that the input sentence can be processed.

14. The method according to claim 13 , further comprising generating the output sentence consisting of words output from each of the plurality of cells with the highest output probability or a highest output rank upon determining that the input sentence can be processed.