Dialog system

1. A dialog system, comprising:

an acquirer acquiring a first sentence, the acquirer including one selected from: a keyboard; a microphone; a touch panel; and

a processor,

the processor referring to a first weighting amount based on a first feature and a first weight, the first feature being of a first dialog state and relating to a first slot, the first weight being based on a weight parameter, the weight parameter being output from a model including a neural network by converting the first sentence into a feature vector and by inputting the feature vector as an input into the model including the neural network,

the processor referring to a second weighting amount based on a first intention amount and a second weight, the first intention amount being of a first intention and relating to the first slot, the first intention being included in the first sentence, the second weight pertaining to the first sentence and being different from the first weight,

the processor outputting a second feature by using the first weighting amount and the second weighting amount, the second feature being of a second dialog state and relating to the first slot, the second dialog state being after the first dialog state,

the processor generating a first dialog sentence based on the second dialog state including the second feature,

the processor outputting the first dialog sentence toward a user.

2. The system according to claim 1 , wherein the first weight pertains to a first degree of a mention of the first slot in the first sentence.

3. The system according to claim 2 , wherein

the first weight is larger than the second weight when the first degree is in a first state,

the second weight is larger than the first weight when the first degree is in a second state, and

the first degree in the first state is smaller than the first degree in the second state.

4. The system according to claim 1 , wherein

the first slot includes a plurality of first slot parameters, and

the processor calculates the first intention amount based on similarities between a first vector corresponding to the first sentence and vectors of the plurality of first slot parameters.

5. The system according to claim 1 , wherein

the first weighting amount is based on a product of the first feature and the first weight, and

the second weighting amount is based on a product of the first intention amount and the second weight.

6. The system according to claim 1 , wherein the second feature is based on a sum of the first weighting amount and the second weighting amount.

7. The system according to claim 1 , wherein

the processor further refers to a third weighting amount based on a third feature and a third weight, the third feature being of the first dialog state and relating to a second slot, the third weight pertaining to the first sentence,

the processor further refers to a fourth weighting amount based on a second intention amount and a fourth weight, the second intention amount being of the first intention and relating to the second slot, the fourth weight pertaining to the first sentence and being different from the third weight, and

the processor further outputs a fourth feature by using the third weighting amount and the fourth weighting amount, the fourth feature being of the second dialog state and relating to the second slot.

8. The system according to claim 1 , wherein

the acquirer further acquires a second sentence after the acquiring of the first sentence,

the processor further refers to a fifth weighting amount based on the second feature and a fifth weight, the fifth weight pertaining to the second sentence,

the processor further refers to a sixth weighting amount based on a second intention amount and a sixth weight, the second intention amount being of a second intention and relating to the first slot, the second intention being included in the second sentence, the sixth weight pertaining to the second sentence and being different from the fifth weight, and

the processor further outputs a fifth feature by using the fifth weighting amount and the sixth weighting amount, the fifth feature being of a third dialog state and relating to the first slot, the third dialog state being after the second d clog state.

9. The system according to claim 8 , wherein

the processor further refers to a seventh weighting amount based on the first feature and a seventh weight, the seventh weight pertaining to the second sentence and being different from the fifth weight and the sixth weight, and

the processor outputs the fifth feature by using the fifth weighting amount, the sixth weighting amount, and the seventh weighting amount.

10. A dialog system, comprising a processor,

when acquiring a first sentence desiring a first object and a second object, the processor using an Recurrent Neural Network (ANN) to extract a first weight parameter of a first slot parameter and a second weight parameter of a second slot parameter, the first slot parameter corresponding to the first object, the second slot parameter corresponding to the second object;

setting a first priority of the first slot parameter and a second priority of the second slot parameter to be high based on the first weight parameter and the second weight parameter, and

generating a first dialog sentence based on the first priority and the second priority and output the first dialog sentence toward a user,

when acquiring a second sentence negating the second object after the acquiring of the first sentence, the processor

setting the first priority to be higher than the second priority by changing the second weight parameter, and

generating a second dialog sentence based on the first priority and the second priority and output the second dialog sentence toward the user.

11. The system according to claim 10 , wherein when acquiring the first sentence, the processor sets the first priority and the second priority to be higher than a third priority of a third slot parameter, the third slot parameter corresponding to a third object.

12. A dialog system, comprising a processor,

when acquiring a first sentence desiring a first object, the processor

using an Recurrent Neural Network (RNN) to extract a first weight parameter of a first slot parameter and a second weight parameter of a second slot parameter, the first slot parameter corresponding to the first object, the second slot parameter corresponding to a second object;

setting a first priority of the first slot parameter to be higher than a second priority of the second slot parameter based on the first weight parameter and the second weight parameter, and

generating a first dialog sentence based on the first priority and the second priority and output the first dialog sentence toward a user,

when acquiring a second sentence desiring the second object after the acquiring of the first sentence, the processor

setting the second priority to be higher than the first priority by changing the second weight parameter, and

generating a second dialog sentence based on the first priority and the second priority and output the second dialog sentence toward the user.