Multi-turn dialogue response generation with persona modeling

1. A computer-implemented method, comprising:

initializing a machine classifier having a sequence to sequence network architecture, wherein the sequence to sequence network architecture comprises an encoder and a decoder;

training the machine classifier based on a training set comprising a plurality of input sequences, each input sequence comprising an encoder sequence and a corresponding decoder sequence, wherein training the machine classifier comprises:

generating an encoding of the encoder sequence and the corresponding decoder sequence for each of the plurality of input sequences in the training set; and

selecting a subset of the encodings;

for each of the subset of the encodings:

appending an informative padding to the encoding;

prepending a start of sequence token to the encoding of the encoder sequence;

appending an end of sequence token to the encoding of the corresponding decoder sequence;

training the encoder using the encoding of the encoder sequence; and

training the decoder using the encoding of the corresponding decoder sequence;

obtaining input data comprising a conversation history and a user utterance of a user;

determining a user intent based on the user utterance;

determining, by the machine classifier and based on the conversation history and the user utterance, a user persona for the user;

generating, by the machine classifier and based on the user intent and the user persona, a response persona;

generating, by the machine classifier and based on the user utterance and the response persona, a response; and

providing the response.

2. The computer-implemented method of claim 1 , wherein the response persona and the response are generated in parallel.

3. The computer-implemented method of claim 1 , wherein generating the response comprises:

generating an input encoding of the input data;

generating an output sequence comprising a start of sequence token;

completing the output sequence by iteratively generating one or more next output sequence tokens by providing the input encoding to the machine classifier and appending each generated next output sequence token, of the one or more next output sequence tokens, to the output sequence until a subsequent generated next output sequence token comprises an end of sequence token; and

generating the response based on the completed output sequence.

4. The computer-implemented method of claim 1 , wherein a response template is selected by the machine classifier based on the user intent.

5. The computer-implemented method of claim 4 , wherein the response template comprises pre-defined response language and one or more target slots.

6. The computer-implemented method of claim 5 , further comprising:

determining, based on the user intent and the user utterance, at least one entity; and

replacing, by the machine classifier, at least one of the one or more target slots in the response template with the determined at least one entity.

7. The computer-implemented method of claim 1 , wherein the user persona comprises attributes selected from a group comprising at least one of: a speaker's identity, a speaker's background, a speaker's location, or a speaker's preference.

8. The computer-implemented method of claim 1 , wherein:

the input data comprises a multi-turn dialog indicating a class of task; and

the method further comprises generating the response persona based on the class of task.

9. The computer-implemented method of claim 8 , further comprising generating the user persona based on the class of task.

10. The method of claim 1 , further comprising selecting, based on the response persona and from a plurality of response templates, a response template having an associated confidence metric that satisfies a threshold indicative of the response template being appropriate for the user persona, wherein the response is formatted in accordance with the response template.

11. A device, comprising:

a processor; and

a memory in communication with the processor and storing instructions that, when read by the processor, cause the device to:

initialize a machine classifier having a sequence to sequence network architecture, wherein the sequence to sequence network architecture comprises an encoder and a decoder;

train the machine classifier based on a training set comprising a plurality of input sequences, each input sequence comprising an encoder sequence and a corresponding decoder sequence, wherein the instructions cause the device to train the machine classifier by causing the device to:

generate an encoding of the encoder sequence and the corresponding decoder sequence for each of the plurality of input sequences in the training set; and

select a subset of the encodings and for each of the subset of the encodings:

append an informative padding to the encoding;

prepend a start of sequence token to the encoding of the encoder sequence;

append an end of sequence token to the encoding of the corresponding decoder sequence;

train the encoder using the encoding of the encoder sequence; and

train the decoder using the encoding of the corresponding decoder sequence;

obtain input data comprising a conversation history and a user utterance of a user;

determine a user intent based on the user utterance;

determine a class of task based on the conversation history and the user utterance;

determine, by the machine classifier and based on the conversation history, the user utterance, and the class of task, a user persona for the user;

generate, by the machine classifier and based on the user intent, the user persona, and the class of task, a response persona;

generate, by the machine classifier and based on the user utterance and the response persona, a response; and

provide the response.

12. The device of claim 11 , wherein the response persona and the response are generated in parallel.

13. The device of claim 11 , wherein the instructions, when read by the processor, further cause the device to generate the response by causing the device to:

generate an input encoding of the input data;

generate an output sequence comprising a start of sequence token;

complete the output sequence by iteratively generating one or more next output sequence tokens by providing the input encoding to the machine classifier and appending each generated next output sequence token, of the one or more next output sequence tokens, to the output sequence until a subsequent generated next output sequence token comprises an end of sequence token; and

generate the response based on the completed output sequence.

14. The device of claim 11 , wherein the instructions, when read by the processor, further cause the device to:

select by the machine classifier and based on the user intent, a response template; and

generate the response further based on the response template.

15. The device of claim 14 , wherein the response template comprises pre-defined response language and one or more target slots.

16. The device of claim 15 , wherein the instructions, when read by the processor, further cause the device to:

determine, based on the user intent and the user utterance, at least one entity; and

replace, by the machine classifier, at least one of the one or more target slots in the response template with the determined at least one entity.

17. The device of claim 11 , wherein the instructions, when read by the processor, further cause the device to select, based on the response persona and from a plurality of response templates, a response template having an associated confidence metric that satisfies a threshold indicative of the response template being appropriate for the user persona wherein the response is formatted in accordance with the response template.

18. A non-transitory, machine-readable medium storing instructions that, when executed by one or more processors, cause the one or more processors to perform steps comprising:

initializing a machine classifier having a sequence to sequence network architecture, wherein the sequence to sequence network architecture comprises an encoder and a decoder;

training the machine classifier based on a training set comprising a plurality of input sequences, each input sequence comprising an encoder sequence and a corresponding decoder sequence, wherein training the machine classifier comprises:

generating an encoding of the encoder sequence and the corresponding decoder sequence for each of the plurality of input sequences in the training set;

selecting a subset of the encodings and for each of the subset of the encodings:

appending an informative padding to the encoding;

prepending a start of sequence token to the encoding of the encoder sequence;

appending an end of sequence token to the encoding of the corresponding decoder sequence;

training the encoder using the encoding of the encoder sequence; and

training the decoder using the encoding of the corresponding decoder sequence;

obtaining input data comprising a conversation history and a user utterance of a user;

determining a user intent based on the user utterance;

determining, by the machine classifier, based on the conversation history and the user utterance, a user persona for the user;

generating, by the machine classifier and based on the user intent and the user persona, a response persona;

generating, by the machine classifier and based on the user utterance and the response persona, a response; and

providing the response.

19. The non-transitory, machine-readable medium of claim 18 , wherein the instructions, when executed by the one or more processors, further cause the one or more processors to generate the response by performing steps comprising:

generating an input encoding of the input data;

generating an output sequence comprising a start of output sequence token;

completing the output sequence by iteratively generating one or more next output sequence tokens by providing the input encoding to the machine classifier and appending each generated next output sequence token, of the one or more next output sequence tokens, to the output sequence until a subsequent generated next output sequence token comprises an end of output sequence token; and

generating the response based on the completed output sequence.

20. The non-transitory, machine-readable medium of claim 18 , wherein the instructions, when executed by the one or more processors, further cause the one or more processors to generate the response by performing steps comprising:

obtaining, by the machine classifier and based on the user intent and the response persona, a response template; and

generating the response further based on the response template.