Dialogue system, a dialogue method, a method of generating data for training a dialogue system, a system for generating data for training a dialogue system and a method of training a dialogue system

1. A dialogue system comprising:

an input for receiving input data relating to a speech or text signal originating from a user;

an output for outputting speech or text information specified by a dialogue act; and

a processor configured to:

generate features from the input signal;

for each of a plurality of classifier models, each classifier model corresponding to a dialogue slot, and for one or more values corresponding to the dialogue slot, input features generated from the input signal, the classifier model outputting a probability corresponding to each of three or more relations, wherein the relations specify the relation of the value to the dialogue slot;

update a belief state based on the outputs of the classifier models;

determine a system dialogue act by inputting information relating to the belief state into a policy model;

output speech or text information specified by the determined dialogue act at the output,

wherein generating features from the input signal comprises generating features for a slot and value combination, comprising:

converting the input signal to vector representations, replacing all instances of the slot and of the value with a slot vector representation and a value vector representation; and

inputting the vector representations to a first trained model, configured to output features,

wherein for the classifier model corresponding to the slot, the features generated for the slot and value combination are inputted to the classifier model for the value, the classifier model outputting a probability that each relation specifies the relation of the value to the slot.

2. The dialogue system according claim 1 , wherein generating features from the input signal further comprises:

converting the input signal to vector representations;

inputting the vector representations to the first trained model.

3. The dialogue system according to claim 1 , wherein features representing the value are also inputted to the classifier model.

4. The dialogue system according to claim 3 , wherein the processor is further configured to:

generate features from the previous dialogue system output or the previous system dialogue act, wherein features generated from the previous dialogue system output or the previous system dialogue act are inputted to the classifier models.

5. The dialogue system according to claim 4 , wherein generating features from the previous dialogue system output or system dialogue act comprises generating features for a slot and value combination, comprising:

converting the previous dialogue system output or system dialogue act to vector representations, replacing all instances of the slot and of the value with a slot vector representation and a value vector representation;

inputting the vector representations to the first trained model;

wherein for the classifier model corresponding to the slot, the features generated for the slot and value combination are inputted to the classifier model for the value, the classifier model outputting a probability that each relation specifies the relation of the value to the slot.

6. The dialogue system according to claim 5 , wherein generating features from the previous dialogue system output or system dialogue act further comprises:

converting the previous dialogue system output or system dialogue act to vector representations;

inputting the vector representations to the first trained model.

7. The dialogue system according to claim 1 , wherein the first trained model comprises a convolutional layer.

8. The dialogue system according to claim 1 , wherein a stored set of possible phrasings for one or more slots and values is used to determine the instances of the slots and of the values.

9. The dialogue system according to claim 1 , wherein for one or more of the classifier models, the plurality of relations comprises: an affirmative relation, a negative relation, and a not mentioned relation.

10. The dialogue system according to claim 1 , wherein for one or more of the classifier models, the plurality of relations comprises: a greater than relation, a less than relation, and a not mentioned relation.

11. A dialogue method, comprising:

receiving input data relating to a speech or text signal originating from a user;

generating features from the input signal;

for each of a plurality of classifier models, each classifier model corresponding to a dialogue slot, and for one or more values corresponding to the dialogue slot, inputting features generated from the input signal, the classifier model outputting a probability corresponding to each of three or more relations, wherein the relations specify the relation of the value to the dialogue slot;

updating a belief state based on the outputs of the classifier models;

determining a system dialogue act by inputting information relating to the belief state into a policy model;

outputting speech or text information specified by the determined dialogue act,

wherein generating features from the input signal comprises generating features for a slot and value combination, comprising:

converting the input signal to vector representations, replacing all instances of the slot and of the value with a slot vector representation and a value vector representation; and

inputting the vector representations to a first trained model, configured to output features,

wherein for the classifier model corresponding to the slot, the features generated for the slot and value combination are inputted to the classifier model for the value, the classifier model outputting a probability that each relation specifies the relation of the value to the slot.

12. A non-transitory computer-readable storage medium comprising computer readable code configured to cause a computer to perform the following:

receiving input data relating to a speech or text signal originating from a user;

generating features from the input signal;

for each of a plurality of classifier models, each classifier model corresponding to a dialogue slot, and for one or more values corresponding to the dialogue slot, inputting features generated from the input signal, the classifier model outputting a probability corresponding to each of three or more relations, wherein the relations specify the relation of the value to the dialogue slot;

updating a belief state based on the outputs of the classifier models;

determining a system dialogue act by inputting information relating to the belief state into a policy model;

outputting speech or text information specified by the determined dialogue act,

wherein generating features from the input signal comprises generating features for a slot and value combination, comprising:

converting the input signal to vector representations, replacing all instances of the slot and of the value with a slot vector representation and a value vector representation; and

inputting the vector representations to a first trained model, configured to output features,

wherein for the classifier model corresponding to the slot, the features generated for the slot and value combination are inputted to the classifier model for the value, the classifier model outputting a probability that each relation specifies the relation of the value to the slot.