System and method for generating manually designed and automatically optimized spoken dialog systems
View Patent ↗Disclosed herein are systems, computer-implemented methods, and tangible computer-readable storage media for generating a natural language spoken dialog system. The method includes nominating a set of allowed dialog actions and a set of contextual features at each turn in a dialog, and selecting an optimal action from the set of nominated allowed dialog actions using a machine learning algorithm. The method includes generating a response based on the selected optimal action at each turn in the dialog. The set of manually nominated allowed dialog actions can incorporate a set of business rules. Prompt wordings in the generated natural language spoken dialog system can be tailored to a current context while following the set of business rules. A compression label can represent at least one of the manually nominated allowed dialog actions.
1. A method comprising:
at each turn in a dialog, nominating via a processor, using a partially observable Markov decision process in parallel with a conventional dialog process, a set of allowed dialog actions and a set of contextual features;
outputting allowed dialog actions from the conventional dialog process to the partially observable Markov decision process;
selecting an optimal action from the set of allowed dialog actions using a machine learning algorithm to yield a selected optimal action;
generating a response based on the selected optimal action at each turn in the dialog.
2. The method of claim 1 , wherein the machine learning algorithm uses reinforcement learning.
3. The method of claim 1 , wherein the machine learning algorithm is partially observable Markov decision process based.
4. The method of claim 1 , wherein the set of allowed dialog actions incorporates a set of business rules.
5. The method of claim 4 , wherein prompt wordings in the generated natural language spoken dialog system are tailored to a current context while following the set of business rules.
6. The method of claim 1 , wherein a lower-dimensional feature vector represents one of the allowed dialog actions.
7. A system comprising:
a processor; and
a computer-readable storage medium having instructions stored which, when executed on the processor, result in the processor performing operations comprising:
at each turn in a dialog, nominating via a processor, using a partially observable Markov decision in parallel with a conventional dialog process, a set of allowed dialog actions and a set of contextual features;
outputting allowed dialog actions from the conventional dialog process to the partially observable Markov decision process;
selecting an optimal action from the set of allowed dialog actions using a machine learning algorithm to yield a selected optimal action;
generating a response based on the selected optimal action at each turn in the dialog.
8. The system of claim 7 , wherein the system uses reinforcement learning.
9. The system of claim 7 , wherein the system is partially observable Markov decision process based.
10. The system of claim 7 , wherein the set of allowed dialog actions incorporates a set of business rules.
11. The system of claim 10 , wherein prompt wordings in the generated natural language spoken dialog system are tailored to a current context while following the set of business rules.
12. The system of claim 7 , wherein a lower-dimensional feature vector represents one of the allowed dialog actions in the set of allowed dialog actions.
13. A computer-readable storage device having instructions stored which, when executed by a computing device, result in the computing device performing operations comprising:
at each turn in a dialog, nominating via a processor, using a partially observable Markov decision process in parallel with a conventional dialog process, a set of allowed dialog actions and a set of contextual features;
outputting allowed dialog actions from the conventional dialog process to the partially observable Markov decision process;
selecting an optimal action from the set of allowed dialog actions using a machine learning algorithm to yield a selected optimal action;
generating a response based on the selected optimal action at each turn in the dialog.
14. The computer-readable storage device of claim 13 , wherein the machine learning algorithm uses reinforcement learning.
15. The computer-readable storage device of claim 13 , wherein the machine learning algorithm is partially observable Markov decision process based.
16. The computer-readable storage device of claim 13 , wherein the set of manually nominated allowed dialog actions incorporates a set of business rules.
17. The computer-readable storage device of claim 16 , wherein prompt wordings in the generated natural language spoken dialog system are tailored to a current context while following the set of business rules.