Machine translation (MT) based spoken dialog systems customer/machine dialog

Patent metadata

Field	Value
Publication number	US-9646001-B2
Application number	US-201113236016-A
Country	US
Kind code	B2
Filing date	Sep 19, 2011
Priority date	Sep 19, 2011
Publication date	May 9, 2017
Grant date	May 9, 2017

Abstract

Operation of an automated dialog system is described using a source language to conduct a real time human machine dialog process with a human user using a target language. A user query in the target language is received and automatically machine translated into the source language. An automated reply of the dialog process is then delivered to the user in the target language. If the dialog process reaches an initial assistance state, a first human agent using the source language is provided to interact in real time with the user in the target language by machine translation to continue the dialog process. Then if the dialog process reaches a further assistance state, a second human agent using the target language is provided to interact in real time with the user in the target language to continue the dialog process.

First claim

What is claimed is:

1. A method of performing a real time human machine dialog process with a human user using a target language, the dialog process employing an automated dialog application stored in memory and implemented by at least one hardware implemented computer processor, the method comprising:
receiving with the automated dialog application a user query in the target language, the dialog application implementing a dialog between a human and a machine;
creating generic translation tables using large parallel corpora;
performing automatic machine translation of the user query into a source language, the source language being a language of the automated dialog application, wherein performing the automatic machine translation utilizes the created generic translation tables and includes mapping the query from the target language to the source language;
providing the dialog between the human and the machine by delivering an automated reply of the dialog application to the user in the target language through use of one or more statistical language models, the reply being a conversational response to the user query originated by the dialog application and selected from memory associated with the processor;
if the dialog process reaches an initial assistance state, providing a hybrid dialog between the human and a non-native agent through the machine by transferring the user to a first human agent using the source language to interact in real time with the user in the target language by machine translation to continue the dialog process; and
if the dialog process reaches a further assistance state, providing a dialog between the human and a native-agent by transferring the user, through the machine, to a second human agent using the target language to interact in real time with the user in the target language to continue the dialog process, the further assistance state reached when the initial assistance state with the first human agent is unsuccessful.

2. A method according to claim 1 , wherein delivering an automated reply includes delivering an existing dialog application reply in the target language.

3. A method according to claim 1 , wherein delivering an automated reply includes:
selecting an existing dialog application reply in the source language;
performing automatic machine translation of the selected reply to the target language; and
delivering the translated reply to the user.

4. A method according to claim 1 , wherein the dialog process is a text dialog process.

5. A method according to claim 1 , wherein the dialog process is a speech dialog process.

6. A dialog process system to conduct a real time human machine dialog with a human user using a target language, the system employing at least one hardware implemented computer processor and an automated dialog application stored in memory and implemented by the computer processor, the system comprising:
means for receiving with the automated dialog application a user query in the target language, the dialog application implementing a dialog between a human and a machine;
means for creating generic translation tables using large parallel corpora;
means for performing automatic machine translation of the user query into a source language, the source language being a language of the automated dialog application, wherein performing the automatic machine translation utilizes the created generic translation tables and includes mapping the query from the target language to the source language;
means for delivering an automated reply of the dialog application to the user in the target language through use of one or more statistical language models, delivering an automated reply providing the dialog between the human and the machine and the reply being a conversational response to the user query originated by the dialog application and selected from memory associated with the processor;
means for providing a hybrid dialog between the human and a non-native agent through the machine by transferring the user, if the dialog process reaches an initial assistance state, to a first human agent using the source language to interact in real time with the user in the target language by machine translation to continue the dialog process; and
means for providing a dialog between the human and a native-agent by transferring the user, through the machine, if the dialog process reaches a further assistance state, to a second human agent using the target language to interact in real time with the user in the target language to continue the dialog process, the further assistance state reached when the initial assistance state with the first human agent is unsuccessful.

7. A system according to claim 6 , wherein the means for delivering an automated reply includes means for delivering an existing dialog application reply in the target language.

8. A system according to claim 6 , wherein the means for delivering an automated reply includes:
means for selecting an existing dialog application reply in the source language;
means for performing automatic machine translation of the selected reply to the target language; and
means for delivering the translated reply to the user.

9. A system according to claim 6 , wherein the dialog process is a text dialog process.

10. A system according to claim 6 , wherein the dialog process is a speech dialog process.

11. A computer program product implemented in a non-transitory computer readable storage medium for performing a real time human machine dialog process with a human user using a target language, the dialog process employing an automated dialog application stored in memory and implemented by at least one hardware processor, the product comprising:
program code for receiving with the automated dialog application a user query in the target language, the dialog application implementing a dialog between a human and a machine;
program code for creating generic translation tables using large parallel corpora;
program code for performing automatic machine translation of the user query into a source language, the source language being a language of the automated dialog application, wherein performing the automatic machine translation utilizes the created generic translation tables and includes mapping the query from the target language to the source language;
program code for providing the dialog between the human and the machine by delivering an automated reply of the dialog application to the user in the target language through use of one or more statistical language models, the reply being a conversational response to the user query originated by the dialog application and selected from memory associated with the processor;
program code for providing a hybrid dialog between the human and a non-native agent through the machine by transferring the user, if the dialog process reaches an initial assistance state, to a first human agent using the source language to interact in real time with the user in the target language by machine translation to continue the dialog process; and
program code for providing a dialog between the human and a native-agent by transferring the user, through the machine, if the dialog process reaches a further assistance state, to a second human agent using the target language to interact in real time with the user in the target language to continue the dialog process, the further assistance state reached when the initial assistance state with the first human agent is unsuccessful.

12. A product according to claim 11 , wherein the program code for delivering an automated reply includes program code for delivering an existing dialog application reply in the target

Assignees

• Sarikaya Ruhi
• Goel Vaibhava
• Nahamoo David
• Tremblay Rèal
• Ramabhadran Bhuvana
• Stewart Osamuyimen
• Nuance Communications Inc

Inventors

• Sarikaya Ruhi
• Goel Vaibhava
• Nahamoo David
• Tremblay Rèal
• Ramabhadran Bhuvana
• Stewart Osamuyimen

Classifications

• H04M3/42

  Systems providing special services or facilities to subscribers (specially adapted for wireless communication networks H04W4/00) · CPC title

• H04M3/42391

  where the subscribers are hearing-impaired persons, e.g. telephone devices for the deaf · CPC title

• G10L15/22 Primary

  Procedures used during a speech recognition process, e.g. man-machine dialogue · CPC title

• H04M2242/12

  Language recognition, selection or translation arrangements · CPC title

• G06F40/58 Primary

  Use of machine translation, e.g. for multi-lingual retrieval, for server-side translation for client devices or for real-time translation · CPC title