System and method of prioritizing automated translation of communications from a first human language to a second human language

• US 20090099835A1
• Filed: 03/07/2008
• Published: 04/16/2009
• Est. Priority Date: 10/16/2007
• Status: Active Grant

• Current Assignee: Lockheed Martin Corporation (Martin Marietta Corporation)
• Original Assignee: Lockheed Martin Corporation (Martin Marietta Corporation)
• Inventors: Elliot, Russell, Poulin, Jeffrey S.
• Granted Patent: US 8,086,440 B2
• Time in Patent Office: Days
• US Class Current: 704/2
• CPC Class Codes: G06F 40/55 Rule-based translation

Abstract

A method of prioritizing the automated translation of communications relating to a predetermined topic includes capturing and inputting into a data processing system a translation-candidate communication rendered in a first human language. A first data set representative of the translation-candidate communication is stored in computer memory and parsed into communication sub-portions. Communication sub-portions are algorithmically selected for translation depending on their relatedness to the predetermined topic as determined by first-language extraction rules. Each selected communication sub-portion is translated to a translated-data-set sub-portion representative of that selected communication sub-portion in the second human language. Translated-data-set sub-portions are subjected to a secondary filtration process in accordance with which their relatedness to the predetermined topic is determined by second-language extraction rules. Those translated-data-set sub-portions determined to contain information sufficiently related to the predetermined topic, or a related sub-topic, are converted to converted-data-set sub-portions representative of the translated-data-set sub-portions in a human-intelligible format and outputted through a machine-to-human interface.

Claims

1. A method of prioritizing for automated translation from a first human language to a second human language communications relating to at least one predetermined topic, the method comprising:

capturing and inputting into a data processing system a translation-candidate communication rendered in the first human language and storing in computer memory associated with the data processing system, in a predetermined machine-readable format, a first data set representative of the contents of the translation-candidate communication in the first human language;

maintaining in computer memory a first-language prioritization protocol including data indicative of first-language extraction rules according to which a selected first-data-set sub-portion representative of a communication sub-portion of the translation-candidate communication is algorithmically one of (i) extracted and (ii) rejected for translation depending on whether the selected communication sub-portion exceeds a first relevancy threshold indicative of the relatedness of the communication sub-portion to the at least one predetermined topic;

consulting the first-language prioritization protocol and algorithmically analyzing, in accordance with the first-language extraction rules, the first data set in order to determine whether at least one communication sub-portion associated with the first data set exceeds the first relevancy threshold; and

one of

(i) selecting for translation to the second human language each communication sub-portion algorithmically determined to exceed the first relevancy threshold and (ii) rejecting for translation to the second human language each communication sub-portion algorithmically determined not to exceed the first relevancy threshold;

wherein

(a) as to a communication sub-portion selected for translation, the method further comprises

causing that sub-portion of the machine-readable first data set representative of the relevant communication sub-portion in the first human language to be translated to a translated-data-set sub-portion representative, in a machine-readable format, of the relevant communication sub-portion in the second human language;

converting at least a portion of the translated-data-set sub-portion into a converted-data-set sub-portion representative of at least a portion of the translated-data-set sub-portion in a human-intelligible format; and

outputting through a machine-to-human interface the converted-data-set sub-portion; and

(b) as to a communication sub-portion rejected for translation, the method further comprises one of (i) deleting from and (ii) archiving in computer memory the first-data-set sub-portion representative of that communication sub-portion.

2. The method of claim 1 further comprising parsing the first data set into first-data-set sub-portions correspondingly representative of equi-durational communication sub-portions of the translation-candidate communication.

3. The method of claim 2 wherein the first-data-set sub-portions are arbitrarily selected.

4. The method of claim 2 wherein each first-data-set sub-portion represents a communication sub-portion temporally centered upon at least one predetermined key term.

5. The method of claim 1 wherein the first-language prioritization protocol is updatable through at least one feedback path including a human-to-machine interface.

6. The method of claim 5 wherein the first-language prioritization protocol is updatable through the input of at least one of (i) text, (ii) images, and (iii) sound.

7. A method of prioritizing for automated translation from a first human language to a second human language communications relating to at least one predetermined topic, the method comprising:

capturing and inputting into a data processing system a translation-candidate communication rendered in the first human language and storing in computer memory associated with the data processing system, in a predetermined machine-readable format, a first data set representative of the contents of the translation-candidate communication in the first human language;

parsing the first data set into first-data-set sub-portions correspondingly representative of communication sub-portions of the translation-candidate communication;

maintaining in computer memory a set of relevancy thresholds including at least first and second relevancy thresholds indicative of the relatedness of a communication sub-portion to the at least one predetermined topic, wherein the first relevancy threshold indicates a greater degree of relatedness to the at least one predetermined topic than does the second relevancy threshold;

maintaining in computer memory a first-language prioritization protocol including data indicative of first-language extraction rules according to which a selected first-data-set sub-portion is algorithmically one of (i) extracted and prioritized for translation;

(ii) extracted and de-prioritized for translation and (iii) rejected for translation depending on whether the selected communication sub-portion, respectively, (a) exceeds the first relevancy threshold, (b) exceeds the second relevancy threshold, but not the first relevancy threshold, and (c) exceeds neither of the first and second relevancy thresholds;

consulting the first-language prioritization protocol and algorithmically analyzing, in accordance with the first-language extraction rules, the first data set in order to determine whether at least one communication sub-portion of the translation-candidate communication associated with the first data set exceeds either of the first and second relevancy thresholds; and

one of

(i) selecting for translation to the second human language each communication sub-portion algorithmically determined to exceed either of the first and second relevancy thresholds and (ii) rejecting for translation to the second human language each communication sub-portion algorithmically determined not to exceed either of the first and second relevancy thresholds;

wherein

(a) as to each communication sub-portion determined to exceed at least one of the first and second relevancy thresholds, the method further comprises causing the first-data-set sub-portion representative of that communication sub-portion in the first human language to be translated to a translated-data-set sub-portion representative, in a machine-readable format, of that communication sub-portion in the second human language;

(b) a first-data-set sub-portion representative of a communication sub-portion determined to exceed the first relevancy threshold is translated to a translated-data-set sub-portion prior to a first-data-set sub-portion representative of a communication sub-portion determined to exceed the second relevancy threshold and not the first relevancy threshold; and

(c) as to a communication sub-portion rejected for translation, the method further comprises one of (i) deleting and (ii) archiving in computer memory the first-data-set sub-portion representative of that communication sub-portion.

8. The method of claim 7 wherein the first-language prioritization protocol is updatable through at least one feedback path including a human-to-machine interface.

9. The method of claim 7 wherein the first data set is parsed into first-data-set sub-portions representative of equi-durational communication sub-portions of the translation-candidate communication.

10. The method of claim 9 wherein the first-data-set sub-portions are arbitrarily selected.

11. The method of claim 9 wherein each first-data-set sub-portion represents a communication sub-portion temporally centered upon at least one predetermined key term.

12. The method of claim 7 further comprising:

converting at least a portion of each translated-data-set sub-portion into a converted-data-set sub-portion representative of at least a portion of the translated-data-set sub-portion in a human-intelligible format; and

outputting through a machine-to-human interface the converted-data-set sub-portion.

13. The method of claim 7 further comprising:

maintaining in computer memory a second-language prioritization protocol including data indicative of second-language extraction rules according to which at least a sub-portion of a selected translated-data-set sub-portion is selected for one of (i) conversion, (ii) non-conversion, (iii) prioritized conversion, and (iv) de-prioritized conversion into a converted-data-set sub-portion representative of at least a portion of the translated-data-set sub-portion in a human-intelligible format depending on whether the selected translated-data-set sub-portion exceeds at least one predetermined extraction threshold;

consulting the second-language prioritization protocol and algorithmically analyzing, in accordance with the second-language extraction rules, at least one translated-data-set sub-portion associated with the first data set in order to determine whether the at least one translated-data-set sub-portion exceeds at least one predetermined extraction threshold; and

one of

(i) selecting for conversion each translated-data-set sub-portion algorithmically determined to exceed at least one extraction threshold and (ii) rejecting for conversion each translated-data-set sub-portion algorithmically determined not to exceed at least one extraction threshold;

wherein

(a) as to each translated-data-set sub-portion selected for conversion, the method further comprises (i) converting at least a portion of the translated-data-set sub-portion into a converted-data-set sub-portion representative of at least a portion of the translated-data-set sub-portion in a human-intelligible format and (ii) outputting through a machine-to-human interface the converted-data-set sub-portion; and

(b) as to a translated-data-set sub-portion rejected for conversion, the method further comprises one of (i) deleting from and (ii) maintaining in computer memory the translated-data-set sub-portion.

14. The method of claim 13 wherein

(i) the second-language prioritization protocol references at least two extraction thresholds indicative of mutually disparate levels of relatedness to a predetermined topic;

and

(ii) a first translated-data-set sub-portion that exceeds an extraction threshold indicative of a higher level of relatedness to the predetermined topic than the highest extraction threshold exceeded by a second translated-data-set sub-portion is converted prior to the second translated-data-set sub-portion.

15. The method of claim 13 wherein at least one of the first-language prioritization protocol and the first-language prioritization protocol is updatable through at least one feedback path including a human-to-machine interface through which can be inputted of at least one of (i) text, (ii) images, and (iii) sound.