Translation quality quantifying apparatus and method

• US 7,653,531 B2
• Filed: 08/25/2005
• Issued: 01/26/2010
• Est. Priority Date: 08/25/2005
• Status: Active Grant

• Current Assignee: MultiLing Corp.
• Original Assignee: MultiLing Corp.
• Inventors: Chandler, Brian R., Sneddon, Michael V.
• Primary Examiner(s): Smits; Talivaldis Ivars
• Assistant Examiner(s): KOVACEK, DAVID M
• Application Number:
  US11/212,962
• Publication Number:
  US 20070050182A1
• Time in Patent Office: 1,615 Days
• Field of Search: 704 1- 10, 704270-271, 704276-277
• US Class Current: 704/10
• CPC Class Codes: G06F 40/47 Machine-assisted translation, e.g. using translation memory
• CPC Class Codes: G06F 40/51 Translation evaluation

Abstract

A system for automating the quality evaluation of a translation. The system may include a computer having a processor and memory device operably connected to one another. A source text in a first language may be stored within the memory device. A target text comprising a translation of the source text into a second language may also be stored within the memory device. Additionally, a plurality of executables may be stored on the memory device and be configured to, when executed by the processor, independently identify a test sample comprising one or more blocks, each comprising a matched set having a source portion selected from the source text and a corresponding target portion selected from the target text.

Claims

1. A system to control discretion of a human reviewer of language translations, the system comprising:

a computer comprising a processor and memory device operably connected to one another;

a source text stored within the memory device and corresponding to a first language;

a target text stored within the memory device and comprising a translation of the source text in a second language, distinct from the first language; and

a plurality of executables stored on the memory device and executed by the processor, the plurality of executables comprising;

a metric module identifying a test sample containing substantially less than all the source text and comprising one or more blocks,

the metric module, wherein the one or more blocks each comprise a matched set having a source portion selected from the source text and a corresponding target portion selected from the target text,

the metric module further comprising a block definer, block selector, and error identifier,

the metric module, wherein the block definer controls the length of text forming the one or more blocks and determines the number of words contained within one of the source portion and target portion of at least one of the one or more blocks,

the metric module, wherein the block selector controls selection of the one or more blocks and determines at random the location of one of the source portions and target portion within the source text of at least one of the one or more blocks,

the metric module, wherein the error identifier defines error categories, each category thereof selected from an error in language translation to be detected if existing in the translation,

an error reporting module constraining error reporting by a human reviewer of the translation by presenting to the human reviewer a listing of the error categories from which the human reviewer may exclusively select and report errors in the translation,

an output module compiling the errors selected by the human reviewer for the test sample, and

the output module further characterizing errors in the translation by scoring the quality of the translation.

2. The system of claim 1, wherein:

the memory device further stores a plurality of dictionary entries matching words in the first language to words in the second language; and

the location is selected according to changes in the concentration of words contained within one of the source text and target text and also contained within the plurality of dictionary entries.

3. The system of claim 1, wherein the memory device comprises one or more electronic storage units.

4. The system of claim 3, wherein at least one of the one or more electronic storage units is positioned remotely from the processor.

5. The system of claim 4, wherein at least one of the one or more electronic storage units is connected to the processor via the Internet.

6. The system of claim 1, wherein the error categories comprise wrong term, omission, grammatical error, misspelling, and punctuation error.

7. A system comprising:

a first computer comprising a first processor and first memory device operably connected to one another;

a second computer comprising a second processor and second memory device operably connected to one another;

a first plurality of executables stored on the first memory device and executed to assist a human translator in translating a source text corresponding to a first language into a target text corresponding to a second language, distinct from the first language; and

a second plurality of executables stored on the second memory device and executed by the second processor, the second plurality of executables operably assisting a human reviewer score the quality of the translating, the second plurality of executables comprising;

a metric module identifying a test sample consisting of one or more blocks,

the metric module, wherein each of the one or more blocks comprises a matched set having a source portion selected from the source text and a corresponding target portion selected from the target text,

the metric module, wherein the one or more blocks collectively are substantially less than the entire source text,

the metric module, further comprising a block definer, block selector, and error identifier,

the metric module, wherein the block definer controls the length of text forming the one or more blocks,

the metric module, wherein the block selector controls the location of the one of the one or more blocks by determining at random the source portions and target portion within the source text of at least one of the one or more blocks,

the metric module, wherein the error identifier defines error categories, each category thereof selected from an error in language translation to be detected in the translating,

an error reporting module constraining error reporting by a human reviewer of the translation by presenting to the human reviewer a listing of the error categories from which the human reviewer may exclusively select and report errors in the translation,

an output module compiling the errors selected by the human reviewer for the test sample, and

the output module further characterizing errors in the translation by scoring the quality of the translation.

8. The system of claim 7, wherein the number of words contained within the source portion of at least one of the one or more blocks is dictated by the block definer.

9. The system of claim 8, wherein the location of the source portion within the source text of at least one of the one or more blocks is dictated by one of an explicit formula and a random selection.

10. The system of claim 7, further comprising an electronic storage device receiving the source text and target text from the first computer.

11. The system of claim 10, wherein the second computer receives the source text and the target text from the electrical storage device.

12. The system of claim 11, wherein the electrical storage device, first computer, and second computer are interconnected via a network.

13. The system of claim 12, wherein the network comprises the Internet.

14. A method to control discretion of a human reviewer of language translations, the method comprising:

selecting a computer having a processor and a memory device operably connected to one another;

selecting a source text corresponding to a first language;

selecting a target text comprising a translation of the source text into a second language, distinct from the first language and comprising substantially less than all of the source text;

executing by the processor, selected computer instructions comprising a metric module, an error reporting module, and an output module;

identifying by the metric module one or more blocks, each block comprising a matched set having a source portion, comprising substantially less than all the source text, selected from the source text and a corresponding target portion selected from the target text;

controlling by a block definer within the metric module a length of text forming the one or more blocks;

controlling by a block selector, in the metric module, at random, a location of the one or more blocks,

controlling by an error identifier, in the metric module, error categories, each category thereof selected from an error in language translation;

constraining, by the error reporting module, a human reviewer of the translation by presenting to the human reviewer a listing of the error categories from which the human reviewer may exclusively select to report errors in the translation;

reviewing by the human reviewer the one or more blocks to identify translation errors therein;

selecting by the human reviewer, from the listing, selected error categories corresponding to the translation errors; and

characterizing, by the output module, errors in the translation by scoring the quality of the translation.