Interactive Multilingual Word-Alignment Techniques

• US 20110246173A1
• Filed: 04/01/2010
• Published: 10/06/2011
• Est. Priority Date: 04/01/2010
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Li, Henry, Liu, Xiaohua, Wei, Hao, Zhou, Ming, Scott, Matthew Robert
• Granted Patent: US 8,930,176 B2
• Time in Patent Office: Days
• US Class Current: 704/2
• CPC Class Codes: G06F 40/45 Example-based machine translation; Alignment
• CPC Class Codes: G06F 40/51 Translation evaluation

Abstract

Techniques for interactively presenting word-alignments of multilingual translations and automatically improving those translations based upon user feedback are described herein. With one or more implementations of the techniques described herein, a word-alignment user-interface (UI) concurrently displays a pair of bilingual sentences, where one is a translation of the other, and interactively highlights linked (i.e., “word-aligned”) words and phrases of the pair. Other implementations of the techniques described herein offer an option for a user to provide feedback about the existing word-alignments or realign the words or phrases. In still other described implementations, word-alignment is automatically improved based upon that user feedback.

Claims

1. One or more computer-readable media storing processor-executable instructions that, when executed, cause one or more processors to perform operations that facilitate interactive exposing of word-alignments between a bilingual sentence pair, the operations comprising:

concurrently displaying each sentence of the bilingual sentence pair via a user-interface (UI);

receiving a user selection of an of-interest word or phrase of a first sentence of the bilingual sentence pair;

in response to the receiving, performing actions including;

highlighting the of-interest word or phrase via the UI;

finding a linked word in a second sentence of the bilingual sentence pair that corresponds to the of-interest word;

highlighting the linked word via the UI.

2. One or more computer-readable media as recited in claim 1, wherein the finding and highlighting actions are performed while still concurrently displaying each sentence of the bilingual sentence pair via the UI.

3. One or more computer-readable media as recited in claim 1, wherein the performing actions further comprise:

finding a linked phrase in the second sentence of the bilingual sentence pair that corresponds to the of-interest word or phrase in the first sentence of the bilingual sentence pair, wherein the linked phrase includes more than one word;

highlighting the linked phrase via the UI.

4. One or more computer-readable media as recited in claim 1, wherein a user-directable position indicator is proximate to the of-interest word or phrase when input from a mouse causes a graphical cursor to hover over or near the of-interest word or phrase in the UI.

5. One or more computer-readable media as recited in claim 1, the operations further comprising, in response to the receiving, performing a dictionary look-up of the of-interest word and displaying the results of the dictionary look-up.

6. One or more computer-readable media as recited in claim 1, the operations further comprising, in response to the receiving, performing an online search engine query of the of-interest word and displaying the results of the query.

7. One or more computer-readable media as recited in claim 1, the operations further comprising emphasizing an emphasis word or phrase in each of the first and second sentences of the bilingual sentence pair, wherein the emphasis word or phrase of each sentence is word-aligned with a query word or phrase of the other sentence.

8. One or more computer-readable media as recited in claim 7, the operations further comprising before the concurrently displaying, receiving, performing, and emphasizing operations, searching based upon the query word or phrase, wherein the emphasis word or phrase includes the query word or phrase.

9. One or more computer-readable media as recited in claim 1, wherein the receiving of the user selection of the of-interest word or phrase includes determining that a user-directable position indicator is proximate to the of-interest word or phrase.

10. One or more computer-readable media as recited in claim 1, the operations further comprising:

determining whether a user selected a first word or phrase of the first sentence of the bilingual sentence pair;

in response to determining that the user selected the first word or phrase, highlighting the first word or phrase via the UI;

determining whether the user selected a second word or phrase of a second sentence of the bilingual sentence pair;

in response to determining that the user selected the second word or phrase, highlighting the second word or phrase via the UI;

storing a user-feedback word-alignment association between the first word or phrase and the second word or phrase.

11. One or more computer-readable media as recited in claim 1, the operations further comprising receiving a sentence-pair rating from a user, wherein the sentence-pair rating is an indication of an opinion of the user regarding a quality of the word-alignments between the bilingual sentence pair.

12. One or more computer-readable media storing processor-executable instructions that, when executed, cause one or more processors to perform operations that facilitate interactive collection of user feedback regarding word-alignments between a bilingual sentence pair, the operations comprising:

concurrently displaying each sentence of the bilingual sentence pair via a user-interface (UI) on an output display;

determining whether a user selected an of-interest word or phrase of a first sentence of the bilingual sentence pair;

in response to determining that the user selected the of-interest word or phrase of the first sentence, highlighting the of-interest word or phrase via the UI;

determining whether the user selected a linked word or phrase of a second sentence of the bilingual sentence pair;

in response to determining that the user selected the linked word or phrase of the second sentence, highlighting the linked word or phrase via the UI;

storing a user-feedback word-alignment association between the of-interest word or phrase of the first sentence and the linked word or phrase of the second sentence.

13. One or more computer-readable media as recited in claim 12, the operations further comprising repeating the operations of claim 12 for multiple of-interest words or phrases and multiple associated linked words or phrases.

14. One or more computer-readable media as recited in claim 12 further comprising measuring properties associated with the user selecting multiple of-interest words or phrases and associated linked words or phrases, wherein the properties include timing of a selection of each word relative to one or more selections of other words.

15. One or more computer-readable media as recited in claim 12, wherein the first sentence of the bilingual sentence pair is in a particular human natural language and the second sentence of the bilingual sentence pair is in another particular human natural language that is different from the language of the first sentence.

16. One or more computer-readable media as recited in claim 12, the method further comprising storing a word-alignment data structure comprising:

a sentence identifier field that identifies at least one of the sentences of the bilingual sentence pair;

a user identifier field that identifies the user who made the selections of the determining operations;

a rating field that indicates an opinion of the user regarding the quality of the word-alignments between the bilingual sentence pair;

a set of user-feedback word-alignment associations;

a user-specific confidence value associated with the user.

17. A method of improving word-alignments between a dataset of bilingual sentence pairs based upon collection of user feedback, the method comprising:

obtaining user-feedback ratings of existing word-alignment of sentence pairs of the dataset of bilingual sentence pairs;

dividing the dataset of bilingual sentence pairs into at least two groups, where a retained group includes sentence pairs that meet or exceed a quality standard based upon the obtained user-feedback ratings and a to-be-revised group includes sentence pairs having the existing word-alignment that fall below the quality standard based upon the obtained user-feedback ratings;

obtaining, from multiple users, user-feedback word-realignment of words in realigned sentence pairs of the dataset of bilingual sentence pairs;

based upon the obtained user-feedback word-realignment of the realigned sentence pairs, calculating a corrected word-realignment of the realigned sentence pairs;

generating a new word-alignment model based upon the retained group of sentence pairs and the realigned sentence pairs.

18. A method as recited in claim 17, the method further comprising applying the new word-alignment model to another dataset of bilingual sentence pairs.

19. A method as recited in claim as recited in claim 17, the method further comprising:

applying the new word-alignment model to another dataset of bilingual sentence pairs;

exposing, via a user-interface (UI) on an output display, bilingual sentence pairs of the another dataset of bilingual sentence pairs.

20. A method as recited in claim as recited in claim 17, wherein the calculating of the corrected word-realignment of the realigned sentence pairs includes:

calculating a user-specific confidence value based, at least, upon factors associated with the user;

repeating the calculating for many users;

weighing each realignment based upon the calculated user-specific confidence value for each user of the many users;

selecting the corrected word-realignment of the realigned sentence pairs based upon the weighted realignments.