Writing assistance using machine translation techniques

• US 7,908,132 B2
• Filed: 09/29/2005
• Issued: 03/15/2011
• Est. Priority Date: 09/29/2005
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Brockett, Christopher, Dolan, William
• Primary Examiner(s): Dorvil; Richemond
• Assistant Examiner(s): ORTIZ SANCHEZ, MICHAEL
• Application Number:
  US11/238,335
• Publication Number:
  US 20070073532A1
• Time in Patent Office: 1,993 Days
• Field of Search: 704/9, 704/2, 434156-185
• US Class Current: 704/9
• CPC Class Codes: G06F 40/253 Grammatical analysis; Style critique
• CPC Class Codes: G06F 40/58 Use of machine translation, e.g. for multi-lingual retrieval, for server-side translation for client devices or for real-time translation

Abstract

A system is configured to provide writing assistance within a monolingual input environment based on statistical machine translation techniques typically utilized to translate from an input language to a different target language.

Claims

1. A computer-implemented method for providing writing assistance to a user, the method comprising:

utilizing a processor that is a component of the computer to automatically generate a mapping between two strings of text that are in the same language, wherein automatically generating the mapping comprises applying statistical analysis so as to produce an alignment of words in a first of the two strings with words in the second one of the two strings;

identifying a syntactic relationship between two words separated by other words within the first of the two strings;

providing, based at least in part on the mapping, and based at least upon the identified syntactic relationship, a list of suggested substitutions that represent potential modifications that will resolve an error contained in an input received from the user, the suggested substitutions in the list reflecting a ranking indicative of which of the potential modifications is most likely to resolve the error, the ranking being based on an evaluation of the suggested substitutions in the list relative to a language model trained based on a collection of well-formed data and

wherein the language model is an n-gram language model smoothed to accommodate one or more unknown words; and

wherein the first of the two strings is an output from a machine translation system that includes the error and is identified as being a low quality translation of a collection of text, the second of the two strings being a higher quality translation of the collection of text.

2. The method of claim 1, further comprising building a model of the error based at least in part on the mapping.

3. The method of claim 1, wherein providing a list comprises providing a list to the user so as to support a user selection process.

4. The method of claim 1, wherein providing a list comprises providing an input to an automated selection algorithm.

5. The method of claim 1, wherein the n-gram language model is trained based on the collection of well-formed data.

6. The method of claim 1, wherein the collection of well-formed data is a collection of sentences collected from a news agency.

7. A computer-implemented method for providing writing assistance to a user, the method comprising:

providing the writing assistance based on an automated statistical comparison, by a computer processor that is a component of the computer, of a well-formed collection of text with a corresponding collection of text that is similar to the well-formed collection of text but includes an intentionally and manually inserted error, wherein the well-formed and corresponding collections of text are in the same language, and wherein the automated statistical comparison includes statistical alignment that factors in syntactical relationships between words in at least one of the well-formed and corresponding collections of text.

8. The method of claim 7, wherein the writing assistance provided is an indication of an error included in an input received from the user.

9. The method of claim 7, wherein the writing assistance provided is a list of modifications that will resolve an error contained in an input received from the user, the modifications in the list reflecting a ranking indicative of which of the modifications is most likely to provide an accurate remedy to resolve the error.

10. The method of claim 9, wherein the ranking is based on analysis relative to a language model trained based on a collection of well-formed data.

11. The method of claim 10, wherein the language model is an n-gram language model.

12. The method of claim 11, wherein the n-gram language model is smoothed to accommodate one or more unknown words.

13. A computer-implemented method for providing writing assistance to a user, the method comprising:

utilizing a computer processor that is a component of the computer to automatically generate a mapping between two strings of text that are in the same language, wherein automatically generating a mapping further comprises applying statistical analysis so as to produce a statistical alignment of the two strings of text. the statistical alignment accounting for syntactical relationships between words in at least one of the two strings;

providing, based at least in part on the mapping, an indication of an error contained in input received from the user; and

wherein the first one of the two strings of text is an output from a machine translation system that includes the error and is identified as being a low quality translation of a collection of text, the second of the two strings being a higher quality translation of the collection of text.

14. The method of claim 13, wherein said statistical alignment accounts for a syntactical relationship between words in at least one of the two strings.

15. The method of claim 13, wherein automatically generating a mapping comprises automatically generating a mapping without translating either of the two strings into a different language.

16. The method of claim 13, wherein the indication of the error includes a list of modifications that will resolve the error, the list reflecting a ranking indicative of which of the modifications is most likely to provide an accurate remedy to resolve the error.

17. The method of claim 16, wherein the ranking is based on analysis relative to a language model trained based on a collection of well-formed data.