Unsupervised learning of paraphrase/translation alternations and selective application thereof

• US 7,552,046 B2
• Filed: 11/15/2004
• Issued: 06/23/2009
• Est. Priority Date: 11/15/2004
• Status: Expired due to Fees

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Brockett, Christopher, Dolan, William, Breck, Eric
• Primary Examiner(s): Smits; Talivaldis Ivars
• Assistant Examiner(s): Ortiz Sanchez; Michael
• Application Number:
  US10/988,721
• Publication Number:
  US 20060106592A1
• Time in Patent Office: 1,681 Days
• Field of Search: 704/9, 704/1, 704/2, 704/3, 704/4
• US Class Current: 704/9
• CPC Class Codes: G06F 40/237 Lexical tools

Abstract

A computer-implemented method determines whether to apply a given paraphrase alternation pattern to an input string. The method includes generating a context model and applying the context model to determine whether the given paraphrase alternation pattern can be applied to the input string so as to preserve meaning.

Claims

1. A computer-implemented method for determining whether to apply a given paraphrase alternation pattern to an input string, the method comprising:

generating a context model based on a data set from which the given paraphrase alternation pattern was derived, the data set being database of news articles;

utilizing a computer processor that is a functional component of the computer to apply the context model to determine whether the given paraphrase alternation pattern can be applied to the input string so as to preserve meaning;

if it is determined that the given paraphrase alternation pattern can be applied so as to preserve meaning, then applying the given paraphrase alternation pattern to the input string, wherein the given paraphrase alternation pattern indicates a pattern of transformation from a first set of words to a second set of words, and wherein applying the given paraphrase alternation pattern comprises transitioning the input string from the first set of words to the second set of words, and wherein applying the given paraphrase alternation pattern further comprises applying the given paraphrase alternation pattern to the input string in a plurality of different ways to produce a plurality of different textual variations, and then applying the language model to the plurality of different textual variations to determine a probable sequence of words.

2. The method of claim 1, wherein generating a context model based on a data set from which the given paraphrase alternation model was derived comprises generating a context model based on data derived from a database of articles having a related subject matter.

3. The method of claim 1, wherein generating a context model based on a data set from which the given paraphrase alternation pattern was derived comprises generating a context model based on data including aligned sets of phrases containing at least one meaning-preserving alternation.

4. The method of claim 1, wherein generating a context model based on a data set from which the given paraphrase alternation pattern was derived comprises generating a context model based on data including aligned, meaning-related textual segments.

5. The method of claim 1, wherein applying the context model to determine comprises applying the context model to determine whether the input string is contextually consistent with data utilized as a rationale for the given paraphrase alternation pattern.

6. The method of claim 1, wherein applying the context model comprises comparing characteristics of the input string to characteristics of data utilized as a rationale for the given paraphrase alternation pattern.

7. The method of claim 1, wherein generating a context model comprises generating an assumption based on a capitalization pattern.

8. The method of claim 1, wherein generating a context model comprises generating an assumption based on a word length pattern.

9. The method of claim 1, wherein generating a context model comprises generating an assumption based on a word clustering pattern.

10. The method of claim 1, wherein generating a context model comprises generating an assumption based on a punctuation pattern.

11. The method of claim 1, wherein generating a context model comprises generating an assumption based on a syntax pattern.

12. The method of claim 1, wherein generating a context model comprises generating an assumption based on a pattern reflected in a usage of certain parts of speech.

13. The method of claim 1, wherein generating a context model comprises generating an assumption based on a pattern reflected in the identity of lexical components.

14. The method of claim 1, wherein if it is determined further comprises applying only application of the context model indicates that doing so will not create an improbable sequence of words.

15. The method of claim 1, wherein applying a language model comprises applying a language model that incorporates data that reflects a subject matter that is also reflected in data that is utilized as a basis for generation of the given paraphrase alternation pattern.