Cross-lingual information extraction program

• US 10,042,846 B2
• Filed: 04/28/2016
• Issued: 08/07/2018
• Est. Priority Date: 04/28/2016
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Akbik, Alan, Chiticariu, Laura, Hailpern, Marina Danilevsky, Li, Yunyao, Zhu, Huaiyu
• Primary Examiner(s): Godbold, Douglas
• Application Number:
  US15/141,577
• Publication Number:
  US 20170315986A1
• Time in Patent Office: 831 Days
• CPC Class Codes: G06F 16/313 Selection or weighting of terms for indexing
• CPC Class Codes: G06F 40/169 Annotation, e.g. comment data or footnotes
• CPC Class Codes: G06F 40/205 Parsing
• CPC Class Codes: G06F 40/263 Language identification
• CPC Class Codes: G06F 40/30 Semantic analysis
• 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

One embodiment provides method for constructing a cross-lingual information extraction program, the method including: utilizing at least one processor to execute computer code that performs the steps of: constructing a plurality of language-specific representations from text expressed in a plurality of languages by parsing the text of each language using a language-specific semantic parser; mapping the plurality of language-specific representations to a single cross-lingual semantic representation, wherein the cross-lingual semantic representation encompasses the plurality of languages; and constructing the cross-lingual information extraction program based on the cross-lingual semantic representation. Other aspects are described and claimed.

Claims

1. A method for constructing a cross-lingual information extraction program, the method comprising:

utilizing at least one processor to execute computer code that performs the steps of;

constructing a plurality of language-specific representations from text expressed in a plurality of languages by parsing the text of each language using a language-specific semantic parser;

mapping the plurality of language-specific representations to a single cross-lingual semantic representation, wherein the cross-lingual semantic representation is language independent, language-invariant, and encompasses the plurality of languages, wherein the cross-lingual semantic representation comprises a graphical representation that identifies the semantic relationship between words contained within a text without regard for language of the text and comprises semantic and syntactic components;

constructing the cross-lingual information extraction program based on the cross-lingual semantic representation, wherein the cross-lingual information extraction program comprises a set of rules created from the cross-lingual semantic representation and wherein the cross-lingual information extraction program is language independent and facilitates extraction of structured information from unstructured text different from the text expressed in a plurality of languages; and

extracting structured information from texts in a language using the constructed cross-lingual information extraction program.

2. The method of claim 1, comprising extracting information from text by querying the cross-lingual information extraction program against a cross-lingual semantic representation of the text.

3. The method of claim 2, wherein the querying comprises requesting an argument associated with a target word contained within the text.

4. The method of claim 2, wherein the extracted information comprises structured information.

5. The method of claim 1, wherein the language-specific semantic parser comprises a semantic role labeler.

6. The method of claim 5, wherein at least one word contained within the text is identified as an action-inducing word.

7. The method of claim 6, wherein at least one word contained within the text is annotated with a semantic role with respect to the action-inducing word.

8. The method of claim 1, wherein the text expressed in a plurality of languages comprises syntactically different text of a single language.

9. An apparatus for constructing a cross-lingual information extraction program, the apparatus comprising:

at least one processor; and

a computer readable storage medium having computer readable program code embodied therewith and executable by the at least one processor, the computer readable program code comprising;

computer readable program code that constructs a plurality of language-specific representations from text expressed in a plurality of languages by parsing the text of each language using a language-specific semantic parser;

computer readable program code that maps the plurality of language-specific representations to a single cross-lingual semantic representation, wherein the cross-lingual semantic representation is language independent, language-invariant, and encompasses the plurality of languages, wherein the cross-lingual semantic representation comprises a graphical representation that identifies the semantic relationship between words contained within a text without regard for language of the text and comprises semantic and syntactic components;

computer readable program code that constructs the cross-lingual information extraction program based on the cross-lingual semantic representation, wherein the cross-lingual information extraction program comprises a set of rules created from the cross-lingual semantic representation and wherein the cross-lingual information extraction program is language independent and facilitates extraction of structured information from unstructured text different from the text expressed in a plurality of languages; and

computer readable program code that extracts structured information from texts in a language using the constructed cross-lingual information extraction program.

10. A computer program product for constructing a cross-lingual information extraction program, the computer program product comprising:

a computer readable storage medium having computer readable program code embodied therewith, the computer readable program code comprising;

computer readable program code that constructs a plurality of language-specific representations from text expressed in a plurality of languages by parsing the text of each language using a language-specific semantic parser;

computer readable program code that maps the plurality of language-specific representations to a single cross-lingual semantic representation, wherein the cross-lingual semantic representation is language independent, language-invariant, and encompasses the plurality of languages, wherein the cross-lingual semantic representation comprises a graphical representation that identifies the semantic relationship between words contained within a text without regard for language of the text and comprises semantic and syntactic components;

computer readable program code that constructs the cross-lingual information extraction program based on the cross-lingual semantic representation, wherein the cross-lingual information extraction program comprises a set of rules created from the cross-lingual semantic representation and wherein the cross-lingual information extraction program is language independent and facilitates extraction of structured information from unstructured text different from the text expressed in a plurality of languages; and

computer readable program code that extracts structured information from texts in a language using the constructed cross-lingual information extraction program.

11. The computer program product of claim 10, wherein the computer readable program code comprises code that extracts information from text by querying the cross-lingual information extraction program against a cross-lingual semantic representation of the text.

12. The computer program product of claim 11, wherein the computer readable program code that queries comprises code that requests an argument associated with a target word contained within the text.

13. The computer program product of claim 11, wherein the extracted information comprises structured information.

14. The computer program product of claim 10, wherein the language-specific semantic parser comprises a semantic role labeler.

15. The computer program product of claim 14, wherein at least one word contained within the text is identified as an action-inducing word.

16. The computer program product of claim 15, wherein at least one word contained within the text is annotated with a semantic role with respect to the action-inducing word.

17. The computer program product of claim 10, wherein the text expressed in a plurality of languages comprises syntactically different text of a single language.

18. A method for creating a cross-lingual information extraction program, the method comprising:

receiving a plurality of phrases, wherein the plurality of phrases comprises phrases each of which is expressed in more than one language;

parsing, using a language-specific semantic parser, each of the plurality of languages;

constructing a plurality of language-specific representations of each of the parsed plurality of languages;

mapping the plurality of language-specific representations to a language-invariant representation, wherein the language-invariant representation is language independent, encompasses the plurality of languages and comprises a graphical representation that identifies the semantic relationship between words contained within a phrase without regard for language of the phrase and comprises semantic and syntactic components;

creating the cross-lingual information extraction program using the language-invariant representation, wherein the cross-lingual information extraction program comprises a set of rules created from the language-invariant representation and wherein the cross-lingual information extraction program is language independent and facilitates extraction of structured information from unstructured text different from the text expressed in a plurality of languages; and

extracting structured information from texts in a language using the constructed cross-lingual information extraction program.

19. The method of claim 18, comprising applying the cross-lingual information extraction program to extract information from text in a language.

20. The method of claim 18, wherein the language-specific representation comprises a sense of an action-inducing word contained within the phrase and a semantic role of at least one argument of the action-inducing word.