Integrated authoring and translation system

• US 6,139,201 A
• Filed: 04/15/1996
• Issued: 10/31/2000
• Est. Priority Date: 12/22/1994
• Status: Expired due to Term

• Current Assignee: Caterpillar Incorporated
• Original Assignee: Caterpillar Incorporated
• Inventors: Baker, Kathryn L., McLaren, Bruce M., Leavitt, John Robert Russell, Nyberg, 3rd, Eric H., Nirenburg, Irene, Holm, Susan E., Nasjleti, David, Sweet, John F., Hill, Dennis A., Bauer, Michael M., Brownlow, Nicholas D., Carbonell, Jaime G., Huettner, Alison K., Busdiecker, Roy F., Hudson, David C., Rennich, Mervin L., Harris, Timothy J., Schmandt, Linda M., Lonsdale, Deryle W., Higdon, James W., Hayes, Philip J., Mitamura, Teruko, Andersen, Peggy M., Gallup, Sharlene L., Riebling, Eric H.
• Primary Examiner(s): Hayes, Gail
• Assistant Examiner(s): Tkacs, Stephen R.
• Application Number:
  US08/632,780
• Time in Patent Office: 1,660 Days
• Field of Search: 395/751, 395/752, 395/754, 395/757, 395/758, 395/759, 395/792, 395/793, 395/798, 395/1, 395/10, 395/12, 395/50, 395/60, 395/63
• US Class Current: 704/2
• CPC Class Codes: G06F 40/10 Text processing natural language analysis G06F40/20; semantic analysis G06F40/30; processing or translation of natural language G06F40/40
• CPC Class Codes: G06F 40/211 Syntactic parsing, e.g. based on context-free grammar [CFG] or unification grammars
• CPC Class Codes: G06F 40/232 Orthographic correction, e.g. spell checking or vowelisation
• CPC Class Codes: G06F 40/253 Grammatical analysis; Style critique
• CPC Class Codes: G06F 40/268 Morphological analysis
• CPC Class Codes: G06F 40/284 Lexical analysis, e.g. tokenisation or collocates
• CPC Class Codes: G06F 40/289 Phrasal analysis, e.g. finite state techniques or chunking
• CPC Class Codes: G06F 40/55 Rule-based translation
• 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

The present invention is a system of integrated, computer-based processes for monolingual information development and multilingual translation. An interactive text editor enforces lexical and grammatical constraints on a natural language subset used by the authors to create their text, which they help disambiguate to ensure translatability. The resulting translatable source language text undergoes machine translation into any one of a set of target languages, without the translated text requiring any postediting.

Claims

1. A computer-based system for monolingual document development, comprising:

a text editor adapted to accept interactively from an author input text written in a source language;

a language editor, which is an extension of said text editor, which interactively enforces lexical constraints and grammatical constraints on a natural language subset used by said author to create said input text, wherein said author is interactively aided in enforcing said lexical constraints and said grammmatical constraints on said input text so as to produce unambiguous constrained text;

a machine translation system, responsive to said language editor that is configured to translate said unambiguous constrained text into a foreign language; and

a domain model, which communicates with said language editor, wherein said domain model provides pre-determined domain knowledge and linguistic semantic knowledge about lexical units and of their combinations, so as to assist said language editor in said enforcement of said lexical and grammatical constraints wherein said domain model is a tripartite domain model, said tripartite domain model comprising,

a kernel which contains lexical information that is required by said language editor and said machine translation system, wherein said lexical information includes lexical items within said natural language subset along with associated semantic concepts, parts of speech, and morphological information,

a language editor domain model which contains information that is required only by said language editor, wherein said information includes at least one of a natural language subset of synonyms for items not within said natural language subset, a dictionary of definitions of said lexical items, and a set of examples of using said lexical items, and

a machine translation domain model which contains information which is required by only said machine translation system, said machine translation domain model includes a hierarchy of concepts used for unambiguous mapping and semantic verification in translation.

2. A computer-based system for monolingual document development, comprising:

a text editor adapted to accept interactively from a author information elements written in a source language;

a language editor, which is an extension of said text editor, which interactively enforces lexical and grammatical constraints on a natural language subset used by said author to create unambiguous constrained information elements, wherein said author interactively aids in enforcing said lexical and grammatical constraints on said input text so as to produce said unambiguous constrained information elements;

memory means for storing said unambiguous constrained information elements for subsequent use;

a machine translation system, responsive to said language editor, that is configured to translate said unambiguous constrained information elements into a foreign language; and

a domain model, which communicates with said language editor, wherein said domain model provides pre-determined domain knowledge and linguistic semantic knowledge about lexical units and of their combinations, so as to assist said language editor in said enforcement of said lexical and grammatical constraints wherein said domain model is a tripartite domain model, said tripartite domain model comprising,

a kernel which contains lexical information that is required by said language editor and said machine translation system, wherein said lexical information includes lexical items within said natural language subset along with associated semantic concepts, parts of speech, and morphological information,

a language editor domain model which contains information that is required only by said language editor, wherein said information includes at least one of a natural language subset of synonyms for items not within said natural language subset, a dictionary of definitions of said lexical items, and a set of examples of using said lexical items, and

a machine translation domain model which contains information which is required by only said machine translation system, said machine translation domain model includes a hierarchy of concepts used for unambiguous mapping and semantic verification in translation.

3. A computer-based method for monolingual document development, comprising the steps of:

(1) entering input text in a source language into a text editor

(2)checking, via a language editor, said input text against a pre-determined set of constraints stored in a domain model that provides pre-determined domain knowledge and linguistic semantic knowledge about lexical units and of their combinations, said pre-determined set of constraints includes a set of source sublanguage rules concerning vocabulary and grammar, wherein said domain model is a tripartite domain model, said tripartite domain model comprising,

a kernel which contains lexical information that is required by said language editor and a machine translation system, wherein said lexical information includes lexical items that satisfy said pre-determined set of constraints along with associated semantic concepts, parts of speech, and morphological information,

a language editor domain model which contains information that is required only by said language editor, wherein said information includes at least one of a subset of synonyms for items that do not satisfy said pre-determined set of constraints, a dictionary definitions of said lexical items, and a set of examples of using said lexical items, and

a machine translation domain model which contains information which is required by only said machine translation system, said machine translation domain model includes a hierarchy of concepts used for unambiguous mapping and semantic verification in translation;

(3) providing to an author interactive feedback relating to said input text, said interactive feedback indicating if said pre-determined set of constraints is met, said interactive feedback is performed subsequent to referring to said domain model which provides the necessary domain knowledge and linguistic semantic knowledge about lexical units and of their combinations, and grammar of a subset of a natural language; and

(4) producing, after completion of step (3), unambiguous constrained text.

4. A computer-based method of claim 3, wherein said pre-determined set of constraints includes a set of source sublanguage rules concerning vocabulary and grammar, wherein said interactive feedback is performed in order to make said input text conform with said set of source sublanguage rules and to eliminate ambiguities.

5. A computer-based method for monolingual document development, comprising the steps of:

(1) entering input text in a source language into a text editor;

(2) checking, via a language editor, said input text against a constrained source language;

(3) providing to an author interactive feedback relating to said source input text if non-constrained source language is present in said source input text until said author modifies said source input text into a constrained source text, said interactive feedback is performed after consulting a domain model which provides the necessary domain knowledge and linguistic semantic knowledge about lexical units and of their combinations, wherein said domain model is a tripartite domain model, said tripartite domain model comprising,

a kernel which contains lexical information that is required by said language editor and said a machine translation system, wherein said lexical information includes lexical items within said constrained source language along with associated semantic concepts, parts of speech, and morphological information,

a language editor domain model which contains information that is required only by said language editor, wherein said information includes at least one of a natural language subset of synonyms for items not within said constrained source language, a dictionary definitions of said lexical items, and a set of examples of using said lexical items, and

a machine translation domain model which contains information which is required by only said machine translation system, said machine translation domain model includes a hierarchy of concepts used for unambiguous mapping and semantic verification in translation;

(4) checking for syntactic grammatical errors and semantic ambiguities in said constrained source text by consulting said domain model; and

(5) providing to said author interactive feedback to remove said syntactic grammatical errors and said semantic ambiguities in said constrained source text to produce unambiguous constrained text.

6. A computer-based method for monolingual document development, comprising the steps of:

(1) entering into a text editor at least one information element created in a source language;

(2) checking, via a language editor, said at least one information element against a constrained source language;

(3) providing to an author interactive feedback relating to said at least one information element if non-constrained source language is present in said at least one information element until said at least one information element has been modified into a constrained source text, said interactive feedback is performed after referring to a domain model which provides the necessary domain knowledge and linguistic semantic knowledge about lexical units and their combinations, wherein said domain model is a tripartite domain model, said tripartite domain model comprising;

a kernel which contains lexical information that is required by said language editor and said a machine translation system, wherein said lexical information includes lexical items within said constrained source language along with associated semantic concepts, parts of speech, and morphological information,

a language editor domain model which contains information that is required only by said language editor, wherein said information includes at least one of a natural language subset synonyms for items not within said constrained source language, a dictionary of definitions of said lexical items, and a set of examples of using said lexical items, and

a machine translation domain model which contains information which is required by only said machine translation system, said machine translation domain model includes a hierarchy of concepts used for unambiguous mapping and semantic verification in translation;

(4) checking for syntactic grammatical errors and semantic ambiguities in said constrained source text by consulting said domain model;

(5) providing interactive feedback to said author to remove said syntactic grammatical errors and said semantic ambiguities in said constrained source text to produce at least one unambiguous constrained information element; and

(6) saving said at least one unambiguous constrained information element for later use.

7. A computer-based system for translating source language input text to a foreign language, comprising:

a text editor adapted to accept interactively from an author the input text written in a source language;

a language editor, which is an extension of said text editor, which interacts with said author to produce from said input text an unambiguous constrained source text by interactively enforcing vocabulary and grammatical constraints against a constrained source language;

a machine translation system, responsive to said language editor, which is configured to translate said unambiguous constrained source text into the foreign language; and

a domain model, which communicates with said language editor and said machine translation system, and which provides predetermined domain knowledge and linguistic semantic knowledge about lexical units and of their combinations, so as to aid in producing said unambiguous constrained source text and in said translation to the foreign language, wherein said domain model is a tripartite domain model, said tripartite domain model comprising,

a kernel which contains lexical informationthat is required by said language editor and said machine translation system, wherein said lexical information includes lexical items within said constrained source language along with associated semantic concepts, parts of speech, and morphological information,

a language editor domain model which contains information that is required only by said language editor, wherein said information includes at least one of a subset of synonyms for items not within said constrained source language, a dictionary definitions of said lexical items, and a set of examples of using said lexical items, and

a machine translation domain model which contains information which is required by only said machine translation system, said machine translation domain model includes a hierarchy of concepts used for unambiguous mapping and semantic verification in translation.

8. The system of claim 7, further comprising means for marking with a tag a portion of said input text which has been rendered unambiguous constrained text by said interactive enforcement, wherein said tag indicates translatability.

9. The system of claim 7, wherein said machine translation system operates in a translation server environment which allows multiple authors to use the system.

10. The system of claim 7, wherein said author operates on a workstation which is part of a computer network.

11. The system of claim 7, wherein said machine translation system includes an interpreter which is configured to translate said unambiguous constrained source text into interlingua.

12. The system of claim 7, wherein said language editor provides said interaction with said author in a batch mode.

13. The system of claim 7, further comprising a graphics editor adapted to create text labels, wherein said text labels can be edited by said author with the aid of said language editor and subsequently translated by said machine translation system.

14. The system of claim 7, wherein said constrained source language is a subset of a natural language, said constrained source language is specified as to lexicon and grammar.

15. The system of claim 7, wherein said language editor comprises a vocabulary checker and a grammar checker.

16. The system of claim 15, wherein said vocabulary checker checks said input text against a permitted lexicon and suggests alternatives to non-lexicon word choices.

17. The system of claim 15, wherein said grammar checker checks for compliance with predefined grammatical rules and suggests alternatives to undefined grammatical structures.

18. The system of claim 15, wherein said grammar checker provides feedback to the author concerning lexical ambiguities and structural ambiguities.

19. The system of claim 15, wherein said grammar checker provides a means for interactive disambiguation.

20. The system of claim 15, wherein said vocabulary checker includes a spell checker.

21. The system of claim 15, wherein said vocabulary checker is configured to identify words not included in said constrained source language.

22. The system of claim 7, wherein said input text is provided in blocks of information elements.

23. The system of claim 22, wherein said information elements contain tags which enable said information elements to be described in terms of their content and logical structure.

24. A computer-based system for monolingual document development and multilingual translation, comprising:

a text editor adapted for accepting interactively from an author information elements written in a source language;

a language editor, which is an extension of said text editor, which interactively enforces lexical and grammatical constraints on a natural language subset used by said author to create said input text, wherein said author is interactively aided in enforcing said lexical and grammatical constraints on said information elements to produce said unambiguous constrained information elements;

machine translation system, responsive to said language editor, which translates said unambiguous constrained information elements into a foreign language; and

a domain model, which communicates with said language editor and said machine translation means, wherein said domain model provides pre-determined domain knowledge and linguistic semantic knowledge about lexical units and their combinations, so as to aid in producing said unambiguous constrained source text and in said translation to said foreign language, wherein said domain model is a tripartite domain model, said tripartite domain model comprising,

a kernel which contains lexical information that is required by said language editor and said a machine translation system, wherein said lexical information includes lexical items within said natural language subset along with associated semantic concepts, parts of speech, and morphological information,

a language editor domain model which contains information that is required only by said language editor, wherein said information includes at least one of a natural language subset of synonyms for items not within said natural language subset, a dictionary definitions of said lexical items, and a set of examples of using said lexical items, and

a machine translation domain model which contains information which is required by only said machine translation system, said machine translation domain model includes a hierarchy of concepts used for unambiguous mapping and semantic verification in translation.

25. A computer-based system for monolingual document development and multilingual translation, comprising:

(A) a text editor adapted to accept interactively from an author input text written in a source language;

(B) a language editor, which is an extension of said text editor, which interactively enforces lexical and grammatical constraints on a natural language subset used by said author to create said input text, said language editor comprising,

(i) a vocabulary checker which identifies occurrences of words that do not conform to said lexical constraints and which interactively aids said author in finding valid lexical replacements for said words that do not conform, and

(ii) a grammar checker which provides interactive feedback to said author concerning syntactic and semantic ambiguity, said interactive feedback producing unambiguous constrained text; and

(C) a domain model which communicates with said language editor, wherein said domain model provides pre-determined domain knowledge and linguistic semantic knowledge about lexical units and their combinations; and

(D) a machine translation system, responsive to said language editor, which is configured to translate said unambiguous constrained text into a foreign language,;

wherein said domain model is a tripartite domain model, said tripartite domain model comprising,

a kernel which contains lexical information that is required by said language editor and said a machine translation system, wherein said lexical information includes lexical items within said natural language subset along with associated semantic concepts, parts of speech, and morphological information,

a language editor domain model which contains information that is required only by said language editor, wherein said information includes at least one of a natural language subset of synonyms for items not within said natural language subset, a dictionary definitions of said lexical items, and a set of examples of using said lexical items, and

a machine translation domain model which contains information which is required by only said machine translation system, said machine translation domain model includes a hierarchy of concepts used for unambiguous mapping and semantic verification in translation.

26. A computer-based method for translating source language text to a foreign language, comprising the steps of:

(1) entering input text in a source language into a text editor;

(2) checking, via a language editor, said input text against a constrained source language;

(3) providing to an author interactive feedback relating to said source input text if nonconstrained source language is present in said source input text until said author modifies said source input text into a constrained source text, said interactive feedback includes allowing said author to select, from a list of at least one synonym, a word or phrase to replace said nonconstrained source language;

(4) checking for syntactic grammatical errors and semantic ambiguities in said constrained source text;

(5) providing interactive feedback to said author to remove said syntactic grammatical errors and said semantic ambiguities in said constrained source text to produce unambiguous constrained source text; and

(6) translating, via a machine translation system, said unambiguous constrained source text into a target language;

wherein steps (2) and (4) further include the step of communicating with a tripartite domain model (DM), wherein said tripartite DM provides predetermined domain knowledge and linguistic semantic knowledge about lexical units and their combinations, said tripartite domain model including,

a kernel which contains lexical information that is required by said language editor and said a machine translation system, wherein said lexical information includes lexical items within said constrained source language along with associated semantic concepts, parts of speech, and morphological information,

a language editor domain model which contains information that is required only by said language editor, wherein said information includes at least one of a set of synonyms for items not within said constrained source language, a dictionary of definitions of said lexical items, and a set of examples of using said lexical items, and

a machine translation domain model which contains information which is required by only said machine translation system, said machine translation domain model includes a hierarchy of concepts used for unambiguous mapping and semantic verification in translation.

27. The system of claim 26, further comprising the step of marking with a tag a portion of said input text which has been rendered unambiguous constrained source text, wherein said tag indicates translatability.

28. The method of claim 26, wherein said step of translating first includes the step of translating said constrained unambiguous text into interlingua.

29. The method of claim 26, wherein said step (2) of check ing comprises the steps of:

(a) checking a term from said source input text against a constrained source language (CSL) lexical knowledgebase;

(b) if the term is not found in said CSL lexical knowledgebase then,

(i) spellchecking said term against a standard dictionary and allowing said author to correct the spelling of said term if it is misspelled;

(ii) checking said term against said CSL lexical database; and

(iii) providing, if available, at least one CSL synonym from said domain model if said term is not in said CSL lexical knowledgebase, and allowing said author to choose one of said at least one synonym.

30. The method of claim 29, further comprising the step of repeating steps (a) and (b) for every term in said source input text.

31. The method of claim 29, further comprising the step of providing a list of related CSL words or phrases to said author if said term has no direct CSL synonyms.

32. The method of claim 29, further comprising the step of allowing said author to rewrite a sentence containing a non-CSL term.

33. The method of claim 26, further comprising the step of inserting a tag into said source input text after said author responds to said request for clarification of ambiguity.

34. The method of claim 26, wherein said source input text is created in blocks of information elements.

35. The method of claim 26, wherein said source input text is a text label in a graphic.

36. The method of claim 26, wherein step (3) comprises the step of presenting an indication of the two or more possible meanings of said source input text to said author.

37. A computer-based method for monolingual document development and multilingual translation, comprising the steps of:

(1) entering input text in a source language into a text editor;

(2) checking, via a language editor, said input text against a predetermined set of constraints stored in a domain model, wherein said predetermined set of constraints includes a set of source sublanguage rules concerning vocabulary and grammar, wherein said interactive feedback is performed in order to make said input text conform with said set of source sublanguage rules and to eliminate ambiguities, wherein said domain model is a tripartite domain model, said tripartite domain model comprising,

a kernel which contains lexical information that is required by said language editor and said a machine translation system, wherein said lexical information includes lexical items that satisfy said pre-determined set of constraints along with associated semantic concepts, parts of speech, and morphological information,

a language editor domain model which contains information that is required only by said language editor, wherein said information includes at least one of a set of synonyms for items that do not satisfy said pre-determined set of constraints, a dictionary of definitions of said lexical items, and a set of examples of using said lexical items, and

a machine translation domain model which contains information which is required by only said machine translation system, said machine translation domain model includes a hierarchy of concepts used for unambiguous mapping and semantic verification in translation;

(3) providing to an author interactive feedback relating to said input text if said predetermined set of criteria is not met, said interactive feedback is performed subsequent to consulting said domain model which provides the necessary domain knowledge and linguistic semantic knowledge about lexical units and their combinations, wherein said author produces, through said interactive feedback, unambiguous constrained source text;

(4) translating said unambiguous constrained source text into a target language.

38. The system of claim 37, further comprising the step of marking with a tag a portion of said input text which has been rendered unambiguous constrained text, wherein said tag indicates translatability.

39. A computer-based method for monolingual document development and multilingual translation, the computer-based method comprising the steps of:

(1) entering input text in a source language into a text editor;

(2) checking, via a language editor, said input text against a constrained source language;

(3) providing to an author interactive feedback relating to said source input text if nonconstrained source language is present in said source input text until said source input text has been modified into a constrained source text, said interactive feedback being done subsequent to consulting a domain model which provides the necessary domain knowledge and linguistic semantic knowledge about lexical units and their combinations, wherein said domain model is a tripartite domain model, said tripartite domain model comprising,

a kernel which contains lexical information that is required by said language editor and said a machine translation system, wherein said lexical information includes lexical items within said constrained source language along with associated semantic concepts, parts of speech, and morphological information,

a language editor domain model which contains information that is required only by said language editor, wherein said information includes at least one of a natural language subset of synonyms for items not within said constrained source language, a dictionary of definitions of said lexical items, and a set of examples of using said lexical items, and

a machine translation domain model which contains information which is required by only said machine translation system, said machine translation domain model includes a hierarchy of concepts used for unambiguous mapping and semantic verification in translation;

(4) checking for syntactic grammatical errors and semantic ambiguities in said constrained source text by consulting said domain model;

(5) providing interactive feedback to said author to remove said syntactic grammatical errors and said semantic ambiguities in said constrained source text to produce at lease one unambiguous constrained source text; and

(6) saving said at least one unambiguous constrained information element for later use;

(7) translating with said machine translation system said at least one unambiguous constrained source text into a foreign language.

40. A computer-based method for monolingual document development and multilingual translation, comprising the steps of:

(1) entering into a text editor at least one information element created in a source language;

(2) checking, via a language editor, said at least one information element against a constrained source language;

(3) providing to an author interactive feedback relating to said at least one information element if nonconstrained source language is present in said at least one information element until said at least one information element has been modified into a constrained source test, said interactive feedback is performed after consulting a domain model which provides the necessary domain knowledge and linguistic semantic knowledge about lexical units and of their combinations, wherein said domain model is a tripartite domain model, said tripartite domain model comprising,

a kernel which contains lexical information that is required by said language editor and said a machine translation system, wherein said lexical information includes lexical items within said natural language subset along with associated semantic concepts, parts of speech, and morphological information,

a language editor domain model which contains information that is required only by said language editor, wherein said information includes at least one of a natural language subset of synonyms for items not within said natural language subset, a dictionary of definitions of said lexical items, and a set of examples of using said lexical items, and

a machine translation domain model which contains information which is required by only said machine translation system, said machine translation domain model includes a hierarchy of concepts used for unambiguous mapping and semantic verification in translation;

(4) checking for syntactic grammatical errors and semantic ambiguities in said constrained text by consulting said domain model;

(5) providing interactive feedback to said author to remove said syntactic grammatical errors and said semantic ambiguities in said constrained source text to produce at least one unambiguous constrained information element;

(6) saving said at least one unambiguous constrained information element for later use;

(7) translating with said machine translation system said at least one unambiguous constrained information element into a foreign language.

41. The method of claim 40, further comprising the step of marking with a tag said information element certifying it to be translatable.

42. The method of claim 40, wherein step (3) of providing interactive feedback includes the step of allowing said author to select from a list of synonyms a word or phrase to replace said nonconstrained language in said at least one information element.