System for Multiligual Machine Translation from English to Hindi and Other Indian Languages Using Pseudo-Interlingua and Hybridized Approach

• US 20080040095A1
• Filed: 04/06/2004
• Published: 02/14/2008
• Est. Priority Date: 04/06/2004
• Status: Abandoned Application

• Current Assignee: Indian Institute Of Technology And Ministry Of Communication And Information Technology, Department Of Information Technology
• Original Assignee: Indian Institute of Technology and Ministry of Communication and Information Technology
• Inventors: Jain, Ajai, Sinha, R. Mahesh
• Application Number:
  US11/547,803
• Publication Number:
  US 20080040095A1
• 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 40/55 Rule-based translation

Abstract

The present invention relates to a method and system for translating a source language into a target language comprising the steps of:—identifying the nature of text extracted from a source document, - filtering and storing the text formatting and structure information of the extracted text,—selecting an appropriate text translation engine based on the nature of the extracted text, —using the text translation engine for analysing and translating the extracted text into an unformatted translated text, and—using the stored text formatting and structure information to process the unformatted text for obtaining a structured translated text document in the target language.

Claims

1-40. -40. (canceled)

41. A method for translating a source language into a target language comprising the steps of:

identifying the nature of text extracted from a source document;

filtering and storing the text formatting and structure information of the extracted text;

selecting an appropriate text translation engine based on the nature of the extracted text;

using the text translation engine for analyzing and translating the extracted text into an unformatted translated text; and

using the stored text formatting and structure information to process the unformatted text for obtaining a structured translated text document in the target language.

42. The method as claimed in claim 41 further comprising the step of performing post editing on the structured translated text document for improving the accuracy of the translation and its presentation style.

43. The method as claimed in claim 42 wherein the post editing step is performed automatically on the structured translated text document for removing target language specific ambiguities and errors that maybe present.

44. The method as claimed in claim 42 wherein the post editing step is performed by a manually on the structured translated text document for removing ambiguities and errors that maybe present.

45. The method as claimed in claim 41 wherein nature of the extracted text is identified by a source language specific base includes running text with full sentences, running text with partial sentences, address, text heading, news heading, mathematical expression, table, a transcripted speech text, a text in mixed languages, footnotes, text within quote marks, parenthesized items and like.

46. The method as claim 41 wherein text portions having different nature are translated using different text translation engines.

47. The method as claimed in claim 41 wherein the step of analyzing the extracted text comprises the steps of:

identifying the sentence unit delimiter of the extracted text for breaking the text into separate sentences;

performing the lexical analysis on each word of the sentence using a domain specific lexical database for disambiguating the meaning and identifying acronyms;

abbreviations and unknown words in the sentence by identifying their domain, and storing the analyzed words (lexicons) along with their properties in an online-lexical and phrasal database and storing the unknown lexicons in a separate database for increasing the translation speed.

48. The method as claimed in claim 41 wherein the step of translating the extracted text comprises the steps of:

converting the analyzed text or a part of it to an intermediate form; and

translating the text in the intermediate form to the unformatted translated text said translation uses an abstracted example base comprising commonly encountered phrases, groups of Words and sentences.

49. The method as claimed in claim 48 wherein the analyzed text is compared with the entries in the abstracted example base and is substituted with its corresponding translation in the pseudo- interlingua, when a match is found, to obtain an intermediate translated text.

50. The method as claimed in claim 48 wherein the example base is expanded by adding new entries based on users'"'"' feedback on accuracy of the obtained translated output for improving the quality of the translation, wherein the example base can be expanded by adding new entries based on statistical information regarding the frequency of occurrence of the phrases in the source language for improving the quality of the translation.

51. The method as claimed in claim 48 wherein a rule based translation is done for the text or part of the text that are not present in the abstracted example base to obtain an intermediate translated text.

52. The method as claimed in claim 48 wherein a target language text generator is used for translating the intermediate text to the unformatted target language text wherein the text generator performs at least one of the following steps for translating the text in the intermediate form to the target language:

morphological synthesis of different lexicons for the target language, transliterating the unknown lexicons, generating an appropriate form for unknown lexicons in the target language;

establishing semantic and ontological relationship, using the history list of nouns and related rules for pronoun reference disambiguation, and composing and restructuring the target language document using the stored text formatting and structure information to obtain a structured translated text document.

53. A system for translating a source language into a target language comprising:

means for identifying the nature of text extracted from a source document wherein the source document includes a language specific knowledge base;

means for filtering and storing the text formatting and structure information of the extracted text;

means for selecting an appropriate text translation engine based on the nature of the extracted text;

means for analyzing and translating the extracted text into an unformatted translated text, using text specific translating engines, said translating and analyzing means further comprising;

means for identifying the sentence unit delimiter of the extracted text for breaking the text into separate sentences;

means for performing the lexical analysis on each word of the sentence; and

means for storing the analyzed words (lexicons) along with their properties in an online-lexical and phrasal database and storing the unknown lexicons in a separate database for increasing the translation speed maintaining a history of nouns for resolving pronoun reference abiguity; and

means for using the stored text formatting and structure information to process the unformatted text for obtaining a structured translated text document in the target language;

optionally comprising editing means for performing post editing on the structured translated text document for improving the accuracy of the translation and its presentation style.

54. The system as claimed in claim 53 wherein means for performing the lexical analysis is a hierarchical domain specific multilingual database that can be expanded by adding new domains and domain specific words, said hierarchical domain specific multilingual database is organized as a Directed Acyclic Graph linking domains and sub-domains and stores verbs and nouns using paradigm coding for morphological synthesis rules in translation.

55. The system as claimed in claim 53 wherein means for translating the lexicons into an intermediate text is an expandable abstracted target language specific example base comprising commonly encountered phrases, groups of words and sentences.

56. The system as claimed in claim 53 further comprising rule based translating means for translating the text or part of text not present in the abstracted example base into an intermediate text.

57. The system as claimed in claim 55 wherein means for translating the intermediate text to the target language text is a target language text generator, said target language text generator comprises:

means for morphological synthesis of different lexicons for the target language, means for transliterating the unknown lexicons;

means for generating an appropriate form for unknown lexicons in the target language, means for establishing semantic and ontological relationship, means for using the history list of nouns and related rules for pronoun reference disambiguation; and

means for composing and restructuring the target language document using the stored text formatting and structure information to obtain a structured translated text document.

58. The system as claimed in claim 53 wherein the computing system nodes for translating a source language into a target language comprises:

at least one system bus, at least one communication unit connected to the system bus, at least one memory unit connected to the system bus, wherein the memory includes a set of instructions, and at least one central processing unit connected to the system bus, wherein the central processing unit executes the instructions in the memory for translating a source language into a target language said system further connected to other similar systems and that may contain means to complement and supplement the aforementioned means.

59. A computer program product comprising computer readable program code stored on computer readable storage medium embodied therein for translating a source language into a target language, comprising:

computer readable program code means configured for identifying the nature of text extracted from a source document;

computer readable program code means configured for filtering and storing the text formatting and structure information of the extracted text;

computer readable program code means configured for selecting an appropriate text translation engine based on the nature of the extracted text;

computer readable program code means configured for analyzing and translating the extracted text into an unformatted translated text;

computer readable program code means configured for using the stored text formatting and structure information to process the unformatted text for obtaining a structured translated text document in the target language;

computer readable program code means configured to expand the example-base interactively; and

computer readable program code means configured to derive abstracted examples from the raw examples.

60. The computer program product as claimed in claim 59 further comprising computer readable program code means configured for performing post editing on the structured translated text document for improving the accuracy of the translation and its presentation style.