Data processing: speech signal processing – linguistics – language – Linguistics – Translation machine
Reexamination Certificate
1998-11-13
2001-03-27
Isen, Forester W. (Department: 2747)
Data processing: speech signal processing, linguistics, language
Linguistics
Translation machine
C704S007000
Reexamination Certificate
active
06208956
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to the machine translation of documents using resources such as dictionaries and rule data bases. The invention is further related to using different dictionaries and data bases for translating different parts of a document. The invention is still further related to creating the document to be translated using tags such as Standard Generalized Markup Language (SGML) to indicate different sections of the document which are to use different translation resources.
2. Discussion of the Background
The concept of machine translation which translates a document from one human language to another has been widely studied. In the infancy of machine translation in the early 1940s, Warren Weaver of the Rockerfeller Foundation considered the concept of translation to be analogous to decoding in a manner using one-for-one substitution. However, it was soon realized that this was an overly simplistic analogy and that the subtleties of human language had to be taken into account in developing translation systems. For instance, the meaning of certain words in context has to be determined since specific phrases may have meanings that are not apparent from the individual words and some words have different meanings in different contexts.
One approach to improving the machine translation of a document is to impose restrictive rules on the words and phrases contained in the source document which is to be translated. In such a manner, one can obtain perfect translations but at the cost of radically restricting what an author can say. The basic idea of limiting the language which can be used in the source document which is to be translated is similar to the use of a translation phrase book often used by tourists which provides the user with a collection of “canned” phrases to use. This is fine, provided the canned text contains what the user wants to say. However, this is not convenient in many instances.
SUMMARY OF THE INVENTION
Accordingly, it is an object of the invention to provide a machine translation system and method which provides a more accurate translation than conventional machine translation systems but is not overly restrictive of the words or phrases which are contained in the source document. It is another object of the invention to provide a machine translation system and method that uses different translation resources such as dictionaries and rule data bases for different parts of a document to produce more accurate translations and minimize post-translation editing. It is a further object of the invention to utilize tags such as SGML tags to define the structure of the document to be translated in order to select the dictionaries and/or rules which are to be used for the translation process.
It is yet another object of the invention to provide an interactive translation system which allows a user to display both the source document and the machine translated document side-by-side which allows the user to interactively change the translated document to have a more preferable translation. It is still another object of the invention to update dictionaries and/or translation rules based on the changes which the user has made to the automatic translation.
These and other objects are accomplished by a method, system, and computer program product which translate, with the assistance of a machine such as a computer, a document from one human language to another. The present invention utilizes different translation resources or translation information such as dictionaries or translation rules for different sections of the document. The sections of the document are indicated through the use of tags which indicate portions or sections of documents. The tags are preferably tags which are created in accordance with the Standard Generalized Markup Language (SGML).
Before any translation can be performed, a Document Type Definition (DTD) must be created which defines the permissible tags and the permissible structure of a document. The DTD defines the permissible structure and hierarchy of a document. For example, a document which describes a device such as a scanner might have major sections such as an overview, a description of the hardware of the scanner, a description of the software utilized with the scanner, and accessories or options which can be used with the scanner. The overview section might include the name or description of the device, a figure showing the scanner, and a list of features of the scanner.
After the Document Type Definition has been established, the dictionary and rule data bases which are to be used with the translation process must be defined. One manner of creating the dictionaries and rules used for the translation of the different sections of the document is to input documents having a known translation. In this manner, the system can automatically build the dictionary and translation rules so that the translation of the document results in the proper known translation. In addition to or alternative to the translation of known documents, the user can manually translate a document so that dictionaries and rules are built based on the input of a user. Further, a combination of automatic and manual translation can be performed which allows the user to correct the translations performed automatically so that the resulting translation is more accurate. As a further alternative, the rules and dictionaries are manually created.
The preferred operation of the invention utilizes a separate data structure for each of the different sections of the document. The data structure stores information indicating information such as names and locations of further dictionaries and rules which may be utilized with the corresponding section or other sections of the document. The use of the data structure to keep track of the dictionaries and rules permits the different dictionaries and rules to be arranged in a hierarchy similar to a hierarchy of the structure of the document.
When the translation system is sufficiently developed and trained, the translation process can become automatic, using the specific dictionary information and translation rules which correspond to a specific section of the document. By using the specific dictionary information and rules for the different sections of the document, the translation becomes more accurate and becomes closer to a one-to-one correspondence between the source document and translated document without the burdens or difficulties which arise from using translation systems which are limited to the use of canned phrases.
REFERENCES:
patent: 4553261 (1985-11-01), Froessl
patent: 4584667 (1986-04-01), Hashimoto et al.
patent: 4890230 (1989-12-01), Tanoshima et al.
patent: 4916614 (1990-04-01), Kaji et al.
patent: 4980829 (1990-12-01), Okajima et al.
patent: 5010486 (1991-04-01), Suzuki et al.
patent: 5029114 (1991-07-01), Makiguchi
patent: 5062047 (1991-10-01), Tanaka et al.
patent: 5091876 (1992-02-01), Kumano et al.
patent: 5109439 (1992-04-01), Froessl
patent: 5132901 (1992-07-01), Yokogawa
patent: 5140522 (1992-08-01), Ito et al.
patent: 5329446 (1994-07-01), Kugimiya et al.
patent: 5418717 (1995-05-01), Su et al.
patent: 5424947 (1995-06-01), Nagao et al.
patent: 5466072 (1995-11-01), McCready
patent: 5477450 (1995-12-01), Takeda et al.
patent: 5497319 (1996-03-01), Chong et al.
patent: 5528491 (1996-06-01), Kuno et al.
patent: 5535120 (1996-07-01), Chong et al.
patent: 5548508 (1996-08-01), Nagami
patent: 5579224 (1996-11-01), Hirakawa et al.
patent: 5587902 (1996-12-01), Kugimiya
patent: 5659765 (1997-08-01), Nii
patent: 5687383 (1997-11-01), Nakayama et al.
patent: 5848386 (1998-12-01), Motoyama
patent: 5983169 (1998-12-01), Kozna
patent: 0 568 319 A2 (1993-11-01), None
Baker et al., “Coping with Ambiguity in Knowledge-based Natural Lnaguage Analysis”, Japan 1994, 5 pp.
Edouard Patrick N.
Isen Forester W.
Oblon & Spivak, McClelland, Maier & Neustadt P.C.
Ricoh & Company, Ltd.
LandOfFree
Method and system for translating documents using different... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and system for translating documents using different..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and system for translating documents using different... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2524857