Data processing: speech signal processing – linguistics – language – Speech signal processing – Application
Reexamination Certificate
2000-05-09
2001-04-17
Hudspeth, David (Department: 2641)
Data processing: speech signal processing, linguistics, language
Speech signal processing
Application
C704S260000, C704S235000, C704S251000, C704S256000, C704S010000, C704S008000, C704S003000
Reexamination Certificate
active
06219646
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates to translating from one language to another. More particularly, the present invention relates to providing translation between languages based, at least in part, on a user selecting a particular topic that the translation focuses on.
Translation between languages is well known. In the most basic sense, translation is performed manually by individuals that are fluent in both the source and target languages. Human translators have the ability to translate written or spoken text, with a very high degree of accuracy, almost instantaneously. Additionally, human translation is often more accurate because the translator is often very knowledgeable regarding the topic or subject matter that the communication is based on.
Even though human translations are often very accurate, they are also very expensive—requiring individuals having very specialized skills. Besides the basic expense, which can be prohibitive, many instances requiring translation require people having additional knowledge that further increases costs. For example, if two biotechnology scientists who spoke different languages needed to communicate, the translator would need to have, in addition to being fluent in both languages, knowledge in biotechnology so that many “terms of art” would be translated with their proper meaning.
An additional problem with human translation is the small availability of qualified individuals to perform such tasks. There have been many attempts to address the problems of labor-intensive human translation practices. Often, these attempts have involved the use of electronic devices to translate written text from one language to another.
For example, Masuzawa et al. U.S. Pat. No. 4,393,460 discloses an electronic translator that uses a voice responsive circuit to input words in one language, processing circuitry that translates the words to a second language, and a voice synthesis circuit for “speaking” the translated words in the second language. The translation is based on the operation of three “analyzers” that analyze the characters of a word, the syntax and the frequency of words. One deficiency of the system described by Masuzawa is that it relies on a single set of rules and “difficult to understand” sentences for each language. This invariably leads to translation errors, especially when specialized topics are the subject of the translation.
Another translation system is described in Okamoto et al. U.S. Pat. No. 4,805,732. Okamoto discloses a machine having an input section, a dictionary section that stores linguistical information, a processing section and an output section. The output section includes a display and the capability to permit the user to edit either the input text or the translated text. Okamoto, however, suffers from at least the same deficiency as Masuzawa, namely that translation is based on a single set of rules for a given language.
Fukumochi et al. U.S. Pat. No. 5,299,124 describes a translation machine that is specifically directed toward syntactic sentence analysis. The input sentence is divided into words by a dictionary/morpheme analyzer. The translated word is matched up with grammatical information for each word to derive tense, person and quantity information (i.e., singular or plural). A syntactic analyzer forms a structure-analyzing tree based on the dictionary translation and grammatical rules for the sentence. If the tree fails, the sentence is broken apart into smaller pieces and reprocessed until a successful translation occurs. Fukumachi, however, also applies a single set of rules for each language.
Frantz et al. U.S. Pat. No. 4,507,750 attempts to address some of the deficiencies described above. Frantz, noting that a word for word translation is insufficient for accurate translation, describes a system that analyzes the context within a given sentence to overcome various problems, such as dealing with homonyms, incurred by other translation devices. While Frantz does deal with simple problems, such as differentiating between to, too and two, Frantz still relies on a single set of rules per language.
For at least the above reasons, it is an object of the present invention to provide a translation system that translates between languages depending on the topic of the information being translated.
It is also an object of the present invention to provide a translation system that relies on multiple rule bases for a given language to increase translation accuracy.
It is a still further object of the present invention to provide a translation system that utilizes multiple dictionaries to translate between a first and a second language.
SUMMARY OF THE INVENTION
The above and other objects of the present invention are accomplished by the systems and methods described herein in which translation between two languages is based, at least in part, on the selection of a topic that the communication is based upon. The translation system and method of the present invention includes what may be referred to as a “three-dimensional” data base that is built for each pair of languages. The three dimensions include the source language, the target language and the topic (or subject matter on which the communication is based). Each cell in the data base includes information for each topic relating to, for example, frequency of association, synonyms and topic-related dictionary definition.
The principles of the present invention may be applied to a variety of systems and methods to automatically perform highly accurate translations. In one embodiment, topical dictionaries are constructed by scanning (or other conventional means) various documents of material. For example, to establish a data base level (i.e., one level of the three-dimensional data base) related to microbiology, various articles from microbiology trade publications would be scanned. The scanned material may be passed through a parser that assigns the words to word classes. The input material is then applied to a conventional dual-language dictionary that produces target language translations.
At the same time as the source language material is being scanned, a similar process is carried out on the target language using various documents in the target language (that are related to the same topic). The end result is two files of words in one language with associated translations. The two files are then processed via some form of pattern recognition routines to compute the forward and backward frequency of association between the two files. This process produces one level of the three-dimensional topic data base (i.e., for the selected topic).
Once the three-dimensional topic data is established for a given topic, the system may be used in many ways. One embodiment is a system in which two individuals that speak different languages can talk to each other over the telephone with the system providing near real-time translation (each caller would pause slightly after speaking). Voice recognition circuitry converts the analog signals received from a microphone (or a digitized version of those signals) into signals for processing. The processing circuitry converts the signals into words that are then applied against the topical data base, and a conventional dual language data base, to translate the text. The translated text is synthesized and transmitted to the other caller.
REFERENCES:
patent: 4706212 (1987-11-01), Toma
patent: 5220503 (1993-06-01), Suzuki et al.
patent: 5268839 (1993-12-01), Kaji
patent: 5442780 (1995-08-01), Takanashi et al.
patent: 5535120 (1996-07-01), Chong et al.
patent: 5695980 (1997-12-01), Brew
patent: 5848389 (1998-12-01), Asano et al.
patent: 6085162 (2000-07-01), Cherny
Chawan Vijay B
Fish & Neave
Gedanken Corp.
Hudspeth David
Morris Robert W.
LandOfFree
Methods and apparatus for translating between languages does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Methods and apparatus for translating between languages, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Methods and apparatus for translating between languages will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2452994