GB2199432A - Translation machine - Google Patents

Abstract

Disclosed is a translation machine system comprising a circuit to sequentially output next candidate translations in the case that plural candidate translation are obtained from a sentence, and a circuit to select a previous candidate translation. The system provides the facility for the user to revert from a currently-displaced candidate to one offered previously without the need first to return to the beginning of the set of candidate translations. <IMAGE>

Description

TRANSLATION MACHINE SYSTEM BACKGROUND OF THE INVENTION This invention relates to a translation machine system to translate sentences one by one, more particularly to a system capable of, after inquiring several next translation candidates with regard to an original sentence, selecting a previous candidacy translation.

Generally speaking, in a translation machine system, it cannot be avoided that plural translated sentences are generated because of the ambiguity present in natural languages.

Conventionally, it was impossible to return to a previous candidate after displaying several candidacy translations so that an#operator should start again from the beginning.

The key operation should be, hence, carried out several times to waste time, which was a great burden for the user.

SUMMARY OF THE INVENTION It is, hence, an object of this invention to present a translation machine system to sequentially output next candidacy translations when plural output sentences can be obtained by translation, wherein it is so designed that a previous candidacy translation can be selected, so allowing the user to quickly and easily select a right translation result from the plural candicacy translations.

Briefly described. in accordance with the present invention, is a translation machine system having means to sequentially output next candidacy translations when a sentence is translated and plural translated sentences are obtained, and having means to select a previous candidacy translation.

BRIEF DESCRIPTION OF THE DRAWINGS The present invention will become more fully understood from the detailed description given hereinbelow and the accompanying drawings which are given by way of illustration only, and thus are not limitative of the present invention and wherein: Fig. 1 is a block diagram of the whole system of a translation machine system according to an embodiment of this invention; Fig. 2 is an explanatory drawing of translation levels; Figs. 3 and 4 are explanatory drawings of translation module; Fig. 5 is a drawing showing an original sentence; Fig. 6 is a drawing showing the results of dictionary look-up; Fig. 7 is a drawing showing the results of syntax analysis; Fig. 8 is a drawing showing the converted results; Fig. 9 is a drawing showing an output sentence; Fig. 10 is an explanatory drawing of result buffer; and Fig. 11 is a processing flowchart.

DESCRIPTION OF THE PREFERRED EMBODIMENT Referring now to the drawings, the constitution of this invention is explained hereinafter.

Fig. 1 is a block diagram of the whole system of an embodiment of the translation machine system according to this invention.

In the drawing, numeral 1 is a central processing unit (CPU), 2 is a memory, 3 is a CRT display and 4 is a keyboard. Numeral 5 is a translation module, which is described in details below, and 6 is a memory to store a dictionary for translation, grammatical rules and structure conversion rules.

Said translation module 5 presents a target language in return when a source language is given.

A source language is input from said keyboard 4, and the input source language is sent to said translation module 5 depending on the control of said CPU 1.

The target language output by the translation module 5 is displayed on said CRT 3.

On said keyboard 4, a next-candidacy selection key 4a and a previous-candidacy selection key 4b are installed.

Said next-candidacy selection key 4a is provided to obtain the next candidacy translation and said previous-candidacy selection key 4b is provided to obtain the previous-candidacy translation.

The system-made translations processed by translation module 5 have generally such levels as shown in Fig. 2.

As shown in the drawing, by inputting a source language located at the upper left, an analysis is proceeded by steps of (1) dictionary look-up, (2) morpheme analysis and (3) syntax analysis.

System-made translation is classified roughly into two depending on said analysis level.

One is called transfer method, which executes analysis up to any one of the levels (2) morpheme analysis, (3) syntax analysis, (4) meaning analysis or (5) context analysis, thereby obtaining the internal constitution of the source language. Sequentially, after converting to the constitution at the same level as said internal constitution, a target language is generated.

The other one is called pivot method in which a source language is analyzed up to a concept (interim language in the diagram), which does not depend on the languages, and thus, a target language is generated therefrom.

The analyses are explained hereafter.

(1) Dictionary look-up and (2) morpheme analysis The system refers to the dictionary in said memory 6 to obtain grammatical information on words such as parts of speech and analyzes their tense, person, and number.

(3) Syntax analysis The system determines the constitution of the sentence such as the relation between words.

(4) Meaning analysis The system identifies correct and incorrect meanings based on the plural syntax analyses.

(5) Context analysis The system determines the subject to remove abbreviation and ambiguity.

The translation module 5 of this embodiment carries out the analysis up to the level of at least (3) syntax analysis.

Accordingly, the translation module 5 of this embodiment comprises, as shown in Fig. 3, (1) dictionary look-up device and morpheme analyzer, (2) system analyzer, (3) converter, and (4) generator.

The structure of said translation module 5 is shown in Fig. 4.

The translation module 5 is equipped with, as shown in the drawing, buffers A, B, C, D and E, central processing unit CPU 5a, and ROM (read only memory) 5b which stores the programs.

Here, an example of translation from English to Japanese is described on an English sentence, "This is a pen".

At first, the original sentence is read in buffer A of said translation module. In buffer A, as shown in Fig. 5, the original sentence is stored word by word.

Sequentially, dictionary look-up of said translation module 5 is actuated, and hence, required information on words is looked up, which is input in buffer B of said translation module 5.

Information of part of speech, which is one of the items of the above information is shown in Fig. 6.

As shown in the drawing, "this" has several parts of speech, but one meaning is selected depending on the syntax analyzer of said translation module 5, and its relation with other words is stored in buffer C of said translation module. The relation with other words is shown in Fig. 7.

At this time, by obtaining the following grammatical rules from said memory 6, analysis is executed.

(1) Sentence ' Subject and predicate (2) Subject + Noun phrase (3) Predicate + Verb and noun phrase (4) Noun phrase + Pronoun (5) Noun phrase #-#rticle and noun The above rules mean, by taking (1) as an example, "A sentence comprises subject and predicate", which is the same in items 2 to 5.

At the converter of said translation module 5, the analysis is carried out, in the same way as syntax analysis, depending on the structure conversion rules, and the analyzed results are as shown in Fig. 8, which are input in buffer D of said translation module 5.

In the next, by the generator of said translation module 5, appropriate particles and auxiliary verbs are attached to be in a form of Japanese as shown in Fig. 9, which is stored in buffer E of said translation module 5. The data of buffer E becomes the output of said translation module 5.

The translation product delivered from said translation module 5 is stored in the result buffer shown in Fig. 10. The result buffer has a capacity of storing plural translation products, among which a translation product stored in a location indicated by pointer P is read out.

Fig. 11 shows the processing flowchart of this embodiment.

At first, the pointer P is set at the beginning of the result buffer (step Sl).

The system translates a sentence (step S2) , stores the obtained translation sentence in said result buffer (step S3), and waits for the next key input (step S#).

In the state of waiting o key input, i- the next-candidacy selection Rey ta is pressed, the value of said pointer P is increased by one (step S3).

Sequentially, whether any translated sentence exists in the location indicated by the pointer or not is checked (stage Sg). Then, when it is judged that no translation data is in saic indicated location, the operation returns to said step S2, and translation processing is executed again to transfer to said step S3. The translation product is, thus, stored in said result buffer. In this way a translation product is stored in said result buffer at each time when said r.e: candicacy selection key 4a is pressed.

In the stage of waiting for key input at said step S#, t if previous-candidacy selection key 4b is pressed, the value of said pointer P is reduced by one (step Ss) Next, it is checked whether the value of said pointer P is less than zero or not (step S6)- And if it is judged that the value of said pointer P is less than zero, the operation returns to the state of waiting for key input at said step Sq. Accordingly, at this time, the content of said result buffer is not displayed.

On the contrary, if the value of said pointer P is zero or more, the data of result buffer indicated by the pointer P is displayed (step S7) and after that, the system returns to said step S4 to wait for key input.

When said previous-candidacy selection key 4b is operated and sequentially next-candidacy selection key 4a is pressed, the operation transfers to step Ss and step SJ, where said pointer P specifies a translation product before the one entered in said result buffer, so that the translation sentence is displayed (step S7).

Afterward, the above processing is repeated each time when said next-candidacy selection key 4a or previous-candidacy selection key 4b is operated.

As described above, according to the translation machine system of this invention, in a system having means to sequentially output next candidacy translations when a sentence is translated and plural translation products are obtained, because of having means to select previous candidacy translation, it is possible to output the desired translation result without restarting its translation from the beginning when the operator wants to output a previous candidacy translation again at the stage of outputting next candidacy translation, so that it becomes possible to obtain a correct translation result easily and quickly.

While only certain embodiments of the present invention have been described, it will be apparent to those skilled in the art that various changes and modifications may be made therein without departing from the spirit and scope of the present invention as claimed.

Claims (1)

1. CLAIMS:

   1. A translation machine system comprising: means to sequentially output next candidacy translations in the case that plural translation products are obtained from a sentence; and means to select a previous candidacy translation.

   2. A translation machine system of claim 1, wherein said selection means comprises buffer means for storing translated sentences and pointer selection means for selecting the position of a pointer for said buffer means to specify one of the translated sentences.

   3. A translation system of claim 2, wherein said pointer selection means is operated by next candidacy selection key means and previous-candidacy selection key means.

   4. A translation machine system which is operable to perform a translation operation in which for some input word groups or passages the user is offered a succession of possible candidate translations from which to choose, the system having selection means which permits the user to revert from a currently offered candidate to a previous one without the need first to return to the beginning of the succession.

   5. A translation machine system according to claim 4 wherein said selection means is selectively operable to permit the user to proceed forward or backward, one-by-one, through the succession.

   7. A translation machine system substantially as hereinbefore described with reference to the accompanying drawings.