JP2870279B2 - Translation selection device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a translation word selecting apparatus in a machine translation system for analyzing an input language expression using a bilingual dictionary and outputting a corresponding target language expression.

[0002]

2. Description of the Related Art Translation is to analyze an input expression of a source language A,
The purpose is to output an expression of the target language B corresponding to the input expression. Generally, the correspondence between the word expression a in the language A and the word expression b in the language B, that is, the translation is not one-to-one.
Therefore, at the time of translation, it is necessary to select the most appropriate one from the translated words b1 to bn in the language B corresponding to the word expression a in the language A according to the appearance environment of the word expression a.

[0003] In the current machine translation system, basically,
A translation word selection method is adopted in which "a constraint on a language structure for a certain expression is described in advance in a bilingual dictionary / grammar, and a translation word is selected by applying these to an input expression." FIG. 4 shows the configuration of a bilingual dictionary for a certain headword a used in this translation word selection method. Language A
Is input, and when determining a translation for the word expression a in the partial expression A ′, this dictionary is selected using the headword a as a key, and from the corresponding translation word groups b1 to bn, translation bi the attendant constraints Ci1～cim _i partial representation a 'satisfies is selected as an appropriate translation.

For example, in selecting a translation for a declinable word,
This constraint is mainly described as a semantic constraint on the essential case element. FIG. 5 shows an example of an English verb “tak”.
e is a part of an English-Japanese bilingual dictionary corresponding to "e". "Take this medicine. When the sentence "is input, the subject is" person "(the sentence is a command sentence and the subject is equivalent to" you "). "
"Drink" is selected, and the whole translation is decided to "Take this medicine."

On the other hand, recently, 1) Sumida / Iida / example-driven machine translation, IPSJ Natural Language Processing Study Group, 82-5, 2) Sato / MBT1: Examples Selection of translations based on Wikipedia, Japan Society for Artificial Intelligence, Vol. 6, No. 4 3) Sato / MBT2: Combined Use of Multiple Translation Examples in Example-Based Translation, Journal of the Japanese Society for Artificial Intelligence, Vol. 6, N
o6 (both in 1991) have been proposed. Case-based translation means that the translation cases actually translated by humans are stored in advance, and the most similar case to the input is searched from the bilingual case database using a thesaurus, etc. It is a technique of translating to. FIG. 6 shows the configuration. In the bilingual case database 24, cases translated by humans are stored in advance. Further, a distance to be calculated using the thesaurus 25 between the input and the case in order to determine the case most similar to the input is defined in advance. When the case-based translation means 21 receives the source language expression as the input 11,
The similar case search means 22 searches the bilingual case database 24 for a case where the distance from the partial expression is the shortest with respect to the partial expression having the source language expression, and determines the similar case. Is modified by the case input adapting means 23 so as to conform to the input, and is used as a partial expression of the translation result.
Let it be 2.

The translation method based on this case differs from the translation word selection based on the constraint description in that a bilingual case similar to the input is searched for from the bilingual case database, instead of completely matching the input. It is characterized by doing.

[0007]

However, the translation word selection method of the above-mentioned format includes "a translation constraint group b1 to bn corresponding to the word expression" a "in the input partial expression A '. If there are multiple translations that satisfy expression A ', it cannot be determined which is the most appropriate. "
"Translation word group b corresponding to word expression a in input partial expression A '
If none of the sub-expressions A ′ satisfies the constraints associated with 1 to bn, the analysis will fail at that point in time ”. Ideally, the word expression a
If the constraints attached to the translation word groups b1 to bn corresponding to are all exclusive and the union covers all the appearance environments of the word expression a, any input including the word expression a There is always only one translation, but in practice it is not possible to describe conditions in this way.

Further, in order to realize the above-described case-based translation method, a large-scale bilingual case database is required. And it is difficult to collect a sufficient amount of bilingual texts for this purpose, and even if it can be collected, there is no means to automatically analyze it and associate it structurally. And costly.
Further, no application is conceived in the form of utilizing knowledge of conventional systems.

As described above, the translation word selection method based on the constraint description includes "a translation word whose partial expression A 'satisfies the accompanying constraint condition in the translation word groups b1 to bn for the word expression a in the input partial expression A'. Cannot be determined which one is the most appropriate. "" The sub-expression A 'defines the constraints attached to the translation word groups b1 to bn corresponding to the word expression "a" in the input sub-expression A'. If none of these are satisfied, the analysis will fail at that point. " On the other hand, the case-based translation method has the drawback that it is difficult to create a bilingual case database, and no practical application utilizing conventional technology has been proposed.

[0010] An object of the present invention is to add an example of a source language expression in addition to a constraint condition for each translation word in a bilingual dictionary, thereby providing a case-based compensation function when a translation word cannot be selected only by constraints. It is an object of the present invention to provide a translation word selection device that utilizes the advantages of both a translation word selection method based on constraint description and a case-based translation method of selecting the most appropriate translation word using a translation function.

[0011]

According to a first aspect of the present invention, there is provided a translated word selection apparatus comprising: a translated word dictionary in which constraints for selecting the translated word and an example of a source language expression are described; and a set of expressions. A similar expression search means for searching for an expression most similar to a specific expression, and a constraint described in a translation for a headword that matches the word expression in the bilingual dictionary when determining a translation for a word expression in the input expression. Applying the condition to the input expression, selecting the associated constraint as the most appropriate translation that satisfies the input expression, and, when there are multiple translations satisfying the associated constraint, the input expression satisfies: From the examples of source language expressions described in the translations that the input expression satisfies the accompanying constraints,
A translation term selecting means for searching for an example most similar to an appropriate partial expression including the word expression in the input expression by the similar expression search means, and selecting a translation word describing the example example as a most appropriate translation word; A source language expression is linguistically analyzed and decomposed into sub-expressions to generate a set of sub-expressions, which is sent to the translated word selecting means, and a target linguistic expression is generated and output from the set of translated words received from the translated word selecting means. It is characterized by having machine translation means.

According to a second aspect of the present invention, there is provided a translation word selecting apparatus, comprising: a bilingual dictionary describing, for each translation word, a constraint condition for selecting the translation word and an example of a source language expression; A similar expression search means for searching for the expressed expression;
When determining a translation for a certain word expression in the input expression, the constraints described in the translation for the headword that matches the word expression in the bilingual dictionary are applied to the input expression, and the accompanying constraints are applied to the input expression. Is selected as the most appropriate translation, and if the input expression does not satisfy any of the constraint conditions, then the source language expression described in the translation for the headword is used as the most appropriate translation. Searching the example most similar to an appropriate partial expression including the word expression by the similar expression search means,
A translation word selecting means for selecting a translation word in which the example is described as the most appropriate translation word, and a linguistic analysis of the input source language expression to decompose it into partial expressions to generate a set of partial expressions, which is sent to the translation word selecting means. Machine translation means for generating and outputting a target language expression from a set of translation words received from the translation word selection means.

[0013]

In the translation word selecting device of the present invention, the machine translation means receives the input source language expression, linguistically analyzes the input expression, decomposes the expression into subexpressions, generates a set of subexpressions, and sends it to the translation word selection means. The translated word selecting means refers to the bilingual dictionary and, if necessary, determines a translated word for the word expression in the received partial expression using the similar expression searching means and sends it to the machine translating means. The machine translation means generates and outputs a target language expression from the set of received translated words.

In the bilingual dictionary, examples are added in addition to the constraint conditions for each translated word when the dictionary is created.

The similar expression retrieval means retrieves an example most similar to a specific subexpression from a set of examples. The method of calculating the distance between the input expression and the example used to determine the most similar example includes a method using a thesaurus similar to the translation method based on the conventional case, and the meaning of each expression is expressed by a pair of a feature and a feature value. And a method of counting the number of features having matching values.

[0016]

Next, an embodiment of the present invention will be described with reference to FIGS.

FIG. 1 is a block diagram showing one embodiment of the first and second embodiments of the present invention. FIG. 2 is a diagram showing a configuration of a bilingual dictionary for a certain headword a used in a translation word selecting apparatus of the present invention.
FIG. 3 is a diagram showing an example of an English-Japanese bilingual dictionary for the English verb "take".

In FIG. 1, a machine translation unit 1 receives a source language expression as an input 11, linguistically analyzes the input expression, decomposes the expression into partial expressions, generates a set of partial expressions, and sends it to a translated word selection unit 2.

The translation word selecting means 2 refers to the bilingual dictionary 3 and
Determine the translation for the word expression in the received sub-expression.

In the bilingual dictionary 3, as shown in FIG.
It has been written in the im _i & Examples ei1~eil _i. The number of examples may be at least one. Figure 3 shows an example of a description example
Shown in

When the partial expression A 'including the word expression a is input, the translation word selecting means 2 refers to the bilingual dictionary shown in FIG. In the same manner as in the description-based word selection method, the constraints attached to the word groups b1 to bn corresponding to the word expression a are applied to the partial expression A ′. The constraint that subexpression A 'satisfies is c
In the case of I1～cim _i only selects the translation bi as appropriate translation.

In the first invention, the word expression a is included in the constraints attached to the translation word groups b1 to bn corresponding to the word expression a.
When there are a plurality of constraints satisfying the partial expression A ′ including the partial expression A ′, the translated word selecting unit 2 uses the similar expression searching unit 4 to select a partial expression from the examples added to the translated word group accompanied by the restricted condition. The example that is most similar to the expression A ′ is searched, and the translated word bj to which the example ejk is added is selected as the most appropriate translated word.

In the second invention, when none of the subexpressions A ′ including the word expression a satisfies the constraint conditions associated with the translation word groups b1 to bn corresponding to the word expression a, the translation word selecting means 2 performs similarity. Using the expression search means 4, from the examples added to the translation word groups b1 to bn corresponding to the word expression a,
Search for the example that is most similar to the sub-expression A '
The translated word bj to which jk is added is selected as the most appropriate translated word.

The determined translated word is sent from the translated word selecting means 2 to the machine translating means 1, and the machine translating means 1 generates a target language expression from the set of received translated words and outputs it as an output 12.

The first invention and the second invention are applied in different cases, and they can be used in any combination. The example added to the dictionary can be used in common for both inventions.

[0026]

As described above, according to the present invention, it is possible to realize a translation word selection device that utilizes both the advantages of the conventional translation word selection method based on constraint description and the case-based translation method. That is, by adding an example to the bilingual dictionary, it is guaranteed that one of the translated words is always selected, and the disadvantage of the translated word selection method using the constraint description can be covered. It is also possible to incorporate the advantage of a translation method based on the case that "the translation result can be given reliability in the form of a distance from the used case".

On the other hand, the translation method based on the case has a drawback that "it is difficult to create a bilingual case database". However, in the translation word selecting apparatus of the present invention, it is not necessary to prepare a bilingual case. Since it is only necessary to add at least one example of the language, it is not difficult to create a dictionary in this form. In particular, when creating a bilingual dictionary with reference to existing English-Japanese / Japanese-English dictionaries, the examples in the dictionary may be used as they are. Also, even if the user is required to add at least one example of the source language when registering a new word or a new translation for an existing word in the dictionary, since the user is considered to register based on a specific example, The increase in the burden on the user is not large. In addition, this method has an effect that the dictionary is easy for humans to see because the example is described in the bilingual dictionary.

[Brief description of the drawings]

FIG. 1 is a block diagram showing one embodiment of the present invention.

FIG. 2 is a diagram showing a configuration of a bilingual dictionary for a certain headword a used in the present invention.

FIG. 3 is a diagram showing an example of a configuration of a bilingual dictionary used in the present invention in the form of an English-Japanese bilingual dictionary for an English verb “take”.

FIG. 4 is a diagram showing a configuration of a bilingual dictionary for a certain headword a used in a translation word selection method based on a constraint description according to the related art.

FIG. 5 is a diagram showing an example of the configuration of a bilingual dictionary used in a translation word selection method using a constraint description according to the related art in the form of an English-Japanese bilingual dictionary for an English verb “take”.

FIG. 6 is a block diagram showing an embodiment of a translation method based on the case of the conventional technique.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Machine translation means 2 Translation word selection means 3 Bilingual dictionary 4 Similar expression search means 11 Input 12 Output 21 Case-based translation means 22 Similar case search means 23 Case input adaptation means 24 Bilingual case database 25 Thesaurus

────────────────────────────────────────────────── ─── Continuation of the front page (56) References JP-A-62-219165 (JP, A) JP-A-3-276367 (JP, A) JP-A-1-135567 (JP, U) Naoto Nakamura, “Example Retrieval and Translation Support System ”, Proc. Of the IPSJ 38th National Convention, (I), p. 357-P. 358 (1989) Shinichi Doi and Kazushi Muraki, "Strengthening Translation / Meaning Selection Function by Adding Cases to Dictionary," Proc. Of the 44th National Convention of IPSJ (III), p. 111-p. 3.112 (1992.2) (58) Field surveyed (Int. Cl. ⁶ , DB name) G06F 17/20-17/28 JICST file (JOIS)

Claims (2)

(57) [Claims] 1. A bilingual dictionary describing, for each translated word, a constraint condition for selecting the translated word and an example of a source language expression, and a similar expression search for searching an expression most similar to a specific expression from a set of expressions. Means for determining a translation for a certain word expression in the input expression, applying a constraint described in a translation for a headword matching the word expression in the bilingual dictionary to the input expression, Selects the translation that satisfies the input expression as the most appropriate translation, and if there are multiple translations that satisfies the input expression, the associated constraint is applied to the translation that satisfies the input expression. From the described example of the source language expression, an example that is most similar to an appropriate partial expression including the word expression in the input expression is searched by the similar expression search means, A translation word selecting means for selecting the translation word described as the most appropriate translation word, and generating a set of sub-expressions by linguistically analyzing the input source language expression and decomposing it into sub-expressions, sending the set to the translation word selecting means, A translation word selecting device, comprising: a machine translation means for generating and outputting a target language expression from a set of translation words received from the translation word selection means. 2. A bilingual dictionary for describing, for each translated word, a constraint condition for selecting the translated word and an example of a source language expression, and a similar expression search for searching an expression most similar to a specific expression from a set of expressions. Means for determining a translation for a certain word expression in the input expression, applying a constraint described in a translation for a headword matching the word expression in the bilingual dictionary to the input expression, The translation that satisfies the input expression is selected as the most appropriate translation, and if the input expression does not satisfy any of the constraints, the translation is performed based on the example of the source language expression described in the translation for the headword. Searching for an example most similar to an appropriate partial expression including the word expression in the expression by the similar expression search means,
A translation word selecting means for selecting a translation word in which the example is described as the most appropriate translation word; and a linguistic analysis of the input source language expression to decompose it into sub-expressions to generate a set of sub-expressions, which are sent to the translation word selection means. And a machine translation means for generating and outputting a target language expression from a set of translation words received from the translation word selection means.