JPH0424767A - Machine translation system - Google Patents

Abstract

PURPOSE:To easily realize appropriate translation even for a beginner by successively displaying a modification relation with the possibility of interpretation as a structure sentence when there is an ambiguity in an original so as to make the user select. CONSTITUTION:A sentence analysis part 2 executes the sentence analysis of the original and generates a syntax tree with the possibility of interpretation, and a structure sentence generation part 3 generates a structure sentence from the generated syntax tree, displays this structured sentence, confirms the ones which are approved by the user as an appropriate sentence, and successively displays other structured sentences if it is rejected. Also, the modification relation is displayed in parentheses as the structure sentence, the contents in an equal modification relation are itemized and displayed, etc. Then, when there is an ambiguity in the original, the modification relation with the possibility of interpretation is successively displayed as the structured sentence to make the user select. Thus, even a beginner can easily and efficiently generate the appropriate translation.

Description

[Detailed Description of the Invention] [Summary] Regarding a machine translation system that displays structured sentences that may be interpreted in the original text, if there is ambiguity in the original text, the system displays a dependency relationship that may be interpreted as a structured sentence. The purpose of this module is to analyze the original text and generate syntactic trees with possible interpretations, with the aim of allowing the user to make a selection by sequentially displaying the following translations, allowing even beginners to easily and efficiently generate appropriate translations. and a structured sentence generation unit that generates a structured sentence from the syntax tree generated by the sentence analysis unit, and the structured sentence generation unit generates a syntax tree that has possible interpretations generated from the original sentence. The structured sentences are sequentially displayed, and the displayed structured sentences that are approved by the user are determined to be appropriate sentences.

[Industrial application field]

The present invention relates to a machine translation system that displays structured sentences that may be interpreted as possible in the original text. However, there is a need for a system that supports this.

[Prior art and problems to be solved by the invention] Conventionally, when an appropriate translation result cannot be obtained when a machine translation system has multiple dependency relationships in the original text, the user has to transfer the original text to the system. I tried to rewrite it in a way that could be interpreted appropriately. This rewriting work requires a great deal of effort and effort by a user who is familiar with system processing, and there is a problem in that machine translation cannot be performed quickly and efficiently.

The present invention enables even beginners to easily and efficiently generate appropriate translations by sequentially displaying dependency relationships that may be interpreted as structured sentences when there is ambiguity in the original sentence and allowing the user to select one. It is an object.

[Means to solve problems]

Means for solving the problem will be explained with reference to No. 17.

In FIG. 1, a sentence analysis unit 2 analyzes an original sentence to generate a syntax tree with possible interpretations.

Structured sentence generation unit No. 3 generates structured sentences from the syntax tree generated by the sentence analysis unit 2.

[Function: 1] As shown in FIG. 1, in the present invention, the sentence analysis unit 2 performs sentence analysis of the original sentence to generate a syntax tree with possible interpretations, and the structured sentence generation unit 3 A structured sentence is generated from the syntax tree generated in step 2, the structured sentence is displayed, and the one that is approved by the user is determined as an appropriate sentence, and when it is rejected, another structured sentence is generated. I am trying to display them sequentially. In addition, as a structured sentence, dependency relationships are displayed in parentheses, and contents in a coordinate connection relationship are displayed as bullet points.

Therefore, when there is ambiguity in the original text, even beginners can easily and efficiently generate appropriate translations by sequentially displaying dependency relationships that may be interpreted as structured sentences and letting the user select one. becomes possible.

〔Example〕

Next, the configuration and operation of the embodiment of the present invention will be explained in detail using FIGS. 1 and 2.

In FIG. 1(a), the writer's blade section 1 is used to input an original text to be subjected to machine translation or the like.

The original text is, for example, the sentence (1) shown on the right.

The sentence analysis unit 2 performs sentence analysis of the original sentence and generates a syntax tree with possible interpretations. For example, for the original text in (1), a syntax tree with possible interpretations as shown in FIG. 1 (b) and (c) is generated. Here, the syntax tree in FIG. 1(b) is for the case where "attached" relates to both "place" and "transformation." In the syntax tree shown in FIG. 1 (c), "attached" relates to "location."

The structured sentence generation unit 3 generates a structured sentence from the syntax tree generated by the sentence analysis unit 2.

Here, for example, from the syntax tree in Figure 1 (B) to the syntax tree in Figure 1 (D),
As shown in display example 12), a structured sentence is generated in which dependency relationships are enclosed in parentheses 1:.

The display unit 4 displays the structured sentence generated in the FIIt structured sentence generation process 3 on the display as shown in the display example shown in FIG. 1(d). This displayed structured sentence,
For example, in the display example in Figure 1 (d), for example, the structured sentence (2) (
When the user approves the structured sentence generated from the syntax tree in FIG. 1(b), it is determined that the structured sentence has the correct dependency relationship of the original sentence. On the other hand, if the user disapproves, another structured sentence is selected 7, and for example, the structured sentence (2) in the display example in Figure 1 (d) generated from the syntax tree in Figure 2 (c) is displayed. do.

The target language generation unit 6 adopts the determined syntax tree of the structural sentence and uses this to translate it into the target language.

For example, if the user accepts (2) in the display example shown in FIG. 1(d), the machine translated result is displayed as shown in (3). Furthermore, if the user accepts (2)'' in the display example shown in FIG. 1(d), the machine translated result is displayed as shown in (3)''.

As described above, the original sentence is parsed to generate syntactic trees with possible interpretations, and structured sentences generated from these syntactic trees, such as structured sentences with dependency relationships enclosed in parentheses, are shown in Figure 1 ( d) Display sequentially as shown in (2) or (2)゛,
What the user accepts is determined to be a structured sentence (syntax tree) with correct dependency relationships, and by machine translation based on this syntax tree, the system displays it as a structured sentence with possible interpretations. Select one of the
It becomes possible to easily perform machine translation into a target language that has appropriate dependency relationships.

Next, the case of coordinate connection will be explained using FIG. In the case of this coordinate connection, the processing by the structured sentence generation unit 3 in FIG. 1(a) is performed as shown in 0 to [phase] below.

In FIG. 2 (a), ■ finds coordinately connected elements. As shown in the original sentence (1) of the display example in FIG. 2 (b), this finds a sentence that has the coordinating elements ``Is the listening level sufficient?'' and ``Is the listening level sufficient?''

(2) performs the following processing.

(1) Convert coordinating elements into pronouns. For example, as shown in the underlined part of (2) using a dotted arrow, the following is transformed into a pronoun, such as "...".

(2) List the contents of coordinately connected elements. For example, as shown using dotted arrows, the coordinating elements of the original text (1) are itemized, such as "Is the level of listening sufficient?" and "Is it uneven?"

0 is displayed. This shows the contents processed in ■ in Figure 2 (
Display as shown in (2) of b).

In (2), the coordinately connected elements are determined in response to the user seeing the display (2) in the display example of FIG. 2 (B) and agreeing to the display. Then machine translate it into the target language.

On the other hand, in response to the user's rejection of the display (2) in the display example in Figure 2 (b), the user goes to the sentence analysis section 2 using ■ to search for another syntax tree, and if it is not We will notify you of this and encourage you to revise the original text.

As described above, coordinate connecting elements are found from the original text, and the found coordinate connecting elements are converted into pronouns, and their contents are listed in bullet points and displayed as a structured sentence, for example, as shown in Figure 2 (b) and d. , in response to the user's OK response, the coordinate connection elements are determined, and by performing machine translation based on this, even beginners can understand the coordinate connection relationships using structured sentences. It is possible to easily display and generate appropriate translations.

〔Effect of the invention〕

As explained above, according to the present invention, when there is ambiguity in the original text, dependency relationships that may be interpreted are sequentially displayed as structured sentences and the user is asked to select one. Even beginners who do not have knowledge of the range of sentences that can be processed by machine translation systems and who do not know the target language can easily and efficiently perform appropriate translations.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of one embodiment of the present invention, and FIG. 2 is a block diagram of another embodiment of the present invention. In the middle, 1 represents a writer's blade section, 2 a sentence analysis section, 3 a structured sentence generation section, 4 a display section, and 6 a target language generation section.

Claims (2)

[Claims] (1) In a machine translation system that displays structured sentences that have possible interpretations of the original text, a sentence analysis unit (2) that analyzes the original text and generates a syntax tree that has possible interpretations; A structured sentence generation unit (3) generates a structured sentence from the syntax tree generated by the analysis unit (2), and the structured sentence generation unit (3) generates a possible interpretation generated from the original sentence. A machine translation system characterized in that structured sentences generated with respect to a syntax tree are sequentially displayed, and a displayed structured sentence that is approved by a user is determined as an appropriate sentence. (2) Claim No. (2) characterized in that the structured sentence is configured such that dependency relationships are displayed enclosed in parentheses.
The machine translation system described in section 1).