JPH0754522B2 - Natural language compound noun analysis and conversion method - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to a device for automatically converting a natural language read from an input device into another natural language. The present invention relates to a natural language compound noun analysis / conversion method that analyzes and converts a sentence that includes two or more words arranged one by one.

(Prior Art) In conventional automatic translation systems, compound nouns are handled as follows.

Human beings intervene before the human-interpreted translation process, and after translating compound nouns into noun phrases that can be analyzed by the system, they are translated.

Direct translation type Each noun in the compound noun of the source language is translated as it is, and translated into the target language in the same order as the source language.

Each of these methods has the following drawbacks.

Since humans are involved, translation takes time and costs increase.

Since it does not take into account the relationship between each noun of a compound noun,
Cannot generate a proper translation.

(Object of the Invention) An object of the present invention is to provide a method for automatically generating an appropriate compound noun in order to solve the above-mentioned drawbacks of the translation generation of the compound noun of human intervention type or direct translation type. It is in.

(Structure of Invention) (Differences Between Features of Invention and Conventional Techniques) The present invention analyzes and converts a compound noun of a source language, and for the two processing units, a part-of-speech and a compound word translation are used as keys. Search the table and determine the translation pattern.

A translated word for one processing unit is generated according to a translation pattern using a bilingual dictionary. (At this time, the face and meaning of the word in the compound word are used.) The superposition method is applied to each unit to obtain the final translated word.

That is the most main feature.

Humans do not intervene with conventional technology.

 Can generate appropriate translations.

That is different.

(Embodiment) FIG. 1 is a diagram showing an example of the entire basic configuration of the automatic translation system of the present invention, in which 1 is an input device for inputting a source language, and 2 is an analysis / conversion for analyzing / converting the source language. A device 3 is a generation device for generating a target language based on the output of the analysis / conversion device 2,
4 is an output device for outputting the generated target language, and 5 is analysis /
A conversion dictionary, and 6 is a generation dictionary.

FIG. 2 is a diagram showing a configuration example of a compound noun analysis / conversion device, in which 7 is a dependency of processing units within a compound noun,
A compound noun analysis device for determining a translation pattern, 8 is a partial result translation device for translating a unit according to a translation pattern based on the output of the compound noun analysis device 7, and 9 is a partial result translation device 8
, A buffer for storing the output of the, a superimposing device for outputting a compound noun in the final target language from the result stored in the buffer 9, a control device for controlling the above flow, and a translation pattern for the relation. Is a compound word translation table for selecting, and 13 is a bilingual dictionary.

FIG. 2 is a part of FIG. 1, in which the compound noun analysis device 7 is provided in the analysis / conversion device 2, the partial result translation device 8, the buffer 9,
The superimposing device 10 causes the generating device 3 to display a compound word translation table.
12 is the analysis / conversion dictionary 5, bilingual dictionary 13 is the generation dictionary 6,
Each corresponds.

FIG. 3 is a diagram showing the flow of processing of the control device 11 in FIG. 2, in which processing is performed by sending a command to the compound noun analysis device 7, and then to the partial result translation device 8 and buffer 9. Processing 15 is performed by sending a command, and processing 16 is performed by sending a command to the superposition device 10.

FIG. 4 is a diagram showing a flow of processing of the superposition method in FIG. 3. First, when the translation words of the processing units have the same translation word, the translation words are superposed using the word as a key. If there is no such word, it is checked whether or not there is * in one of the processing units, and if there is, the translated words of other processing units of * are superposed. If not, connect the translated words as they are. This is applied to all processing units to obtain translated words.

It should be noted that 18 to 22 in the figure indicate the number of each process.

FIG. 5 is a diagram showing a part of the compound word translation table, in which the part of speech of the front element of the processing unit is a sahen noun, the part of speech of the rear element is a general noun, and the front element and the rear element It shows that the processing unit is translated as a past poem + a pattern of a single noun when it has a target relation with the element of. The part of speech of the element in front is a general noun, and the part of speech of the element in the back is a general noun. Then, the processing unit is translated as a single noun + single noun pattern.

The operation of this apparatus will be described below with reference to FIGS. 6 and 7.

FIG. 6 is a diagram showing an example of the operation of this device taking the “retirement businessman group” as an example.

First, the compound noun “retirement businessman group” is processed by the compound noun analysis device 7, the partial result translation device 8 and the buffer 9. First, "Retirement businessman", which is one of the processing units, is processed.

The part-of-speech of a word in a processing unit is a sahen noun and a common noun, respectively. Since "businessman" has a relationship with "retirement" as the subject, a translation pattern [past participle + single Noun] is selected, and the translated word "retired businessman" is obtained from this pattern and the bilingual dictionary. Similarly, the translation pattern of “businessmen group” is selected as “single noun + single noun” and “busi
The translated word "nessman group" is obtained.

When these partial results are input to the superimposing device 10, the same word in the processing unit is used as a key to superimpose the translated word, and “reti
Get the translation "red businessman group".

FIG. 7 is a diagram showing an example of the operation of this device, taking the "machine translation dictionary" as an example.

First, a compound noun called “machine translation dictionary” is processed by the compound noun analysis device 7, the partially combined translation device 8 and the buffer 9. First, "machine translation", which is one of the processing units, is processed.

The part-of-speech of the stored word in the processing unit is a general noun and a general noun, respectively, and the translation pattern "single noun + single noun" is selected, and the translated word "machine translation" is obtained.

Next, for "translation", it is a general noun and an affix, and [* + for + single noun] is selected as the translation pattern.
As a result, the translated word "* for translation" is obtained.

Finally, for the "dictionary", affixes and common nouns, "single noun" is selected as the translation pattern, and
The translated word "ctionary" is obtained.

When these partial results are input to the superimposing device 10, the translated words are superposed with the same word in the processing unit as a key, and "* fo
r michine translation "and" dictionary ".

Next, consider superimposing the two. Since there is no same word, but there is *, this matches with "dictionary" and the translated word "dictionary for michine translation" is obtained.

(Effects of the Invention) As described above, according to the present invention, an appropriate translated word of a compound noun can be generated without human intervention.

[Brief description of drawings]

FIG. 1 is a diagram showing an example of the entire basic configuration of the automatic translation system of the present invention, FIG. 2 is a diagram showing an example of the configuration of a compound noun automatic translation device, and FIG. 3 is a control flow in the control device of FIG. Fig. 4 is a diagram showing the flow of processing by the superposition method, Fig. 5 is a diagram showing an example of a compound word translation table, and Fig. 6 is this device taking "retirement businessman group" as an example. FIG. 7 is a diagram showing an example of the operation of the present device, taking the “machine translation dictionary” as an example. 1 ... input device, 2 ... analysis / conversion device, 3 ... generation device, 4 ... output device, 5 ... analysis / conversion dictionary, 6 ... generation dictionary, 7 ... compound noun analysis device, 8 ... … Partial result translation device, 9 …… buffer, 10 …… superposition device, 11 …… control device, 12 …… compound word translation table, 13 …… translation dictionary.

Claims (1)

[Claims] 1. A source language of Japanese input from an input device is translated into a target language which is another natural language,
In an automatic translation system that outputs the result from an output device, a compound noun analysis means that searches a compound word translation table created in advance based on the part of speech and the relation for the processing unit in the compound noun and determines the translation pattern A partial result translating means for generating a translated word for each processing unit according to a translation pattern using a bilingual dictionary stored in advance in a predetermined storage device based on the output of the compound noun analyzing means, and storing the translated word in a predetermined buffer area. And a superposition means for outputting a compound noun of the target language by applying a superposition method to each unit from the result stored in the predetermined buffer area by the partial result translating means. Natural language compound noun analysis and conversion method.