JPH03225468A - Machine translation device - Google Patents

Abstract

PURPOSE:To perform processing by covering a part on which correction or inspection is performed in translation by automatically extracting the key word of an input original text by using a key word information knowledge base, and expediting the input of the correction by outputting and displaying the data base information of the key word. CONSTITUTION:A text is read in with an original text input means 1, and a key word extraction means 2 segments an inputted original text to words, and extracts the word considered as the key word in the text in the translation. The key word extraction means 2 performs the verification of a significant key word selected based on the disribution of appearance frequence of an independent word (noun, verb, adjective, and adjective verb). Then, an unregistered word not being registered on a language information data base 4 is checked, and the key word information knowledge base 3 storing a constraint item not being set as the object of processing, a phrase to be corrected in the translation, and a point to which special attention should be paid is referred. In such a way, it is possible to reduce the omission of a significant part to be corrected, and to improve the efficiency of pre-edit.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a machine translation device, and particularly to a machine translation device that uses a computer to support translation and perform translation.

[Conventional technology]

There are many linguistic phenomena in natural language, and it is difficult to handle all of them with a machine translation device. Conventional machine translation devices handle linguistic phenomena within a limited range, and for linguistic phenomena that cannot be handled, a user who is familiar with translation and operation must read the original text before translation and handle it correctly. We perform a process called pre-editing, which involves identifying possible parts and modifying them to make them easier for machine translation equipment to analyze. Appropriate pre-editing is essential for obtaining high-quality translations.
Work efficiency varies depending on the skill and experience of the user. In addition, pre-editing has the aspect that it is difficult to understand which parts need to be corrected unless you translate the text twice.

[Problem to be solved by the invention]

The conventional machine translation device described above relies on the user's estimation for pre-editing, and has the disadvantage that the user must be familiar with the device's capabilities and translation.

In addition, conventional machine translation devices vary depending on the user's skill level.
Pre-editing can lead to differences in work speed and translation quality, as well as problems being discovered only after the actual translation, or conversely, corrections being made even though they don't need to be made. This has the problem that the work becomes wasteful.

The purpose of the present invention is to mechanically extract keywords that should be corrected or inspected in the original text, such as words that appear frequently or words that are treated by the device, and to encourage the user to input them. By doing so, it is an object of the present invention to provide a machine translation device that reduces oversight of important parts that require correction and improves the efficiency of pre-editing.

[Means to solve the problem]

The machine translation device of the first invention includes (A) an original text input means that reads and inputs the original text to be translated; (B) restrictions that are not processed by the translation machine and words and phrases that must be taken into account when translating. (C) dividing the original text input by the original text input means into words and using the information in the keyword information knowledge base;
(D) A linguistic information database that stores word ideas and grammatical information of the source language; (E) A keyword extraction means that extracts keywords from the original text; (E) A linguistic information database that stores the grammatical information of the keywords extracted by the keyword extraction means; (F) a user corrects or changes the grammar information of the keyword outputted by the grammar information output means;
Grammar information correction means to be stored in the linguistic information database; (G) After making necessary corrections in the grammar information correction means, syntactic analysis and morphological analysis are performed using the corrected information in the linguistic information database.

The system is equipped with a syntactic/semantic analysis means that performs translation through semantic analysis, etc.

In addition, the machine translation device of the second invention includes: (A) keyword extraction criteria such as constraints that are not processed by the translation machine and words and phrases that must be taken care of when translating, in advance in the keyword information knowledge base; (B) Store the word clouds and grammatical information of the source language in a linguistic information database in advance; (C) Read and input the original text to be translated; (D) Convert the input source text into words. (E) extracting the keywords in the input original text using the information in the keyword information knowledge base, and (E) extracting the grammatical information of the extracted keywords from the linguistic information database for the user. (F) the user corrects or changes the grammatical information of the output keyword and stores it in the linguistic information database; (G) uses the corrected information in the linguistic information database; It consists of performing translation using syntactic analysis, morphological analysis, semantic analysis, etc.

〔Example〕

Next, embodiments of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram showing an embodiment of the machine translation device of the present invention.

As shown in Figure 1, first, an optical character reader (OCR)
), etc., reads and inputs the original text to be translated.

Further, the keyword information knowledge base 3 stores information that serves as keyword extraction criteria, such as restrictions that are not processed by the translation machine and words and phrases that must be taken into account during translation.

Then, the keyword extracting means 2 divides the original text input by the original text inputting means 1 into words, and uses the information in the keyword information knowledge base 3 to extract keywords from the original text.

On the other hand, the linguistic information database 4 stores vocabulary and grammatical information of the source language.

Therefore, the grammar information output means 5 such as a display extracts the grammar information of the keyword extracted by the keyword extraction means 2 from the language information database 4 and outputs it for the user.

Further, the grammar information modification means 6 allows the user to modify or change the grammar information of the keyword outputted by the grammar information output means 5 and stores it in the language information database 4.

After the syntactic/semantic analysis means 7 performs the necessary corrections by the grammatical information correction means 6, the syntactic analysis/semantic analysis means 7 uses the information in the corrected language information database 4 to perform syntactic analysis, morphological analysis, etc.
Translation is performed using semantic analysis, etc.

Next, "This text describes the operating system. The operating system consists of the following three parts. (The text continues below)
The operation of the machine translation device of this embodiment will be explained using a sentence starting with "..." as an example.

First, this sentence is read by the original text input means 1, such as an optical character reader (OCR), and the keyword extraction means 2 divides the input original text into words, which are considered to be keywords for translation in this sentence. Extract words.

The keyword extraction means 2 verifies important keywords selected based on the appearance frequency distribution of independent words (nouns, verbs, adjectives, adjective verbs), and checks for unregistered words that are not registered in the linguistic information database 4. The keyword information knowledge base 3 is referred to, which stores constraints that are not subject to processing, words and phrases that must be corrected during translation, and points to note when describing the original text.

Note that these extraction criteria can also be changed by the user by setting parameters.

In this example, the independent words that appear frequently are ``state'' and ``state''.
``Operating system'' is extracted as an unregistered word, and ``naru'' is extracted as a highly ambiguous word requiring caution.

and "state", "operating system",
Linguistic information database 4 using “naru” as a keyword
The grammatical information stored in is searched and output word by word to grammatical information output means 5 such as a display.

FIG. 2 is an information configuration diagram showing an example of grammatical information in the language information database of this embodiment.

As shown in FIG. 2, the user first selects rstate J
J, reXpress J, rdescr
You can select lbeJ and confirm that it is translated using rstate J.

In this example, it is more appropriate to translate "state" as r descrlbeJ, so the user uses the grammar information correction means 6 to input the language using the mouse and keyboard to set the highest priority language to r descrlbeJ, and then translates the language information into r descrlbeJ. Change information in database 4.

FIG. 3 is an information configuration diagram showing an example of the grammar information editing screen of the grammar information correction means 6 of this embodiment.

Next, the grammar information output means 5 displays that "operating system" is an unregistered word, and then the third
The grammar information editing screen of the grammar information correction means 6 shown in the figure is displayed to prompt the user for input.

Therefore, the user uses the translated word operating sys.
By inputting temJ from the keyboard and selecting grammatical information such as part of speech and semantic classification using the mouse, these are stored as information in the linguistic information database 4.

Next, the grammar information output means 5 displays a warning that "naru" is a highly ambiguous word, and outputs a message prompting the user to "re-enter in kanji or specify a translation." , the grammar information correction means 6 re-inputs in kanji or specifies a translated word.

Then, the syntax/semantic analysis means 7 performs translation using the language information database 4 in which all such corrections have been made. In addition, as a reference for the syntactic analysis/semantic analysis means 7, there is Japanese Patent Application No. 183435/1983.

〔Effect of the invention〕

As explained above, the machine translation device of the present invention automatically extracts keywords from the input original text using the keyword information knowledge base, and outputs and displays the database information of the keywords to the user to facilitate input of corrections. Therefore, it has the effect of being able to cover all the parts that are important in translation and need to be corrected or inspected.

In addition, the machine translation device of the present invention eliminates significant differences in work efficiency depending on the skill level of the user and the oversight of items that require correction due to carelessness, and obtains uniform pre-editing results.
This has the effect of improving translation quality, reducing the number of retranslations due to insufficient pre-editing, and improving translation efficiency.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of the machine translation device of the present invention, FIG. 2 is an information configuration diagram showing an example of grammatical information of the language information database of the present embodiment, and FIG. 3 is a grammar diagram of the present embodiment. 3 is an information configuration diagram showing an example of a grammar information editing screen of the information correction means 6. FIG. 1...Original text input means, 2...Keyword extraction means, 3...Keyword information knowledge base, 4...Language information database, 5...・
- Grammar information output means, 6... Grammar information correction means, 7... Syntax analysis/semantic analysis means.

Claims (1)

[Claims] 1. (A) Original text input means for reading and inputting the original text to be translated; (B) Restrictions that are not processed by the translation machine and words and phrases that must be taken care of during translation. a keyword information knowledge base that stores information serving as keyword extraction criteria; (C) dividing the original text input by the original text input means into words, and using the information in the keyword information knowledge base;
(D) a linguistic information database that stores vocabulary and grammatical information of the source language; (E) extracting grammatical information of the keywords extracted by the keyword extracting means from the linguistic information database; grammatical information output means for extracting and outputting for the user; (F) allowing the user to modify or change the grammatical information of the keyword output by the grammatical information output means;
grammatical information correction means to be stored in the linguistic information database; (G) after making necessary corrections in the grammatical information correction means, syntactic analysis, morphological analysis, and semantic analysis using the corrected information in the linguistic information database; A machine translation device characterized by comprising syntactic analysis and semantic analysis means for performing translation using methods such as syntactic analysis and semantic analysis means. 2. (A) In advance, store in the keyword information knowledge base information that serves as criteria for keyword extraction, such as restrictions that are not subject to processing by the translation machine and words and phrases that must be noted during translation; (B) Store the vocabulary and grammatical information of the source language in the language information database; (C) read and input the original text to be translated; (D) divide the input original text into words and use the information in the keyword information knowledge base. (E) extracts the grammatical information of the extracted keywords from the linguistic information database and outputs them for the user; (F) extracts the keywords from the input original text; The user corrects or changes the grammatical information and stores it in the linguistic information database, and (G) performs syntactic analysis, morphological analysis, semantic analysis, etc. using the corrected information in the linguistic information database. A machine translation device that performs translation.