JPH03157766A - Machine translation device - Google Patents

Abstract

PURPOSE:To facilitate the designation of a part of speech and to improve the processing speed by providing a part-of-speech display column for displaying all part-of-speech information which a specific word designated at the time of designating a part of speech of pre-editing has, on a screen of a display device. CONSTITUTION:The translation system is constituted of input devices 1 - 3, a spelling check part 4, a pre-editing part 5, a translation body (translation processor) 6, a dictionary 7, a grammar dictionary 8, a post-editing part 9, a file 10, a printer (output device) 11, and a display device (CRT) 12. In such a state, a part-of-speech display column for displaying all part-of-speech information which a specific word designated at the time of designating a part of speech of pre-editing has is provided on a screen of the display device 12. In such a way, a part of speech can be designated easily, and the processing efficiency can be improved.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of the Invention The present invention relates to a machine translation device, such as a Japanese-English machine translation system.

2. Description of the Related Art Generally, in this type of machine translation system, a translated text is obtained by performing processes such as morphological analysis, syntactic analysis, conversion, and generation on an input original text.

In such natural language parsing, processing multi-part speech words is a very difficult problem. This is because when a word has multiple meanings, various analysis results are derived, and anything other than the correct answer is an incorrect analysis.

Therefore, polysemy resolution is required for words with multiple parts of speech.
This has two meanings in terms of syntactic analysis. One is that ambiguity resolution significantly reduces erroneous analysis and improves analysis accuracy. Another point is that since other parts of speech are excluded, there is almost no needless analysis other than the analysis that leads to the correct answer, so the analysis speed is improved. For example, if an original sentence with multiple parts of speech words is analyzed as it is and eight possible solutions are obtained, eliminating the multiple parts of speech can be expected to greatly reduce the number of solution candidates, and the number of times the analysis rules are applied. The number of combinations is reduced, and the burden of syntax analysis is significantly reduced.

One method for achieving such resolution of ambiguity is to specify the part of speech of words that have ambiguity in the original text at the pre-editing stage before translation. For example, JP-A-61-740
According to Publication No. 69, the user is allowed to specify the part of speech of single δ1 for translation.

However, the problem that the invention aims to solve is that the user must have knowledge of what part of speech a certain word has, and it is not easy to specify the part of speech.

Means for solving the problem An input device for inputting the original text in the first language, a display device for displaying the input information, and after parsing the input original text, converting it into a single syntactic unit. A translation processing device that generates a translated text and an output device that outputs a translated text in a second language are installed, and the translation processing device performs syntax analysis on a specific word as a pre-editing of the original text before translation. In the machine translation device in which a part of speech is specified and input, the invention according to claim 1 provides a part-of-speech display field that displays all part-of-speech information of a specific word specified when specifying a part of speech in pre-editing. In the invention described, the word display field that displays all words within the specified source text range that has a specific part of speech specified when specifying the part of speech in @edit is set to +11f.
f In the invention described in claim 3, word display columns are provided on the screen of each display device to display all words within the specified source text range that have multiple specific parts of speech specified when specifying the parts of speech in pre-editing. .

According to the invention described in claim 1, when specifying the part of speech in pre-action editing, if a target word with multiple meanings is specified, all part of speech information of that word is displayed to the user in the part of speech display field. , the user can easily specify the part of speech by selecting the appropriate part of speech.

On the other hand, according to the invention as claimed in claim 2, if a certain part of speech is specified, all words having that part of speech within the specified range of the original text are displayed to the user in the word display field. When specifying the same part of speech for words, the specification processing can be performed continuously, improving processing efficiency. especially,
According to the invention described in claim 3 of 51'1, multiple parts of speech are specified and the corresponding words are displayed, so even if the parts of speech for these words are not necessarily the same,
Specified processing can be performed continuously, further improving processing efficiency.

Example An embodiment of the present invention will be described based on the drawings.

This example is applied to an English-Japanese machine translation system in which the input text is English and the output text is Japanese.
The figure shows an overview of the system configuration.

This is a translation algorithm that uses a syntactic conversion method using semantic information, and the original text obtained by either 1 file/Tsuba keyboard 2 input or 0CR3 input is spell checked by the spell check section 4, and then It undergoes necessary pre-editing by the pre-editing section 5 and is subjected to translation processing by the translation main body (translation processing device) 6. For spell checking, dictionary '87 in the basic dictionary, user dictionary, etc. is referred to. Pre-editing also includes a process of specifying the part of speech of a specific word in the text in order to perform sentence analysis on the correct target text (original text). In addition to this dictionary 7, a grammar dictionary 8 is used in the translation process.

The translation output obtained by the translation main body 6 is appropriately edited by a post-editing section 9 using translation information, and is stored in a file 10. Further, the input sentence and the output sentence can be printed using a printer (output device) 11. Furthermore, during these processes, input sentences and output sentences are displayed on CR as a display device.
Displayed on the T12 screen.

The processing units of the translation main body 6 can be roughly divided into a morphological analysis unit 13, a sentence analysis unit 14, a conversion unit 15, and a generation unit 16, as shown in FIG. First, the morphological analysis unit 13
The input text is looked up in the dictionary. Next, the syntactic analysis unit 14 obtains information about each word and performs syntactic analysis (parsing) according to the grammar rules. A tree structure is created from this analysis result.

The conversion unit I5 transforms the tree structure of the original sentence into the main structure of the outgoing sentence. The generation unit (6) translates the obtained book structure node by node and generates a translated sentence.

Therefore, this embodiment is characterized by specifying parts of speech in pre-editing in the pre-editing section 5, and the screen transition on the editing screen (editor) 17 of the CRT 12 shown in FIG. 1 is referred to. I will explain.

First, the editing screen 17 in FIG. 1(1) shows the initial state,
For example, the original text (input text) “Time flies”
ike an arrow, is displayed on the left side of the screen.

"Work", "Edit", etc. at the top of the screen 17 indicate menus. For example, when "Pre-edit" is selected, a pre-edit menu window 18 is opened as shown in FIG. It becomes possible to call the part-of-speech specification function.

First, let's look at the example sentence "Time flies 1ike"
Considering the parts of speech of the six words anarrow and , Time: noun, verb, adverb flies: verb 9 nouns 1ike: preposition, verb, adjective, conjunction 2 nouns an: article arrow: noun. However, here the parts of speech are classified into major categories, not subdivisions, but in actual use.

It is preferable to display both the detailed classification level and the major classification level for selection.

In the pre-editing menu display state shown in FIG. 1(b), select the part-of-speech designation function. This part-of-speech specification includes "word specification"
There are two submenus: and "Specify part of speech.""Wordspecification" is according to the invention as claimed in claim 1, and when a target word is specified with a cursor or the like, all the words of that word are displayed in a window 19 that is a part of speech display column on the editing screen 17. It is something. For example, if you select [word specification J] in the state shown in FIG.
As shown in c), the message [Please specify a word with the cursor] is displayed, and following this message, for example, "1ik
When the word “e” is specified, “
All part-of-speech information of “1ike” “Preposition J “Verb”
"Adjectives""Conjunctions""Nouns" are in window 19
is displayed, and the message ``Please specify the part of speech'' is also displayed. Then, the user specifies and inputs an appropriate part of speech from among the parts of speech displayed in the window 19. At the time of syntax analysis, only the part of speech specified at this time is referenced for ``1ike''. Therefore, the word is treated as having no ambiguity, and analysis is performed reliably and efficiently. At this time, it is not necessary to specify the part of speech, but if it is not specified, all parts of speech are referred to for "1ike".

On the other hand, [part of speech specification] is based on the invention as claimed in claim 2 or 3, and when any desired part of speech is specified within the specified range of the original text, words having that part of speech within the specified range are displayed on the editing screen 17. The words are displayed in a highlighted display 2o, which is a word display column.

In this case, multiple parts of speech may be specified. For example, in the same figure (b
) If you select "specify part of speech" from the state shown in (e
), various parts of speech are displayed in the window 21, and a message "Please specify the part of speech" is also displayed. Therefore, the user refers to the window 21 and specifies the part of speech. For example, if you specify the noun J, the message ``Please specify the target range'' is displayed as shown in FIG. 5(f), prompting you to specify the range in the original text. Here, if the entire example sentence is targeted, all words with the part of speech "noun" specified in the example sentence (specified range), that is, "T i
me” “1ias” “1ike” “arro
w is highlighted 2o as shown in FIG. 2(g).

The highlighted display 20 may be an underline display as shown in the figure, a shaded display, a different type of character display such as Gothic/Mincho typeface, etc. This highlighted display 2o allows the user to at least know which words have the part of speech ``noun J,'' and to determine whether it is appropriate to call these words ``noun'' or whether it is better to use another part of speech. It can be used for reference. So, for example, ``” Time
When you specify , the same figure (
As shown in g), all parts of speech that “Time” has, i.e.,
Not only ``nouns,'' but also ``verbs'' and ``adverbs'' are displayed in the window 19, and selections are made according to the message ``Please specify part of speech.'' Therefore, one part of speech is specified.

In this case as well, you do not need to specify it, but if you do, only that part of speech will be referenced during syntax analysis. Hereinafter, the part-of-speech designation is performed in the same way for other words that have the part-of-speech "noun."

By the way, consider a case where multiple parts of speech are specified. For example, if you specify "verb" and "noun" from the screen state shown in Figure 1(e), the display will go through the display shown in Figure 1(f) (with the entire original text as the specified range), and then as shown in Figure 1(h). As in “verb” and “
The word “Time” fl has both parts of speech with “Noun”
ies "1ike" is displayed by highlighting 20. Therefore, for example, if you specify "Time", the "Time
All parts of speech possessed by are displayed in the window 19 for selection. Therefore, one part of speech is specified. "verb"&
The part of speech is specified in the same way for the word "ike" which has the part of speech "noun".

Effects of the Invention As described above, when specifying the part of speech in pre-editing, in the invention according to claim 1, if a target word with multiple meanings is specified, all the part of speech information of that word is displayed in the part of speech display field. Since the user can easily specify the part of speech by selecting the appropriate part of speech, the invention described in JlIJ Claim 2,
When a certain part of speech is specified, all words with that part of speech within the specified source text are displayed to the user in the word display field, so it is possible to specify the part of speech for these words as if they were the same part of speech. In this case, the parts of speech can be specified continuously, and the processing efficiency can be improved. In particular, in the invention as claimed in claim 3, multiple parts of speech are specified and the corresponding words are displayed. Even if the part-of-speech designations for these words are not necessarily the same part-of-speech, the designation can be processed continuously, and the processing efficiency can be further improved.

[Brief explanation of the drawing]

The drawings show one embodiment of the present invention; Fig. 1 is a front view of an editing screen showing the transition of screen display, Fig. 2 is a block diagram of the translation system, and Fig. 3 is a block diagram of the translation main body. It is. 1.2.3... Input device, 6... Translation processing device, 7
...Dictionary, 8...Grammar rules, l■...Output device,
12...Display device, 17...Screen, 19...Part of speech display column, 20...Word display column diagram (violet)

Claims (1)

[Claims] 1. An input device for inputting an original text in a first language, a display device for displaying the input information, and an input device for parsing the input original text and syntactically combining it into one unit. It is equipped with a translation processing device that converts and generates a translated text, and an output device that outputs a translated text in a second language. A machine translation device in which a part of speech is specified and input, characterized in that a part of speech display column is provided on the screen of the display device for displaying all part of speech information of a specific word specified when specifying a part of speech in pre-editing. Machine translation device. 2. An input device for inputting the original text in the first language, a display device for displaying the input information, and after parsing the input original text, it is syntactically converted into a single unit and a translated text is generated. and an output device that outputs a translated sentence in a second language, and is configured to input the part of speech of a specific word in order to perform syntactic analysis as a pre-editing of the original text before translation by the translation processing device. The machine translation device according to the present invention is characterized in that a word display column is provided on the screen of the display device for displaying all words within a specified source text range that have a specific part of speech specified when specifying the part of speech in pre-editing. Machine translation device. 3. An input device for inputting the original text in the first language, a display device for displaying the input information, and after parsing the input original text, it is syntactically converted into a single unit and a translated text is generated. and an output device that outputs a translated sentence in a second language, and is configured to input the part of speech of a specific word in order to perform syntactic analysis as a pre-editing of the original text before translation by the translation processing device. The machine translation device according to the present invention is characterized in that a word display column is provided on the screen of the display device for displaying all words within a specified source text range that have a plurality of specific parts of speech specified when specifying parts of speech in pre-editing. Machine translation device.