JPH0566630B2 - - Google Patents

Description

[Detailed description of the invention]

<Industrial Application Field> The present invention relates to a machine translation device that can classify and translate arbitrary syntax of an input sentence. <Prior Art> Conventionally, there are the following machine translation devices. This machine translation device inputs the source language input from the keyboard into the translation module under the control of the main CPU, and the input source language is transferred by the translation module to the dictionary, grammar rules, and It uses tree structure conversion rules to translate into the target language. The above grammar rules are used to determine the syntactic structure of an input sentence (hereinafter referred to as the original sentence). <Problems to be Solved by the Invention> However, when translating with the above-mentioned conventional machine translation device, the syntactic structure of the original text is determined according to the grammatical rules stored in memory. If the grammar rules stored in memory do not correspond to a particular construction in the source text,
There is a problem that the original text may be misinterpreted or become untranslatable. Furthermore, even if the above-mentioned grammar rules correspond to specific syntax in the original text, there is a problem that it is easy to misinterpret the original text if the dependency is very complicated. Furthermore, when a long sentence with complex dependencies is input, there is a problem that the processing time increases in proportion to the length of the sentence. Therefore, an object of the present invention is to attach a divided translation mark to a part of an arbitrary syntax in the original text, so that only the syntax in that part can be divided and translated. To provide a machine translation device which can dilute an original text correctly and at high speed even when the original text is not compatible with the original text or has complicated dependencies. <Means for Solving the Problems> In order to achieve the above object, the present invention includes a morphological analysis unit that divides an input sentence into morphemes using a dictionary and obtains information such as parts of speech, and the morphological analysis unit. a syntactic analysis unit that analyzes the syntax of the morpheme sequence decomposed by using a dictionary and grammar rules; a conversion unit that exchanges the syntactic structure obtained by the syntactic analysis unit into a syntax structure in the target language; In a machine translation device having a translated sentence generation section that generates a translated sentence according to the syntactic structure of the target language obtained by the conversion section, a divided translation section that divides and translates arbitrary syntax of an input sentence. a mark input means for inputting a divided translation part mark for specifying the divided translation part mark; a mark discrimination means for determining whether or not the divided translation part mark is present in the input word string; Determine whether the grammar rule is a rule for split translation, and if it is not a rule for split translation, enable the use of the imported grammar rule,
When the rule is for split translation, grammar rule discriminating means is provided that enables the use of the split translation rule only when the mark discriminating means determines that there is a split translation part mark in the word string. It is characterized by <Operation> A divided translation part mark is inputted from the mark input means to an arbitrary syntax of an input sentence to specify that this syntax is a divided translation part that is divided and translated. When a sentence to which the divided translation part mark has been given is input to the divided translation part mark determination means, it determines whether there is a divided translation part mark for specifying the divided translation part in the word string of the inputted sentence. is discriminated by the mark discrimination means of the divided translation unit, and then divided into morphemes by the morphological analysis unit to obtain information such as parts of speech.Based on this information and grammar rules, the syntactic analysis unit determines the syntax of the morpheme string. Analysis is performed. and,
The syntactic structure obtained by the syntactic analysis section is converted into the syntactic structure of the target language by the conversion section, and the translated sentence generation section generates a translated sentence according to the syntactic structure of the target language. . In such a translation process, when executing the syntactic analysis, the grammar rule determining means determines whether the imported grammar rule for syntactic analysis is a split translation rule. As a result, when the rules are not for split translation, the imported grammar rules can be used, but when the rules are for split translation, the split translation part mark discriminating means detects that there is a split translation part mark in the word string. The above divisional translation rule can be used only when it is determined. Therefore, only when a divided translation part in the input word string is specified by the divided translation part mark, the rules for divided translation are used to perform syntax analysis, and the divided translation is distinguished from others. It is translated as follows. <Examples> The present invention will be described in detail below with reference to illustrated examples. FIG. 1 is a blog diagram of an embodiment of the machine translation apparatus of the present invention. 1 is the main CPU (central processing unit), 2 is the main memory, 3 is the CRT (cathode ray tube), 4 is the keyboard, 5
6 is a translation module, and 6 is a memory that stores a translation dictionary, grammar rules, tree structure conversion rules, and the like. When a source language is input, the translation module 5 translates it and outputs a target language. That is, the source language input from the keyboard 4 is sent to the translation module 5 under the control of the main CPU 1. Translation module 5
uses the dictionary, grammar rules, tree structure conversion rules, etc. stored in the memory 6 to translate the input source language into the target language as will be detailed later. The results are temporarily stored in the main memory 2 and displayed on the CRT 3. The specific configuration of the translation module 5 of this embodiment is shown in FIG. The operation of the translation module will be explained below with reference to FIGS. 1 and 2. First, the read original text is divided into each morpheme string (word string) by the dictionary reference morphological analysis unit 11 using the dictionary in the memory 6, and grammatical information such as part of speech and translation for each word are obtained. Tense, person, number, etc. are analyzed. In the syntactic analysis unit 12, a structural analysis tree indicating the dependency relationship between each word is determined as shown in FIG. 5, as will be described later, according to the dictionary in the memory 6 and the grammar rules. The conversion unit 13 uses the tree structure conversion rules in the memory 6 to convert the structure of the parse tree for the input original text into the structure of the parse tree for the translation. Then, the translated sentence generation unit 14 adds particles and auxiliary verbs appropriate for translation in the target language, and the obtained translated sentence is output from the translation module 5 and stored in the main memory 2.
Displayed on CRT3. Next, in the operation of the above-mentioned translation module, a divisional translation process in which an arbitrary syntax of an input sentence is separated from others and translated will be explained in detail. Here, it is assumed that the memory 6 stores only the grammar rules shown in Table 1.

【table】

[Table] The grammar rules in Table 1 above represent the following. In other words, grammar rule (a) indicates that a prepositional phrase is made up of a preposition and a noun. Grammar rules (b)
indicates that a noun phrase is made up of a determiner and a noun. Grammar rule (c) states that a synonymous phrase is made up of a BE verb and an adjective. Grammar rule (d) is
A declarative sentence is made up of a noun phrase and a verb phrase. Grammar rule (e) indicates that a THAT clause consists of THAT and a declarative clause. Grammar rule (f) is
The minute indicates that it is made up of a declarative minute and the end of a sentence.
Grammar rule (g) indicates that a sentence is made up of a subtree and the end of the sentence. Grammar rule (h) indicates that a subtree is made up of subtrees, double slashes, and subtrees. Grammar rule (i) indicates that a subtree is made up of prepositional phrases. Grammar rule (j) is that a subtree is
Indicates that it is made up of a THAT clause. Grammar rule (k) indicates that the subtree is made up of BE verbs. Among the grammar rules (a) to (k) above, grammar rule (g)
~(k) is the split translation rule used in the split translation process. FIG. 3 is a flowchart of translation processing according to this embodiment. The operation of the translation process will be explained below with reference to FIG. In step _S1 , a split translation flag indicating execution of split translation processing is reset. In step _S2 , it is determined whether or not there is a double slash "" in the input word string, which is a divided translation part mark that designates a divided translation part. the result,
If "" is present, proceed to step _S3 ; otherwise, skip step _S3 . In step _S3 , the split translation flag is turned on to indicate the execution of the split translation process. In step _S4 , dictionary lookup processing is executed by the dictionary lookup morphological analysis unit 11 to obtain grammatical information such as parts of speech and translated words, and further analysis of tense, person, number, etc. is performed. In step _S5 , the syntax analysis unit 12 performs syntax analysis and determines the structure of the syntax analysis tree. In step _S6 , the conversion unit 13 converts the structure of the parse tree of the source language into the structure of the parse tree of the target language. In step _S7 , the translated text generating section 14 generates an appropriate translated text. Next, the syntax analysis process in step _S5 above is performed in the fourth step.
A more detailed explanation will be given according to the flowchart shown in the figure. In step _S51 , the next grammar rule is imported from among the grammar rules shown in Table 1. In step _S52 , it is determined whether the imported grammar rule is a rule for split translation. As a result, if the rule is for split translation, the process proceeds to step _S54 ; otherwise, the process proceeds to step _S53 . In step _S53 , parsing is performed using the captured grammar rules and the process proceeds to step _S56 . In step _S54 , it is determined whether the divided translation flag is on. If the result is on, proceed to step _S55 , otherwise proceed to step S55.
Proceed to S ₅₆ . In step _S55 , parsing is performed using the captured grammar rules (ie, rules for segmented translation). In step _S56 , it is determined whether or not there is a word string that has not been parsed in the input word string. As a result, if there is an unprocessed word string, the process returns to step _S51 .
Otherwise, the syntax analysis process ends. The above translation processing operation will be further explained using a specific input sentence. First, input sentence (a) “Among those
advantages are that instruments are effective.”
Assume that . This input sentence (a) has an inverted construction, and the normal word order is "That the
instrument is efficient is among those
advantages.” When the input sentence (a) is input, in the flowchart of FIG. 3, since there is no "" in the input word string, the process goes through steps S ₁ , S ₂ , and S ₄ and reaches step S ₅ . Analysis processing is executed according to the flowchart shown in the figure. However, as shown in Table 1, the grammar rules of this embodiment lack a grammar rule that can handle inverted syntax, so that syntactic analysis fails in normal translation processing. Then, from the keyboard 4, input sentence (b) with a double slash inserted into the above input sentence (a), ``Among
those advantages are that the instrument is
"efficient." is input. Then, in the flowchart of FIG. 3, since there is "" in the input word string, the divided translation flag is turned on in step _S3 via steps _S1 and _S2 . and,
Further, the process goes to step _S5 via step _S4 , and the syntax analysis process is executed as follows according to the flowchart of FIG. That is, let us first assume that the grammar rule (g) in Table 1 is imported. This grammar rule (g) is a rule for split translation as described above, and the split translation flag has already been turned on in step _S3 of FIG. Therefore, step S ₅₅ is passed through steps S ₅₁ , S ₅₂ , and S ₅₄ .
Using the grammar rule (g), the input sentence (b) is divided into a subtree and the end of the sentence, as shown in the top row of FIG. Next, assume that the grammar rule (h) in Table 1 is imported. As mentioned above, this grammar rule (h) has a rule for split translation, and the split translation flag is already turned on as described above. Therefore, steps S ₅₁ , S ₅₂ ,
Using grammar rule (h) in step S ₅₅ via S ₅₄ ,
As shown in the second row of FIG. 5, the above subtree is further divided into subtree+double slash+subtree.
Thereafter, the input sentence (b) is parsed in the same manner, and if the parsing is successful, a parsing tree of the source language is determined as shown in FIG. Thereafter, based on this parse tree, the split translation part between "" and "" is separated from the others, and conversion processing and generation processing are executed. As a result, as shown in Figure 6, the divided translation part between "" and "" in the original sentence is separated from others and translated, and the output sentence "The device between these advantages is highly efficient. “That” is
Output to CRT3. In this way, even in long sentences with complicated dependencies, “”
It is possible to divide the text into blocks and translate each block separately.There is no need to prepare special grammar rules even for special constructions such as inverted structures, and the overview of the grammar rules as a whole becomes better. Therefore, it becomes possible to interpret the original text correctly and quickly. Also, since it can parse even special syntax,
It is possible to identify the part of speech of an input word string, making it easy to change translations of words. <Effects of the Invention> As is clear from the above, the present invention provides a mark input means for inputting a divided translation unit mark into a machine translation device including a morphological analysis unit, a syntactic analysis unit, a conversion unit, and a translated sentence generation unit. , a mark determining means for determining whether or not the divided translation section mark is present in the input word string; and a mark determining means for determining whether or not the divided translation section mark is present in the input word string; A grammar rule determination means is provided that enables parsing to be performed using the above-mentioned rules for split translation only when there is a split translation mark, and the split translation part specified by the split translation part mark is distinguished from other parts. This makes it possible to interpret the original text correctly and quickly even when the grammar rules do not correspond to the measurement syntax or when a long sentence with complex dependencies is input. Become.

[Brief explanation of the drawing]

Fig. 1 is a block diagram of an embodiment of the machine translation device of the present invention, Fig. 2 is a functional block diagram of the translation module in the above embodiment, Fig. 3 is a flowchart of translation processing according to the above embodiment, and Fig. 4 is a flowchart of the parsing process, Fig. 5 is a diagram showing an example of a determined parsing tree, and Fig. 6 is an example of an input sentence to the first machine translation device and an example of an output sentence (translated sentence) for it. FIG. 1...Main CPU, 2...Main memory,
3...CRT, 4...Keyboard, 5...Translation module, 6...Memory, 11...Dictionary lookup morphological analysis section, 12...Syntax analysis section, 13...Conversion section, 14...Translation sentence generation section .

Claims (1)

[Scope of Claims] 1. A morphological analysis unit that divides an input sentence into morphemes using a dictionary and obtains information such as parts of speech; a parsing unit that parses using rules;
an exchange unit that exchanges the syntactic structure obtained by the syntactic analysis unit into a syntactic structure in the target language; and a translated sentence generation unit that generates a translated sentence according to the syntactic structure of the target language obtained by the conversion unit. In a machine translation device having a division, a mark input means for inputting a division translation unit mark for specifying a division translation unit that divides and translates arbitrary syntax of an input sentence; mark discriminating means for determining whether or not the above-mentioned divided translation part mark is present; On the other hand, when the rules are for split translation, the split translation rules are used only when the mark discrimination means determines that there is a split translation part mark in the word string. A machine translation device characterized by comprising a grammar rule discriminating means that makes it possible to