JPH0715691B2 - Automatic translator - Google Patents

Description

Detailed Description of the Invention

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an automatic translation device that uses a computer system to translate a sentence written in one language into a sentence written in another language. [Prior Art] Automatic translation between different languages has been attempted, but due to the diversity of natural languages, perfect translation is impossible, and errors often occur during processing, especially during analysis. However, in the conventional automatic translation apparatus, when an error occurs during analysis, the following two methods are used. Backtrack the process and look for another interpretation. In the part after the point where the error occurred, the translated words are simply arranged in word or phrase units. [Problems to be Solved by the Invention] By the way, in the conventional method of backtracking the above-mentioned processing to search for another interpretation, the memory capacity and time are very large, and in addition, Since there is no guarantee that the interpretation will always exist, it has the drawback that the entire sentence may fail to be parsed over time. Also, in the method of simply arranging translated words in units of words or phrases in the part after the error occurs, it is easy to associate the words with each other, but the part after the error occurs There was a drawback that was not a sentence and became a very hard to read translated sentence. The present invention has been proposed in view of the above points, and an object of the present invention is to provide an easy-to-read translated text about the whole when an error occurs, and easily know the error location. It is to provide an automatic translation device capable of doing the above. [Means for Solving the Problem] In order to achieve the above object, the present invention analyzes an input unit for inputting a sentence described in a language to be translated and a syntax of the input sentence according to grammar rules. The syntax analysis unit, an error detection unit that detects an error that occurs when determining the party involved in the parsing processing of the syntax analysis unit, and the error detection unit that is activated to convert the part in which the error occurred into a normal syntax. An error processing unit that temporarily performs a process of associating, determines the related party, and returns the processing to the syntax analysis unit, and the syntax analysis unit determines by the error processing unit when an error occurs when determining the related party. The analysis result is continued by adopting the determined dependency destination, and the analysis result showing the portion of the analysis result that adopts the dependency destination determined by the error processing unit is output, and the syntax analysis In department The obtained analysis results as well as translation in the other party language, part of adopting the dependency destination determined by the error processing unit includes an output unit for instructing the translation in at a given symbol as part of an error. [Operation] In the automatic translation apparatus of the present invention, the input unit inputs a sentence described in a language to be translated, and the syntax analysis unit analyzes the syntax of the input sentence according to grammatical rules, and analyzes the sentence. If an error occurs when determining the destination in processing, the error detection unit detects it and activates the error processing unit, and the error processing unit temporarily associates the error-caused part with the normal syntax. The destination is determined and the process is returned to the parsing unit. Then, the syntactic analysis unit adopts the dependency determined by the error processing unit and continues the analysis process, and outputs the analysis result indicating the portion adopting the dependency determined by the error processing unit in the analysis result. Then, the output unit translates the analysis result obtained by the syntactic analysis unit in the other language, and the part that adopts the party determined by the error processing unit is the part in which the error occurred Give instructions. [Embodiment] Next, an embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing an embodiment of the automatic translation apparatus of the present invention. Referring to FIG. 1, in this embodiment, a sentence described in a language to be translated is input, and an input unit 1 is used to add dictionary information such as a part of speech to each word using a dictionary 2. The syntax analysis unit 3 that analyzes the syntax of the sentence according to the dictionary information and the grammatical rules, the error detection unit 4 that detects the error when the related party in the analysis process of the syntax analysis unit 3 occurs, and the error When it is detected, it is started by the error detection unit 4, and the error processing unit 5 that returns the process to the syntax analysis unit 3 by performing the process of temporarily associating the error occurrence part with the normal syntax and determining the related party, It is composed of an output unit 6 for translating the analysis result of the unit 3 in the other party's language and designating a portion in which an error has occurred in the translated sentence. Then, when the input unit 1 obtains a sentence in the translation target language, the dictionary 2
Is added to add dictionary information to each word, and the result is transmitted to the syntax analysis unit 3. The syntax analysis unit 3 obtains the result of the input unit 1 and analyzes the syntax based on the dictionary information and the grammar rules,
The result is transmitted to the output unit 6. If an error occurs during the analysis, the analysis data and the error information are transmitted to the error detection unit 4. When the error detection unit 4 obtains the error information from the syntax analysis unit 3, the error detection unit 4 transfers the processing from the syntax analysis unit 3 to the error processing unit 5, and transmits the analysis data obtained from the syntax analysis unit 3 to the error processing unit 5. When the error processing unit 5 obtains the analysis data, the error processing unit 5 performs a process of putting together phrases and determining a related party in order to temporarily associate it with a normal syntax, and determines that the syntax analysis unit 3 will follow a normal analysis flow. Parsing part 3 at the time of processing
Then, the analysis data is transmitted to the syntax analysis unit 3. When the parsing unit 3 obtains the parsing data, the parsing unit determined by the error processing unit 5 is used to continue the normal parsing process,
The analysis result indicating the portion of the analysis result that employs the dependency determined by the error processing unit 5 is transmitted to the output unit 6. When the output unit 6 obtains the result of the syntax analysis unit 3, the output unit 6 converts the result into the other language and indicates the portion in which the error has occurred in the translated sentence. Next, the operation of each component will be described using specific example sentences. FIG. 2 shows an example of input and output of each constituent part for two example sentences using English-Japanese translation as an example.
Is the same configuration diagram as FIG. 1, (b) is the first example (example i),
(C) is a second example (example ii). The sentence of the example i is a case where "that" is omitted and is Mary said she could speak Japanese. Note that "that" is omitted between "said" and "she". When this sentence is input to the input unit 1, the dictionary 2 is searched and various dictionary information such as a part of speech is added to each word. Next, the syntax analysis unit 3 analyzes the syntax based on the dictionary information and grammar rules of each word. There are several possible ways of expressing the results of the analysis, but here, the relation between the words is shown by the direction of the arrow, and the word above is expressed by a tree structure indicating that attention is paid. For example, said ↓ actor ↓ Mary means that the actor of "said" is "Mary" and "said" is getting more attention, meaning "Mary said." If the omission of "that" cannot be processed in the processing of the syntactic analysis unit 3 due to the lack of dictionary information of "said", an error will occur when determining the contact person of "could speak". . When the error detector 4 detects the occurrence of this error, it transmits the analysis data up to that point to the error processor 5, and shifts the processing. Since “could speak” is collected as a verb phrase, the error processing unit 5 takes the immediately preceding verb “said” as a reference, with the relation being provisionally modified. As a result, since it can be regarded as a normal syntax, the processing of the newly appearing verb "speak" is returned to the syntax analysis unit 3 and put into the normal analysis processing. The syntactic analysis unit 3 continues the analysis processing, and when the analysis is completed, the result is transmitted to the output unit 6. Here, in the analysis result transmitted to the output unit 6, as shown in FIG. 2 (b), the portion that employs the dependency determined by the error processing unit 5 is shown as (temporary modification). The output unit 6 converts the analysis result into a partner language and outputs it.
At that time, the structure below the temporary modification

Enclose in [...] symbols and clearly indicate in the translated text that an error has occurred. That is, in this case, "Mary said

[... She was able to speak Japanese ...]. Is output. With this, the user can easily determine where the error has occurred by looking at the translated text, and can further obtain a translated text that has been prepared as a text to some extent. The sentence in Example ii is when "should" is omitted, and they require that data be placed in valiabl
e. Note that “should” is omitted between “data” and “be”. In the processing of the syntactic analysis unit 3, if the omission of "should" cannot be processed due to the lack of dictionary information, "be" cannot be the definite verb, so "data" cannot be the subject and therefore the relation of "data". An error occurs when determining the destination. In addition, an error occurs because "that" is not absorbed as a conjunction. Therefore, the error processing unit 5 is activated via the error detection unit 4, the error processing unit 5 determines the related party of “that” and “data” as the temporary modification, and returns the processing to the syntax analysis unit 3. That is, the contact of “data” is “that”
And the reference of "that" is "require". Therefore, as shown in FIG.
Adopt "that" as the contact of "ta" and "require" as the contact of "that", and (temporarily modify) them.
The analysis result instructed in step 1 is transmitted to the output unit 6. Then, after the parsing by the parsing unit 3, the output unit 6 says "they need

〔The invention's effect〕

As described above, in the automatic translation device of the present invention, when an error occurs in the syntactic analysis unit at the time of determining the related party in the analysis process of the input sentence, the error processing unit normally performs the operation in the part in which the error occurs. The syntactical analysis unit adopts the determined modification destination and continues the analysis process, and adopts the modification destination determined by the error processing unit in the analysis result. The output part outputs the analysis result, and the output part translates the analysis result obtained by the syntactic analysis part in the partner language, and the part that adopts the dependency determined by the error processing part is the part where the error occurred. It is an instruction in the translated text with a predetermined symbol as it is, and since the part after the error is translated, it is possible to obtain a translated text that is incomplete but easy to read. There are easily grasp can effect.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of the automatic translation apparatus of the present invention, and FIG. 2 is a view showing a concrete example. In the figure, 1 ... input section 2 ... dictionary 3 ... syntax analysis section 4 ... error detection section 5 ... error processing section 6 ... output section

Claims (1)

[Claims] 1. An input unit for inputting a sentence described in a language to be translated, a syntactic analysis unit for analyzing the syntax of the inputted sentence according to grammatical rules, and a party in the parsing process of the syntactic analysis unit. When an error occurs at the time of determining, an error detection unit that detects the error, is started by the error detection unit, temporarily associates the part in which the error has occurred with a normal syntax, determines an affiliated party, and An error processing unit that returns the processing to the syntax analysis unit, wherein the syntax analysis unit, if an error occurs at the time of determining the engagement destination, adopts the engagement destination determined by the error processing unit and continues the analysis processing. Then, it has a configuration for outputting an analysis result showing a portion that adopts the dependency determined by the error processing unit in the analysis result, and translates the analysis result obtained by the syntax analysis unit in a partner language. Then both , Part employing the dependency destination determined by the error processing unit automatic translation apparatus characterized by comprising an output unit for instructing the translation in at a given symbol as part of an error.