JP6312243B2 - Translation apparatus, translation method, and translation program - Google Patents

Description

  The present invention relates to an apparatus, a method, and a program for translating enumeration data of a large number of noun phrases such as so-called map notes that are character information described on a map.

  Conventionally, various inventions relating to machine translation devices have been made. For example, Patent Document 1 described later describes that even when a proper noun is translated, it can be accurately translated by properly preparing a proper noun dictionary. In addition, the same document also includes a captured word dictionary that stores captured words for each proper noun registered in the proper noun dictionary, so that the proper result is included in the translation result. It is described. This is advantageous in that the translation result can be made easier to understand.

JP 2004-220416 A

  In the case of a conventional machine translation device, various dictionaries are required for translation. The invention described in Patent Document 1 described above also requires a captured word dictionary in addition to a translation dictionary such as proper nouns. Since it takes a certain amount of time to search the dictionary for translation, it takes time to translate a list of noun phrases such as a large number of proper nouns. For example, consider the case of translating map notes as an example of translating a large number of noun phrases.

  There are various maps such as city maps, road maps, regional maps, regional maps, and national maps, and there are a plurality of maps of different scales even with the same type of map. Thus, when all types of maps of different types and scales are targeted, there are tens of millions of map notes. And even if map annotations consisting only of numbers, alphabets, and icons are excluded, those that do not need to be translated are excluded, and even if multiple overlapping map notes are consolidated into one, millions of map notes are subject to translation. Exists.

  It takes several weeks to machine-translate these millions of map notes into multiple languages using a conventional dictionary reference method. In addition, map notes always change when features on the map are lost or newly created, so it is necessary to conduct a field survey at a predetermined timing to update and re-translate.

  As described above, since map annotation needs to be repeatedly translated, there is a conventional request to translate as fast as possible. This also occurs when translating a large number of noun phrases such as place names, names, names, addresses, etc., not only in map notes but also in address lists, phone book lists, customer lists, etc. It is a problem.

  In view of the above, an object of the present invention is to make it possible to translate a large amount of noun phrase enumeration data much faster and more accurately than conventional machine translation devices.

In order to solve the above problem, a translation device according to claim 1 is provided:
Select and extract the dictionary to be used from multiple translation dictionaries with different types of noun phrases to be searched according to the language to be translated. A dictionary creation means for creating a dictionary for translation of the dictionary contents;
Storage that stores reading-kana-added data having one or more noun phrases to be translated and the reading-kana of each of the one or more noun phrases in a variable that is distinguishable from other reading-kana-giving data. Means,
The dictionary data of the dictionary for translation created by the dictionary creating means is read one by one, and the noun phrase to be translated of the data with the kana added stored in the variable is converted to the dictionary data of the dictionary for translation. And a full-text replacement means for replacing the full-text into the target language.

  According to the translation device of the first aspect of the present invention, the dictionary creating means creates the translation dictionary according to the language to be translated. The storage means stores the reading-kana-added data in one variable in a manner distinguishable from other reading-kana-giving data. The full-text replacement means reads the dictionary data of the dictionary for translation created by the dictionary creation means one by one, and the noun phrase of the reading-kana-given data stored in a predetermined variable by the storage means Full-text replacement with the target language using dictionary data.

  As a result, it is possible to store the translation-ready kana-given data in a predetermined variable and collectively process the stored many noun phrases of the read-kana-given data. Then, the dictionary data of the dictionary for translation is read one by one, and the noun phrase of the read-named kana-given data to be translated is replaced with the target language by using the dictionary data having the same reading-kana. Can translate noun phrases. In this case, the processing time required for translation can be greatly shortened as compared with the case where the translation dictionary is referred to based on each reading kana.

  According to the present invention, enumerated data of a large number of noun phrases can be translated significantly faster and more accurately than conventional machine translation devices.

It is a block diagram for demonstrating the example of a structure of the translation apparatus of embodiment, and the outline | summary of operation | movement. It is a figure for demonstrating the example of the storage data (dictionary data) of the translation object data file. It is a figure for demonstrating the example of the stored data of the reading pseudonymized data file 113. FIG. It is a block diagram for demonstrating the process of the use dictionary creation part. It is a figure for demonstrating the 1st process performed in the full text replacement process part. It is a figure which shows an example of the storage process program of the reading pseudonym to the variable performed in the full text replacement process part. It is a figure which shows the example of the state by which the reading pseudonym of each reading pseudonym assignment | providing data was stored in variable $ jp. It is a figure for demonstrating the 2nd process performed in the full text replacement process part. It is a figure which shows an example of the program which substitutes the full text to the target language for the reading pseudonym stored in the variable $ jp performed in the full text replacement processing part. It is a figure for demonstrating the example of the use dictionary data for translation prepared in the use dictionary file for translation. It is a figure which shows the state at the time of carrying out the full-text replacement process for the storage data of variable $ jp with the dictionary shown in FIG. It is a figure for demonstrating the example of the stored data of the translated data file. It is a flowchart for demonstrating the process which provides a reading pseudonym with respect to the translation object data performed with the translation apparatus of embodiment. It is a flowchart for demonstrating the translation process of the map note performed with the translation apparatus of embodiment.

  Hereinafter, embodiments of the apparatus, method, and program of the present invention will be described with reference to the drawings. The present invention can be applied to translation of enumerated data of a large number of noun phrases such as place names, names, names, and addresses described in map notes, telephone directory lists, customer lists, and the like. The term “noun phrase” means a phrase indicating a specific object such as an object, a substance, a person, or a place. In this specification, as a specific example of “a list of noun phrases” to be translated, for example, a place name, a name, a name, an address described in a map note, a phone book list, a customer list, etc. Yes.

  In the following, a specific description will be given by taking as an example the case of translating a map note. Map annotations exist in large quantities as described above, and are frequently updated. Therefore, map annotations must be re-translated each time they are updated. In recent years, the number of tourists and business people visiting Japan from abroad has increased, and the latest maps for foreign languages, which are translated from the map annotations of the latest maps into the specified foreign languages, can be quickly displayed. It is hoped to provide. Moreover, foreign languages are required to correspond to a plurality of languages such as Chinese (simplified), Chinese (traditional), English, and Korean.

  Under such a background, the apparatus, method, and program according to the embodiments described below are capable of converting a large amount of map annotations significantly faster and more appropriately than a conventional machine translation apparatus, and It has been translated into multiple languages.

[Configuration example of big data translation device]
First, a configuration example and an outline of operation of a big data translation apparatus (hereinafter simply referred to as a translation apparatus) of this embodiment will be described. FIG. 1 is a block diagram for explaining a configuration example and an outline of operation of the translation apparatus according to this embodiment. In FIG. 1, the letter “F” in the description of the translation target data F111 or the like is used as an abbreviation for a file.

  In the translation apparatus shown in FIG. 1, the control unit 101 realizes a function for controlling each unit of the translation apparatus, the storage device 102 realizes an information storage holding function, and the operation unit 103 realizes a user interface function.

  The reading kana dictionary group 104 includes a plurality of reading kana dictionaries such as a proper noun reading kana dictionary, a general noun reading kana dictionary, a prefix reading kana dictionary, a suffix reading kana dictionary, and a katakana reading kana dictionary, and a Japanese map. Provides reading kana for note data. Although not shown, the prefix reading kana dictionary includes a proper noun prefix reading kana dictionary, a general noun prefix reading kana dictionary, and a katakana prefix reading kana dictionary. Similarly, the suffix reading kana dictionary includes a proper noun suffix reading kana dictionary, a general noun suffix reading kana dictionary, and a katakana suffix reading kana dictionary.

  Each of the proper noun reading kana dictionary, the general noun reading kana dictionary, and the katakana reading kana dictionary can replace the reading kana by partial matching. Each prefix reading kana dictionary provides the reading kana of the prefix, so that the reading kana can be replaced by front matching. Each suffix reading kana dictionary provides a reading kana for the suffix, so that the reading kana can be replaced by backward matching.

  In this embodiment, the language-specific translation dictionary group 105 includes language-specific translation dictionaries for translating reading kana of Japanese map note data into a target foreign language. In this embodiment, a four-language translation dictionary is provided to support four languages: Chinese (simplified), Chinese (traditional), English, and Korean. Furthermore, the translation dictionaries for each language are roughly divided into a dictionary for general nouns and a dictionary for katakana. The dictionary for general nouns consists of a general noun suffix dictionary, a general noun prefix dictionary, and a general noun dictionary. The dictionary for Katakana is Katakana proper noun suffix dictionary, Katakana proper noun prefix dictionary, Katakana proper noun dictionary, Katakana general noun suffix dictionary, Katakana general noun prefix dictionary, Katakana general noun dictionary Consists of.

  Each of the general noun dictionary, the katakana proper noun dictionary, and the katakana general noun dictionary can replace translation data (translation words) corresponding to reading kana and the like by partial matching. Each prefix dictionary provides translation data corresponding to the prefix, so that prefix translation data can be replaced by prefix matching. Also, each suffix dictionary provides suffix translation data, so that suffix translation data can be replaced by backward matching.

  The translation object data file 111 holds a large number of map note data to be translated. FIG. 2 is a diagram for explaining an example of data stored in the translation object data file 111. As shown in FIG. 2, the translation target data file 111 holds a large number of translation target data composed of “area code, attribute number, map annotation data”. The translation target data file 111 stores, for example, translation target data provided from an external device through an external interface (not shown).

  The reading kana giving unit 112 reads the kanji portion included in each of a large number of map annotation data stored in the translation object data file 111 by using the reading kana dictionary of the reading kana dictionary group 104. Convert to The reading kana giving unit 112 forms reading kana-given data including the converted reading kana and records it in the reading kana-given data file 113. The reading-kana-added data file 113 is created on a large-capacity recording medium such as a hard disk, and stores reading-kana-added data formed by the reading-kana adding unit 112.

  FIG. 3 is a diagram for explaining an example of data stored in the read-named data file 113. As shown in FIG. 3, the reading-kana-added data stored in the reading-kana-added data file 113 includes “area code, attribute number, map note data, and reading-kana”. In addition, if the dictionary which acquired the reading pseudonym differs, they will be shown separately. That is, in FIG. 3, “Akishijima” is obtained by reading the kana of the “Evilstone” part from the proper noun dictionary, for example, and the kana of the “island” part from the general noun suffix dictionary. Like “Akaseki Jima”, “Akaseki” and “Jima” are stored in the kana-given data in such a manner that it can be seen that they are obtained from different dictionaries.

  In this embodiment, the use dictionary creation unit 114 creates a translation use dictionary to be used for translation in the translation use dictionary file 115 according to the language to be translated (in which language it is translated). Specifically, the use dictionary creation unit 114 selects and extracts a dictionary to be used from the translation dictionaries in the language-specific translation dictionary group 105 according to the translation target language. Then, the extracted dictionaries are arranged and combined, and the translation use dictionary in which the contents of the dictionaries (dictionary data) are arranged in the order of the longest readings of the map notes is used. To create.

  In addition, information indicating whether the search word can be replaced by partial match, forward match, or backward match is added to each dictionary data according to the dictionary to which the dictionary data belongs by the use dictionary creation unit 114. Is done. For example, information used for partial matching is added to the dictionary data of the general noun dictionary, the Katakana proper noun dictionary, and the Katakana general noun dictionary. In addition, information that is used by forward matching is added to the general noun prefix dictionary, the katakana proper noun prefix dictionary, and the katakana general noun prefix dictionary. Further, information used for backward matching is added to the dictionary data of the general noun suffix dictionary, the katakana proper noun suffix dictionary, and the katakana general noun suffix dictionary.

  The full-text replacement processing unit 116 includes a large-capacity memory 116M and performs the following two processes. The first process reads the reading-kana-given data in the reading-kana-given data file 113, and stores all the reading-kana-given data in one variable while adding a separator to each reading-kana-given data. I do. In the second process, the dictionary data in the translation use dictionary file 115 created by the use dictionary creation unit 114 is read one by one, and the reading pseudonym of the reading-kana-given data stored in the variables is converted into the translation data of the dictionary data. Perform the replacement process.

  As a result, the full-text replacement processing unit 116 forms translated data including translation data for the map annotation data of the reading-kana-added data, and records this in the translated data file 117. The translated data file 117 is created on a large-capacity recording medium such as a hard disk, and stores translated data formed by the full-text replacement processing unit 116.

  As described above, the translation apparatus according to this embodiment stores the full-reading kana-given data in the reading-kana-given data file 113 in a predetermined variable by the function of the full-text replacement processing unit 116, thereby giving the full-reading kana The stored data is stored in the memory 116M and is collectively processed. Then, processing is performed for reading the prepared dictionary data in the translation use dictionary file 115 one by one and assigning the translation data in the translation use dictionary file 115 to all the read-named data on the memory 116M. Do. In other words, all the reading pseudonyms of all the reading pseudonym-added data on the memory 116M are replaced by the full text with the translation data of the translation use dictionary file 115. This enables ultra-high-speed translation of many existing map notes.

  Hereinafter, the details of the processing performed in the use dictionary creation unit 114 and the processing performed in the full-text replacement processing unit 116, which are the main parts of the translation apparatus of this embodiment, will be described.

[Process of Use Dictionary Creation Unit 114]
FIG. 4 is a block diagram for explaining the processing of the use dictionary creation unit 114. As described above and as shown in FIG. 4, the language-specific translation dictionary group 105 includes, for example, a general noun suffix dictionary, a general noun prefix dictionary, a general noun dictionary, a katakana proper noun suffix dictionary for each language. , A Katakana proper noun prefix dictionary, a Katakana proper noun dictionary, a Katakana general noun suffix dictionary, a Katakana general noun prefix dictionary, and a Katakana general noun dictionary are prepared.

  Therefore, the use dictionary creation unit 114 selects and extracts a dictionary to be used from the translation dictionaries in the language-specific translation dictionary group 105 according to the translation target language. Then, the use order is determined and the dictionaries are rearranged and combined. Further, the translation use dictionary in which the contents of the dictionaries (dictionary data) are arranged in the order of the longest readings of the map notes are stored in the translation use dictionary file 115. create. The reason why the map annotations are arranged in the order of the longest reading kana is to replace the noun phrases in the data having been given the reading kana by the longest match method. In addition, as described above, information indicating whether a search word can be replaced by partial match, forward match, or backward match is added to each dictionary data according to the dictionary to which the dictionary data belongs. The

  Further, the selection and arrangement order of dictionaries are performed according to the priority and necessity of each language to be translated. An example of dictionary selection and arrangement order will be described. Korean is a phonetic character, and Japanese expressed in katakana can be replaced with a Korean (Hangul character) corresponding to the sound. For this reason, when the language to be translated is Korean, the use dictionary creation unit 114 does not use a katakana dictionary, but a general noun suffix dictionary, a general noun prefix dictionary, a general noun dictionary for translation into Korean. Select as your dictionary.

  Then, the use dictionary creation unit 114 translates the suffix and prefix as the selected dictionary order, for example, and translates the remaining general nouns by (1) General noun suffix dictionary → ( 2) General noun prefix dictionary → (3) General noun dictionary arranged in order. As described above, each dictionary is arranged in the order of long reading kana corresponding to the map notes. In this manner, a translation use dictionary for translating map notes into Korean is formed in the translation use dictionary file 115.

  In addition, Japanese common nouns expressed in Chinese characters are often expressed in the same Chinese characters in Chinese. For this reason, when the language to be translated is Chinese, all the nine dictionaries described above are used. However, in order to prioritize the translation of Katakana, the dictionary related to Katakana is used first, and then the general noun is used. Rearrange the used dictionaries to use dictionaries and combine them. As described above, each dictionary is arranged in the order of long reading kana corresponding to the map notes. In this manner, a translation use dictionary for translating map notes into Chinese is formed in the translation use dictionary file 115.

  In addition, when the translation target language is English, it is a completely different language from Japanese, so all nine of the above-mentioned dictionaries are used, and the order of use is first a dictionary for general nouns and then a dictionary for katakana. Can be combined side by side. Even in this case, the respective dictionaries are arranged in the order of the long reading kana corresponding to the map notes as described above. In this manner, a translation use dictionary for translating map notes into English is formed in the translation use dictionary file 115.

  Note that the selection and arrangement order of dictionaries according to the priority and necessity for each translation target language described here are merely examples, and various other modes can be used. Needless to say, if there are dictionaries to be used other than the nine dictionaries described above, these dictionaries are also considered. In addition, there may be a dictionary specially provided for each translation target language. In that case, when the translation target language is selected, the dictionary is also selected as a dictionary to be used. Become.

[Process of Full Text Replacement Processing Unit 116]
[First processing (storage processing to variable)]
Each of FIGS. 5, 6, and 7 relates to a process of storing all reading-kana-given data in the reading-kana-given data file 113 as a first process performed by the full-text replacement processing unit 116 in one variable. It is a figure for demonstrating.

  Among these, FIG. 5 is a figure for demonstrating the said 1st process performed in the full text replacement process part 116. FIG. As shown in FIG. 5, the first process performed by the full-text replacement processing unit 116 reads the reading-kana-given data in the reading-kana-given data file 113 and attaches a separator to each reading-kana-given data. Then, the fully read pseudonymized data is stored in one variable. As a result, all of the reading-kana-added data in the reading-kana-added data file 113 is expanded on the memory 116M of the full-text replacement processing unit 116.

  FIG. 6 is a diagram illustrating an example of a program for storing the reading-kana-given data that is executed in the full-text replacement processing unit 116. As described with reference to FIG. 3, the program shown in FIG. 6 separates each reading-kana-assigned data composed of “area code, attribute number, map note, and reading-kana” from other reading-kana-giving data. Stored in variable $ head [$ i] in a possible manner. Furthermore, only each reading pseudonym of the reading-kana-given data is stored in the variable $ jp. Since the reading-kana-added data stored in these variables and the reading-kana of the reading-kana-added data are stored in each variable in the same order, the corresponding relationship is maintained.

  FIG. 7 is a diagram showing a state where the reading pseudonym of each reading pseudonym-assigned data of the reading pseudonym assigned data file 113 shown in FIG. 3 is stored in the variable $ jp. By executing the program shown in FIG. 6, only the reading kana of each reading kana-given data is stored in the variable $ jp in a manner separable from other reading kana. As a result, each reading-kana of each reading-kana-given data is stored on the memory 116M and can be processed at once.

[Second processing (full-text replacement processing)]
8, 9, 10, and 11, as described above, the variable $ Jp is set in the variable $ Jp using the dictionary data of the translation use dictionary file 115, which is the second process performed by the full-text replacement processing unit 116. It is a figure for demonstrating the process which replaces the stored reading pseudonym with the target language.

  Among these, FIG. 8 is a figure for demonstrating the said 2nd process performed in the full-text replacement process part 116. FIG. As shown in FIG. 8, the second process performed by the full-text replacement processing unit 116 reads the dictionary data of the translation use dictionary file 115 one by one and stores the reading-kana-added data stored in the variable $ jp. Processing to replace the reading pseudonym with the translation data of the dictionary data is performed. Then, the kana-given data stored in the variable $ head [$ i] is associated with the translation data in which the variable $ jp is replaced with this to form translated data, which is converted into the translated data file 117. To record. Thereby, the map note to be translated can be translated into the target language to be translated.

  FIG. 9 is a diagram showing an example of a program that replaces the full text with the target language of the reading pseudonym stored in the variable $ Jp, which is executed by the full text replacement processing unit 116. The program shown in FIG. 9 performs replacement processing by backward matching using dictionary data of a dictionary used by backward matching. In addition, the program uses the dictionary data of the dictionary used for forward matching to perform replacement processing by forward matching. Similarly, the program performs replacement processing by partial matching using dictionary data of a dictionary used by partial matching.

  Then, after the replacement process is completed, the program shown in FIG. 9 associates the read-named data stored in the variable $ head [$ i] with the translation data in which the variable $ jp is replaced. Form translated data and output it. Actually, the formed translated data is recorded in the translated data file 117. In this way, the map note to be translated is translated into the target language to be translated.

  Here, as shown in FIG. 7, the second sentence performed by the full-text replacement processing unit 116, taking as an example the case where the reading kana corresponding to the map note to be translated stored in the variable $ jp is translated into English. The full-text replacement process, which is the above process, will be specifically described.

  FIG. 10 illustrates an example of the translation use dictionary data prepared in the translation use dictionary file 115 when the kana corresponding to the map note to be translated stored in the variable $ jp is translated into English. FIG. Here, in order to simplify the description, it is assumed that five dictionaries A to E are selected and extracted as dictionaries to be used and arranged in the order of dictionaries A to E, as shown in FIG. It is assumed that only one dictionary data exists in each of the dictionaries A to E.

  FIG. 11 is a diagram showing a state when the full text replacement processing is performed on the stored data of the variable $ jp using the dictionary shown in FIG. FIG. 11 (A) is the same as that shown in FIG. 7, and in the variable $ jp, three readings of map annotation data “Akase Jima”, “Hizue Jima”, “Owaku Kahan” "Is stored.

  In this example, the full text replacement processing unit 116 first reads the dictionary data of the dictionary A from the translation use dictionary file 115, and performs full text replacement of the stored data of the variable $ jp by partial matching. That is, full-text replacement processing unit 116 issues an instruction to replace “Wobo” with “Wobou” according to the dictionary data of dictionary A. Here, “Wow” means that “Wow” in any position of the noun phrase is to be replaced. In this case, as shown in FIG. 11 (B), the reading name “Wobo” of the third noun phrase is replaced with the translation data “Wobo” of the dictionary data of the dictionary A.

  In this example, in the command issued by the full-text replacement processing unit 116, the distinction between partial match, forward match, and backward match can be determined by the position of “, (comma)” added to the reading target kana. That is, when the replacement is instructed by the front matching, “, BACK” is indicated, and when the replacement is instructed by the backward matching, the “OUT,” is indicated. Further, when the replacement is instructed by partial matching, “NO” is given and “, (comma)” is not added.

  Next, the full-text replacement processing unit 116 reads the dictionary data of the dictionary B from the translation use dictionary file 115 and performs full-text replacement of the stored data of the variable $ jp by forward matching. That is, the full-text replacement processing unit 116 issues an instruction to replace “, Akuseki” with “Akuseki” based on the dictionary data of the dictionary B. Here, “, akuseki” means that “akuseki” in front of the noun phrase is to be replaced. In this case, as shown in FIG. 11C, “akuseki” located in front of (in front of) the reading of the first noun phrase is replaced with translation data “Akuseki” of the dictionary data of the dictionary B. .

  Next, the full text replacement processing unit 116 reads the dictionary data of the dictionary C from the translation use dictionary file 115 and performs full text replacement of the stored data of the variable $ jp by backward matching. That is, the full-text replacement processing unit 116 issues an instruction to replace “KAHAN,” with “River side” based on the dictionary data of the dictionary C. Here, “Kahan,” means that “Kahan” behind the noun phrase is to be replaced. In this case, as shown in FIG. 11 (D), “Kahan” located behind (behind) the reading noun of the third noun phrase becomes the translation data “Reverse Side” of the dictionary data of the dictionary C. Replaced.

  Next, the full text replacement processing unit 116 reads the dictionary data of the dictionary D from the translation use dictionary file 115 and performs full text replacement of the stored data of the variable $ jp by forward matching. That is, the full-text replacement processing unit 116 issues an instruction to replace “, haze” with “Hizue” based on the dictionary data of the dictionary D. Here, “, hizue” means that “hizue” in front of the noun phrase is to be replaced. In this case, as shown in FIG. 11E, “hizue” located in front of (in front of) the reading of the second noun phrase is replaced with the translation data “Hizue” of the dictionary data of the dictionary D. .

  Finally, the full text replacement processing unit 116 reads the dictionary data of the dictionary E from the translation use dictionary file 115 and performs full text replacement of the stored data of the variable $ jp by backward matching. That is, the full-text replacement processing unit 116 issues an instruction to replace “Jima,” with “Island” based on the dictionary data of the dictionary E. Here, “Jima,” means that “Jima” behind the noun phrase is to be replaced. In this case, as shown in FIG. 11 (F), “Jima” located at the back side (back) of the first and second noun phrases is “translated data“ of the dictionary data of the dictionary E “ Island ”. In this way, full-text replacement is performed in which the reading pseudonym in the variable $ jp is replaced with the translation data by using the dictionary data of the translation usage dictionary created in the translation usage dictionary file 115.

  Then, as described above, the full-text replacement processing unit 116 translates the kana-given data stored in the variable $ head [$ i] in association with the translation data in which the variable $ jp is replaced. Complete data is recorded and recorded in the translated data file 117. FIG. 12 is a diagram for explaining an example of data stored in the translated data file 117. As shown in FIG. 12, the translated data that is stored data of the translated data file 117 is composed of “area code, attribute number, map note, reading kana, translation data”. In this way, translated data obtained by adding corresponding translation data to each translation target data in the translation target data file 111 is obtained in the translated data file 117.

  In the example shown in FIG. 11, since there are only two words constituting the noun phrase, the partial match example and the forward match example are the same. However, the original partial match replacement has, for example, translation target data “XX movie village”, and it is instructed to translate with dictionary data “substitution method = partial match, translation target = movie, translation data = movie”. Suppose that In this case, an instruction to replace “movie” with “movie” is issued. By this command, “movie movie village” is replaced with “movie movie village”, and “movie” is replaced with “movie” in the original partial match replacement.

[Summary of Translation Processing in Translation Device of Embodiment]
Next, the map annotation translation processing performed by the translation apparatus of this embodiment will be summarized with reference to the flowcharts of FIGS. The processing performed in the translation apparatus of this embodiment can be broadly divided into reading kana assignment processing for map annotation data that is data to be translated and translation processing for map annotation data to which reading kana is given. it can. Since the map note data to which the reading pseudonym is assigned can be used in common when the map note data is translated into various languages, it may be performed only once unless the map note data is changed. On the other hand, the map annotation translation process is a process that needs to be executed separately if the language to be translated is different.

[Yomikana assignment process]
FIG. 13 is a flowchart for explaining a process of assigning a reading pseudonym to translation target data performed by the translation apparatus of this embodiment. The processing shown in the flowchart of FIG. 13 is executed by the control unit 101 of the translation apparatus, and is performed by the control unit 101 controlling the reading pseudonym assignment unit 112.

  When the control unit 101 of the translation apparatus receives an instruction to execute the reading pseudonym assignment process from the user of the translation apparatus through the operation unit 103, the control unit 101 executes the translation process illustrated in FIG. In this case, the control unit 101 controls the reading kana giving unit 112 to use each dictionary of the reading kana dictionary group 104 for the map annotation of each translation target data stored in the translation target data file 111. Processing for assigning a reading pseudonym is performed (step S101).

  In the process of step S101, the reading pseudonym assigning unit 112 assigns a reading pseudonym for the map note to each translation target data in the translation target data file 111 shown in FIG. 2, and the reading pseudonym shown in FIG. Create granted data. The created reading-kana-added data is stored in the reading-kana-given data file 113.

  In this way, when the reading pseudonymized data corresponding to all the translation target data in the translation target data file 111 is stored in the reading pseudonymized data file 113, the processing shown in FIG.

[Translation processing of map notes]
FIG. 14 is a flowchart for explaining a map annotation translation process performed by the translation apparatus of this embodiment. The processing shown in the flowchart of FIG. 14 is executed by the control unit 101 of the translation apparatus, and is performed by the control unit 101 controlling the use dictionary creation unit 114 and the full-text replacement processing unit 116.

  When receiving a translation processing execution instruction from the user of the translation device through the operation unit 103, the control unit 101 of the translation device executes the translation processing illustrated in FIG. First, the control unit 101 receives a selection input of a language to be translated from the user through the operation unit 103 (step S201).

  Next, the control unit 101 controls the use dictionary creation unit 114 to perform a process of creating a translation use dictionary corresponding to the selected translation target language in the translation use dictionary file 115 (step S202). In the process of step S202, as described above, the use dictionary creation unit 114 selects and extracts a dictionary to be used for translation from the language-specific translation dictionary group 105 according to the translation target language. Then, the use dictionary creation unit 114 rearranges the extracted dictionaries in the order of use, and creates a use dictionary for translation in the use dictionary file 115 for translation in which dictionary data is arranged in the order of the longest readings of map notes. To do.

  Next, the control unit 101 controls the full-text replacement processing unit 116 to perform processing for storing the reading-kana-added data stored in the reading-kana-added data file 113 in a predetermined variable (step S203). . In the process of step S203, as described with reference to FIG. 5 to FIG. 7, the read-named data is stored in the variable $ head [$ i], and the read-named data of the read-named data is stored in the variable $ jp. It is processing.

  Thereafter, the control unit 101 controls the full text replacement processing unit 116 to read the dictionary data in the translation use dictionary file 115 one by one, and replaces the reading pseudonym stored in the variable $ jp with the translation data of the dictionary data. The process which performs is performed (step S204). The processing in step S204 is the processing described with reference to FIGS. 8 to 11, and the reading kana of the map note stored in the variable $ jp is used for each translation stored in the translation usage dictionary file 115. Full text is replaced with translation data of dictionary data.

  The control unit 101 further controls the full-text replacement processing unit 116, and the full-text replacement processing unit 116 creates translated data and records it in the translated data file 117 (step S205). The translated data created in step S205 is formed by associating the reading-kana-added data stored in the variable $ head [$ i] with the translated data in which the variable $ jp is replaced with this. It is. After the processing in step S205, the translation processing shown in FIG. 14 ends. As a result, the translated data in the form shown in FIG. 12 is accumulated in the translated data file 117.

  As described above, the translated data in the example shown in FIG. 12 is a map note translated into English. Therefore, in the translation process shown in FIG. 14, by selecting the target language as the translation target language in step S201, translated data translated into the target language for the map annotation can be created.

[Special handling for translation into Chinese]
In the embodiment described above, it is assumed that translation is performed into four languages of Chinese (simplified), Chinese (traditional), English, and Korean. The present invention can be applied to any of the four languages. However, when translating into Chinese (simplified) and Chinese (traditional), the kanji part of the map note in Japanese can be translated by single kanji conversion without using the reading kana given to it. Is possible.

  That is, there are many Japanese kanji characters that can be used in common with Chinese (simplified) and Chinese (traditional). Usually, the corresponding Chinese simplified characters, Chinese Traditional characters are determined. For this reason, the kanji portion of the Japanese map note is translated by single kanji conversion without using the reading kana. For simplicity, Japanese Kanji characters may be used as they are, or Simplified Chinese characters and Traditional Chinese characters may be identified from Japanese Kanji characters and used as translation results. . Note that Japanese Katakana and Hiragana can be translated by the above-described invention.

[Effect of the embodiment]
According to the translation apparatus of the embodiment described above, a large amount of map notes can be translated significantly faster and more accurately than conventional machine translation apparatuses.

  Moreover, since the reading kana is given to the map note, and the given reading kana is translated, the translation according to the actual reading can be performed. There is no unnatural translation.

  And when translating millions of map notes, it used to take weeks to translate, but it could be reduced to hours.

[Use of area code and attribute number]
In the above-described embodiment, as described with reference to FIG. 3, each storage data of the reading pseudonymized data file 113 is assigned an area code and an attribute number. For this reason, translation variables are prepared for each area and attribute. Then, the reading-kana-assigned data with the reading-kana added to the data to be translated is distributed to each variable and stored, and by combining them, one variable storing the reading-kana-added data by area and attribute Can be formed. In this case, the area code and the attribute number serve as header information of each reading-kana given data.

  In the above-described embodiment, the case where each translation dictionary in the language-specific translation dictionary group has no area code or attribute number has been described as an example. However, the present invention is not limited to this. In practice, dictionary data is also organized by area and attribute. That is, the area code and the attribute number are assigned to each dictionary data included in each translation dictionary in the language-specific translation dictionary group as in the case of the translation target data. This is because even a map note made up of the same characters may be read differently if the area is different, and a map note made up of the same characters may be read differently if the attributes are different. As an example of the former, the description of “Shinjuku” includes a place name that reads “Shinjuku” and a place name that reads “Arajuku”. As an example of the latter, there is an example in which “Asakusa” is read as the place name, and “Senso” is read as the temple name, even though “Asakusa” is described.

  Therefore, the use dictionary creation unit 114 constructs dictionary data including an area code and an attribute number included in each dictionary data of each translation dictionary when creating the translation use dictionary. The dictionary data of each dictionary is arranged in order of area code, in order of attribute number, and in descending order of the number of characters in the reading kana. Then, only when the area code, the attribute number, and the reading kana match, the reading kana of the reading kana given data is replaced with the translation data of the dictionary data in the translation use dictionary file 115. By doing in this way, the appropriate translation result according to the appropriate reading pseudonym can be reliably provided also to the map notation of the same notation with different readings and kana depending on an area and an attribute.

  In addition, only when the area code and the reading kana match, the reading kana of the data to which the reading kana has been added can be replaced with the translation data of the dictionary data in the translation use dictionary file 115. Similarly, only when the attribute number and the reading kana match, the reading kana of the reading-kana-given data can be replaced with the translation data of the dictionary data in the translation use dictionary file 115. In the former case, for example, the present invention can be applied when translating data that does not have an attribute number such as an address to be translated. The latter case can be applied when the area to which the data to be translated belongs is limited to a single area, such as when a map note of a single area is translated.

[Modifications, etc.]
In the above-described embodiment, the case where the map annotation data is translated has been described as an example. However, the present invention is not limited to this. For example, the present invention can be applied to a case where a large number of noun phrases such as a phone book list, a customer list, and an address list are included, but a large number of noun phrases are translated.

  In the above-described embodiment, the language to be translated when the map annotation data is translated is, for example, capable of supporting four languages: Chinese (simplified), Chinese (traditional), English, and Korean. Although explained, it is not limited to this. A list of noun phrases such as map notes can be translated into various languages simply by preparing a translation dictionary for the target language.

  Moreover, since the reading-kana-added data formed by adding the reading-kana can be used in common when translated into each language, the process of giving the reading-kana to the translation target data shown in FIG. As long as there are no changes to the map notes, this only needs to be done once.

  In the above-described embodiment, the use dictionary creation unit 114 selects and extracts necessary dictionaries from the language-specific translation dictionary group 105 according to the language to be translated, and arranges the arrangement order thereof. The description has been made assuming that the use dictionary is created in the use dictionary file 115 for translation. However, it is not limited to this. Of course, the necessary dictionaries may be read sequentially in the required order. In this case, the use dictionary creation unit 114 instructs the full text replacement processing unit 116 to use the dictionary and the order of use.

  In addition, the map annotation data shown in FIG. 2 and the like is shown as a simple example, but in reality, kanji, katakana, hiragana, and alphanumeric symbols may be mixed. For this reason, kanji and katakana common nouns are translated in the manner described above. In addition, for kanji proper nouns, their kana are converted to Roman characters. If katakana is translated into Korean, it is replaced with Korean according to the sound. Use alphanumeric characters as they are without doing anything. Thus, a hybrid translation is performed.

[Others]
As can be seen from the description of the above-described embodiment, the translation dictionary of the translation device described in the claims corresponds to each translation dictionary of the language-specific translation dictionary group of the translation device of the embodiment. The dictionary creation means of the translation device described corresponds to the use dictionary creation unit 114 of the translation device of the embodiment. Moreover, the reading-kana dictionary of the translation apparatus described in the claim corresponds to each reading-kana dictionary of the reading-kana dictionary group 104 of the translation apparatus according to the embodiment. This corresponds to the reading pseudonym assigning unit 112 of the translation apparatus according to the embodiment.

  Each of the storage unit and the full text replacement unit of the translation device described in the claims corresponds to the full text replacement processing unit 116 of the translation device of the embodiment.

  The translation process described with reference to FIG. 14 is the one to which an embodiment of the translation method of the present invention is applied. The program for executing the translation process described with reference to FIG. An embodiment of the program is applied. 1 may be realized as a function of the control unit 101 by a program executed by the control unit 101. The reading-kana adding unit 112, the use dictionary creation unit 114, and the full-text replacement processing unit 116 illustrated in FIG. .

  DESCRIPTION OF SYMBOLS 101 ... Control part, 102 ... Memory | storage device, 103 ... Operation part, 104 ... Reading kana dictionary group, 105 ... Translation dictionary group according to language, 111 ... Use dictionary creation part, 112 ... Use dictionary file for translation, 113 ... Translation object data File: 114 ... Yomikana adding unit, 115 ... Yomikana-added data file, 116 ... Full text replacement processing unit, 116M ... Memory, 117 ... Translated data file

Claims (7)

Select and extract the dictionary to be used from multiple translation dictionaries with different types of noun phrases to be searched according to the language to be translated. A dictionary creation means for creating a dictionary for translation of the dictionary contents;
Storage that stores reading-kana-added data having one or more noun phrases to be translated and the reading-kana of each of the one or more noun phrases in a variable that is distinguishable from other reading-kana-giving data. Means,
The dictionary data of the dictionary for translation created by the dictionary creating means is read one by one, and the noun phrase to be translated of the data with the kana added stored in the variable is converted to the dictionary data of the dictionary for translation. A translation device comprising: full-text replacement means for replacing the full-text with a target language using The translation device according to claim 1,
The translation apparatus, wherein the plurality of translation dictionaries include a plurality of dictionaries having different search word matching methods for front matching, backward matching, and partial matching. A translation device according to claim 1 or 2, wherein
Attribute information indicating the attribute of the noun phrase included in the reading-kana-given data is added to the reading-kana-giving data stored in the variable,
The dictionary data of the translation dictionary is associated with attribute information indicating the attribute of the noun phrase included in the dictionary data,
The full-text replacing means converts the noun phrase of the reading-kana-given data stored in the variable into a target language using dictionary data of the translation dictionary having the same attribute information as the reading-kana-given data. A translation device characterized by replacing. A translation device according to claim 1 or 2, wherein
Area information indicating a region to which a location relating to the noun phrase included in the reading pseudonymized data belongs is added to the reading pseudonymized data stored in the variable,
Area information indicating a region to which a place relating to the noun phrase included in the dictionary data belongs is added to the dictionary data of the translation dictionary,
The full-text replacement means converts the noun phrase of the reading-kana-given data stored in the variable into a target language using dictionary data of the translation dictionary having the same area information as the reading-kana-given data. A translation device characterized by replacing. A translation device according to claim 1 or 2, wherein
The phonetic affixed data stored in the variable includes attribute information indicating attributes of the noun phrase included in the phonetic affixed data, and a place related to the noun phrase included in the phonetic affixed data. Area information indicating the region to which it belongs,
In the dictionary data of the translation dictionary, attribute information indicating the attribute of the noun phrase included in the dictionary data, and area information indicating a region to which a place related to the noun phrase included in the dictionary data belongs, are added.
The full-text replacement means uses the dictionary data of the translation dictionary having the same attribute information and area information as the reading-kana-given data for the noun phrase of the reading-kana-giving data stored in the variable. A translation device characterized in that it is replaced with a language. The dictionary creation means selects and extracts dictionaries to be used from multiple translation dictionaries with different types of noun phrases to be searched according to the language to be translated, determines the order of use, rearranges them, and reads noun phrases A dictionary creation step for creating a dictionary for translation in which the dictionary contents are arranged in the order of long kana;
One variable in a manner in which the storage means can distinguish the reading-kana-added data having one or more noun phrases to be translated and the respective reading-kana of the one or more noun phrases from other reading-kana-giving data. Storing process to store in,
Full-text replacement means reads the dictionary data of the dictionary for translation created by the dictionary creation means one by one, and translates the noun phrase to be translated of the reading-kana given data stored in the variable And a full-text replacing step of replacing the full-text with a target language using the dictionary data of the dictionary. The computer installed in the translation device
Select and extract the dictionary to be used from multiple translation dictionaries with different types of noun phrases to be searched according to the language to be translated. A dictionary creation means for creating a dictionary for translation of the dictionary contents;
Storage that stores reading-kana-added data having one or more noun phrases to be translated and the reading-kana of each of the one or more noun phrases in a variable that is distinguishable from other reading-kana-giving data. Means,
The dictionary data of the dictionary for translation created by the dictionary creating means is read one by one, and the noun phrase to be translated of the data with the kana added stored in the variable is converted to the dictionary data of the dictionary for translation. A translation program that functions as full-text replacement means for replacing the full-text with a target language using.

Images