JP2007034621A - Translation apparatus and translation method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a translation apparatus and method capable of flexibly selecting a translation engine when the translation engine is selected from among the plurality of translation engines prepared to translate document data. <P>SOLUTION: This translation apparatus is provided with a text and use environment receiving part 21 for receiving document data in first language and use environment information relating to the document data from user terminal equipment 12, a plurality of translation engines 1 to 4 for translating the first language into the second language, a translation managing part 25 referring to the storage content of an evaluation database 26 from the received use environment information, and for selecting the first translation engine on the basis of the preliminarily decided rule, and for making the translation engine translate the received document data in the first language into the document data in the second language, and for returning the document data to the user terminal equipment 12, and a translation engine selection updating part 32 for updating the storage content of an evaluation database 26. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a translation apparatus and a translation method for performing translation using a translation engine selected from a plurality of translation engines capable of performing translation from a first language to a second language.

As a technology related to a translation device, a certainty factor of a translation result for a natural language sentence is calculated, and the post-editing work efficiency is improved by controlling the presentation order of translated sentences to be post-edited by this value (for example, patent document) 1) and the certainty of the translation candidates are displayed in an identifiable manner, so that the translation results can be easily verified, the translation is evaluated by a computer, and the quality of the machine translation is automatically and objectively evaluated. There are some which are evaluated (for example, refer to Patent Document 2).
Japanese Patent Laid-Open No. 7-200602 JP-A-5-158773

  By the way, when performing translation using only one translation engine, it takes longer to translate as the translation is performed with higher translation accuracy. However, the user's requirements are diverse, for example, the translation speed is allowed to allow a slight decrease in translation accuracy. If only one translation engine with high translation accuracy is used, such a variety of requests are required. Cannot respond to requests. For this reason, a plurality of translation engines are prepared and the translation engine selected from among them is considered for translation. However, there is a problem that it is difficult to select an optimal translation engine. It is also possible to find an optimum translation engine by automatically and objectively evaluating the quality of a machine translation sentence by evaluating the translation sentence with a computer using the technique described in Patent Document 2 described above. However, the technique described in Patent Document 2 has a problem that since the translation is evaluated by a computer, the evaluation is uniform and there is no flexibility.

  The present invention has been made in view of such problems. The purpose of the present invention is to prepare a plurality of translation engines, and to perform a translation using a translation engine selected from the plurality of translation engines. A translation apparatus and a translation method that can be used.

  In order to achieve the above object, the present invention can translate source document information input means for inputting document data in the first language and usage environment information related to the document data, and can translate the first language into the second language. A plurality of translation engines; storage means storing evaluation values corresponding to use environment information for each translation engine; use environment information input by the translation source document information input means; and storage contents of the storage means. A translation control means for referring to and selecting one translation engine from the plurality of translation engines based on a predetermined rule and translating the document data by the translation engine; and the translation selected by the translation control means There is provided a translation apparatus comprising translation result output means for outputting document data of the second language translated by an engine.

  Thus, when the translation source document information input means inputs the document data in the first language and the usage environment information related to the document data from the user terminal, the translation control means stores the stored contents of the storage means from the input usage environment information. , One translation engine is selected from a plurality of translation engines based on a predetermined rule, and the input document data of the first language is translated into the second language by this translation engine. Thus, the translation result reply means outputs the translated document data in the second language. Thereby, the user can obtain a translation.

  In this case, the translation control unit for evaluation for translating the document data of the first language input by the source document information input unit into the second language by each of the plurality of translation engines, and the source document information input unit The evaluator terminal that the evaluator uses the input document data of the first language, the usage environment information related to the document data, and the document data of the second language obtained by translating the document data by each of the plurality of translation engines. The evaluation information transmitting means for transmitting to the evaluation information, the evaluation information receiving means for receiving the evaluation information from the evaluator terminal, and the storage contents of the storage means are updated based on the evaluation information received by the evaluation information receiving means. Updating means.

  Thus, the evaluation translation control means translates the document data of the first language input by the translation source document information input means into the second language by each of the plurality of translation engines, and the evaluation transmission means The document data in the second language is transmitted to the evaluator terminal together with the document data in the first language that is the basis and the usage environment information. Then, the evaluator comprehensively looks at these and evaluates each translation engine, and outputs the evaluation information. This is received by the evaluation information receiving means, and based on this evaluation information, the updating means updates the storage contents of the storage means as necessary. Next, when the document data in the first language and the usage environment information related to the document data are received from the user terminal, the translation control unit converts the input usage environment information into the storage contents of the storage unit updated as necessary. Based on this, one translation engine is selected from a plurality of translation engines. In other words, the evaluation result of the evaluator is reflected in the selection of the translation engine at the time of the next translation, so that the selection can be made with great flexibility without being uniform.

  In this case, the evaluation transmission unit selects an optimal evaluator terminal based on the usage environment information received by the translation source document information input unit, and the translation source document is sent to the evaluator terminal. The first language document data input by the information input means, the usage environment information relating to the document data, and the second language document data obtained by translating the document data by each of the plurality of translation engines may be transmitted. good.

  As a result, the evaluation transmitting means selects and transmits the optimum evaluator terminal based on the usage environment information input by the translation source document information input means. it can. Therefore, the translation engine selection is reliable.

  Further, the present invention corresponds to a translation source document information input step for inputting document data in a first language and usage environment information related to the document data from a user terminal used by the user, and the usage environment information for each translation engine. A storage step of storing the evaluation value in the storage means; the use environment information input in the source document information input step among the plurality of translation engines capable of translating the first language into the second language; and the storage A translation control step of referring to the stored contents of the means, selecting one translation engine from the plurality of translation engines based on a predetermined rule, and translating the document data by the translation engine, and the translation control A translation result output step of outputting the second language document data translated by the translation engine selected in the step It is provided as a translational wherein.

  In this case, an evaluation translation control step of translating the document data of the first language input in the source document information input step into the second language by each of the plurality of translation engines, and in the source document information input step The evaluator terminal that the evaluator uses the input document data of the first language, the usage environment information related to the document data, and the document data of the second language obtained by translating the document data by each of the plurality of translation engines. The content of the storage means is updated based on the evaluation information received in the evaluation information received step, the evaluation information receiving step for receiving evaluation information from the evaluator terminal, and the evaluation information receiving step. And an update step.

Next, an embodiment as the best mode for carrying out the invention will be described.
As shown in FIG. 1, the translation apparatus 11 according to the present embodiment is connected to a plurality of user terminals 12 including personal computers used by a user via a network such as the Internet, and is used by an evaluator. Similarly, a plurality of evaluator terminals 14 composed of personal computers are connected via a network (all of which are not shown).

  The translation apparatus 11 according to the present embodiment translates a first language (for example, English) into a second language (for example, Japanese) different from this, and has the configuration shown in FIG. 2, via a network. Original text / use environment receiving unit (translation source document information input) which is connected to a plurality of user terminals 12 and receives document data in the first language as the original text of translation and use environment information related to the document data from each user terminal 12 Means) 21 and a plurality of, specifically four, translation engines 1, 2, 3, 4 and translation engines 1 to 4 capable of translating the first language into the second language. It has translation engine dictionaries 1A, 2A, 3A, 4A.

  The translation device 11 also translates the document data in the first language received by the original text / usage environment receiving unit 21 into the second language using a translation engine appropriately selected from the plurality of translation engines 1 to 4. (Translation control means, translation result reply means, evaluation translation control means) 25 and a plurality of usage environment information received by the original / usage environment receiving unit 21 based on the storage contents of the evaluation database (storage means) 26, etc. A translation engine selection unit (translation control means) 27 that selects one translation engine from the translation engines 1 to 4 is provided.

  The translation management unit 25 returns the document data in the second language translated by the one selected by the translation engine selection unit 27 from the translation engines 1 to 4 to the user terminal 12 that has transmitted the original text. . The translation management unit 25 translates the document data of the first language received by the original / usage environment receiving unit 21 into the second language by each of the plurality of translation engines 1 to 4.

  The translator 11 translates the second language document data translated by each of the plurality of translation engines 1 to 4 together with the original language document data of the first language and the usage environment information related to the document data. An evaluation transmission unit (evaluation transmission means) 30 that transmits to the evaluator terminal 14 to be used, and an evaluation information reception unit (evaluation information reception) that receives evaluation information output from the evaluator terminal 14 at an arbitrary time Means) 31 and a translation engine selection updating unit (updating means) 32 for updating the stored contents of the evaluation database 26 based on the evaluation information received by the evaluation information receiving unit 31. That is, the evaluation transmitter 30 sends the document data of the second language translated by each of the plurality of translation engines 1 to 4 together with the document data of the first language as the original text and the usage environment information related to the document data. If it transmits to the machine 14, an evaluator will evaluate the translation result of each translation engine 1-4 using the evaluator terminal 14, and will transmit evaluation information to the evaluation information receiving part 31 of the translation apparatus 11. FIG.

  Moreover, the translation apparatus 11 updates the translation engine dictionary update data of the translation engine dictionary units 1A to 4A of the translation engines 1 to 4 with the dictionary information by the evaluator received from the evaluator terminal 14 by the evaluation information receiving unit 31. A portion 33 is provided. That is, in the evaluator terminal 14, the evaluator evaluates the translation result of each translation engine 1 to 4 and inputs evaluation information. At this time, the first language that cannot be translated by each translation engine 1 to 4 When the evaluator knows the translated word of the first language phrase into the second language, the evaluation information receiving unit 31 transmits the phrase. Then, the translation engine dictionary updating unit 33 updates the data of the translation engine dictionary units 1A to 4A of the translation engines 1 to 4 that have not been able to translate this phrase in correspondence with the first language phrase and the second language. . Further, in the evaluator terminal 14, when the evaluator inputs the translation result of each translation engine 1 to 4, if there is a sentence that cannot be translated by each translation engine 1 to 4, the evaluator can use the first language. A translation of the sentence of the sentence into the second language is created by itself and transmitted to the evaluation information receiving unit 31. The evaluation information receiving unit 31 outputs this to the translation management unit 25, and the translation management unit 25 Can be returned to the user terminal 12.

  Here, the evaluation transmitter 30 selects one or a plurality of optimal evaluator terminals 14 based on the usage environment information received by the original text / usage environment receiver 21, and selects the selected evaluator terminals 14. Along with this use environment information, the document data in the first language received by the original text / use environment receiving unit 21 and the document data in the second language obtained by translating the document data in the first language by each of the plurality of translation engines 1 to 4 are transmitted. To do.

  Next, a translation method using the translation apparatus of this embodiment will be described with reference to specific examples shown in FIGS.

  First, the translation apparatus 11 receives from the predetermined user terminal 12 used by the user, the original language / use environment receiving unit 21 receives the document data of the first language as the translation target sentence and the use environment information related to the document data ( Source document information input step). Here, when the user terminal 12 is in a state of communicating with the translation apparatus 11 via the network and the work for requesting the translation to the translation apparatus 11 is used, usage environment information is input in the user terminal 12. This usage environment information includes, for example, the purpose of the user's use of the translation target sentence, the current situation, what kind of field the document is, and how much time the translation result is desired. What is the required translation accuracy?

  Next, the translation apparatus 11 converts the received usage environment status data into numerical values, that is, vectors, in the original text / usage environment reception unit 21. For example, as shown by X1 in FIG. 3, the document type is a communication document as the usage environment information together with the document data of the sentence to be translated “A cherry tree sessions to new in New York. How is that?” If the purpose of use is private use, the category includes seasonal sentences and greetings, the translation time is within 30 seconds, and the required translation accuracy is 80%, for example, as shown by X2 in FIG. Create a usage environment vector. In this case, the variable 1 indicating the document type is 0 indicating that the communication document has been selected, the variable 2 indicating the purpose of use is 1 indicating private use, and the variable 3 indicating the category is 0 indicating the seasonal sentence. Then, the variable 4 indicating another category is 2 indicating the greeting sentence, the variable 5 indicating the translation time is 30 indicating 30 seconds or less, and the variable 6 indicating the required translation accuracy is 80 indicating 80%. The received usage environment situation is represented by a vector having such six variables.

  Further, the original text / usage environment receiving unit 21 may digitize, that is, vectorize, the received document data of the translation target sentence as a part of the data of the usage environment status. In this case, the tf. A document vector using tf, idf and the like as a vector component is created. In the case of the above translation target sentence, for example, as indicated by X3 in FIG. 3, the variable 22 indicating “tree” is 1, the variable 131 indicating “York” is 1, and the variable 240 indicating “New” is The variable 579 indicating “cherry” is 1, the variable 1300 indicating “that” is 1, and the variable 2200 indicating “bloom” is 1.

  Next, the translation device 11 translates from the use environment information received in the translation source document information input step among the plurality of translation engines 1 to 4 capable of translating the mounted first language into the second language. The management unit 25 causes the translation engine selection unit 27 to select one translation engine based on the contents stored in the evaluation database 26 (described later) (translation engine selection step). The translation engine selection unit 27 inputs the above-described use environment vector and the like, and selects any one of the translation engines 1 to 4 as the optimum according to the translation engine selection rule. Then, the translation management unit 25 converts the translation target sentence of the first language received in the translation source document information input step to any one of the translation engines 1 to 4 selected by the translation engine selection unit 27. Sending and translating to be translated into the second language, and receiving the translation result and translation time (translation step). Next, the translation management unit 25 returns the received translation result and translation time to the user terminal 12 (translation result reply step).

  Here, as the above-described translation engine selection rule of the translation engine selection unit 27, the evaluators 1B to 4B that produce appropriateness for each of the translation engines 1 to 4 are created, and the evaluations of the evaluators 1B to 4B are performed. A rule for comparing the values and selecting the optimum one from the translation engines 1 to 4 (hereinafter, all translation engine evaluation rules) and a translation engine evaluation for selecting the optimum one of the translation engines 1 to 4 One of the rules for selecting the optimum one from the translation engines 1 to 4 (hereinafter, the optimum translation engine selection rule) is adopted by the device 36.

  In the all translation engine evaluation rules, the usage environment vector is input to the translation engine selection unit 27, and appropriateness is determined according to a preset map or the like by appropriately weighting each variable of the input usage environment vector, for example. Evaluators 1B to 4B are prepared for each of the translation engines 1 to 4, and the translation engine selection unit 27 inputs the use environment vector with the evaluator 1B for one translation engine 1 to determine the appropriateness of the translation engine 1. Similarly, for each of the other translation engines 2 to 3, the respective evaluators 2 </ b> B to 4 </ b> B give the appropriateness. Then, the translation management unit 25 causes the translation engine having the highest appropriateness to perform translation. In addition, it is also possible to input a document vector instead of the usage environment vector, or to input the usage environment vector and the document vector together. Become.

  As shown in FIG. 4, for example, when there are translation engines 1 to 4, the above variable 1 is 0, variable 2 is 1, variable 3 is 0, variable 4 is 2, variable 5 is 30, variable 6 is 80. Is input to the evaluator 1B for the translation engine 1, the evaluator 1B outputs the appropriateness 60 of the translation engine 1, and the same usage environment vector is input to the evaluator 2B for the translation engine 2. When this evaluator 2B outputs the appropriateness 75 of the translation engine 2 and inputs the same use environment vector to the evaluator 3B for the translation engine 3, the evaluator 3B outputs the appropriateness 40 of the translation engine 3, and the same When the usage environment vector is input to the evaluator 4B for the translation engine 4, the evaluator 4B outputs the appropriateness 50 of the translation engine 4. Since the appropriateness of the translation engine 1 is second, the appropriateness of the translation engine 2 is first, the appropriateness of the translation engine 3 is fourth, and the appropriateness of the translation engine 4 is third, the translation management unit 25 Replies to the corresponding user terminal 12 along with the translation time, the translation result of the translation engine 2 “Cherry blossoms are likely to bloom in New York.

  In the optimum translation engine selection rule, the use environment vector is input to the translation engine selection unit 27, and the optimum translation engine is set according to a map set in advance by appropriately weighting each variable of the input use environment vector. The translation engine evaluator 36 for determining the translation engine is prepared, the translation engine selector 27 inputs the use environment vector to the translation engine evaluator 36, selects the optimum one from the translation engines 1 to 4, and the translation manager 25 Let the translation engine with the most appropriate level translate. As shown in FIG. 5, for example, the use environment vector in which the variable 1 is 0, the variable 2 is 1, the variable 3 is 0, the variable 4 is 2, the variable 5 is 30, and the variable 6 is 80 is used as the translation engine evaluator 36. The translation engine evaluator 36 outputs the optimum translation engine identification number from the translation engines 1 to 4 as a selection result.

  Apart from the translation process of the translation engine selection step, the translation step, and the translation result reply step described above, the translation management unit 25 receives the document data of the translation target sentence received in the translation source document information input step as a plurality of translation engines 1. Each of ˜4 is translated into a second language (evaluation translation step). Then, the translation management unit 25 translates the document data of the sentence to be translated, the usage environment information related to the document data, and the document data by each of the plurality of translation engines 1 to 4 in the evaluation translation step. The second language document data, which is the translation result, and the translation time translated in each are collectively transmitted to the evaluator terminal 14 used by the evaluator (evaluation transmission step). At this time, the evaluation transmission unit 30 selects one or a plurality of optimum evaluator terminals 14 based on the usage environment information received together with the translation target sentence, and transmits the selected evaluator terminals 14 to the selected evaluator terminal 14. That is, the evaluation transmitting unit 30 groups evaluators in advance using each attribute (special field) of the usage environment vector, and sets the evaluator terminals 14 used by the evaluator optimal for evaluation. Select and send. Specifically, for the above translation target sentence, for example, as shown by Y1 in FIG. 6, the translation result of the translation engine 1 “Cherry seems to bloom in New York. It seems to bloom. How are you? And the translation time of 0.8 seconds, and the translation result of translation engine 2 “Cherry blossoms are likely to bloom in New York. How about that?” And the translation time of 0.3 seconds. “Translation not possible”, which indicates that the translation was impossible in step 3, and “translation engine 4”, “Sakura is likely to bloom in the new York. How about that?” And its translation time of 0.2 seconds. As shown by Y2 in FIG. 6, the type of document that is the usage environment information is a communication document, the purpose of use is private use, and the category includes seasonal sentences and greetings. Time is within 30 seconds, to the evaluator terminal 14 of the selected and it is translated accuracy 80% evaluator seeking.

  The evaluator evaluates the translation result of each translation engine 1 to 4 received from the translation device 11 using the evaluator terminal 14.

  When all the translation engine evaluation rules are adopted as the translation engine selection rules described above, the evaluator determines the translation accuracy and the appropriateness considering the translation accuracy and the usage environment for each of the translation engines 1 to 4. The appropriateness is output from the evaluator terminal 14 to the evaluation information receiving unit 31 of the translation apparatus 11. For example, as indicated by Y3 in FIG. 6, the translation result of the translation engine 1 has a translation accuracy of 80 and an appropriateness of 60, the translation result of the translation engine 2 has a translation accuracy of 90 and an appropriateness of 80, and the translation of the translation engine 3 The translation accuracy is 0 and the appropriateness is 0, and the translation result of the translation engine 4 is output to the translation device 11 that the translation accuracy is 50 and the appropriateness is 60.

  Further, when the optimum translation engine selection rule is adopted as the translation engine selection rule described above, the evaluator selects and determines the most suitable one of the translation engines 1 to 4 and determines the selection result as the evaluator terminal. 14 to the evaluation information receiver 31 of the translation apparatus 11. For example, as indicated by Y4 in FIG. 6, the translation engine 2 as an optimum translation engine is output to the translation device 11. For example, when the evaluator determines that the translation results of all the translation engines 1 to 4 are not appropriate, the evaluator may arbitrarily create a translation and output it from the evaluator terminal 14 to the translation device 11. Is possible. In this case, in the translation apparatus 11, the translation management unit 25 also returns the translated text generated by the evaluator received by the evaluation information receiving unit 31 to the user terminal 12. For example, as indicated by Y5 in FIG. 6, a translated sentence “Looks like cherry blossoms in New York. How are you?” Is sent back to the user terminal 12.

  In the translation apparatus 11, the evaluation result of the evaluator received by the evaluation information receiving unit 31 is accumulated in the evaluation database 26 by the translation engine selection / updating unit 32, thereby updating the stored contents of the evaluation database 26.

  When all the translation engine evaluation rules are adopted as the translation engine selection rules, when the same use environment vector is input next, as shown by Z1 in FIG. The evaluators 1B to 4B created for all the translation engines 1 to 4 of the selection unit 27 have the translation accuracy indicated by Z2 in FIG. 7 attached by the evaluator stored in the evaluation database 26 in addition to the use environment vector. Each evaluator 1B-4B adds the translation accuracy and appropriateness assigned by the evaluator by adding, for example, a separate weight to determine the appropriateness of each translation engine 1-4. Find out. For example, as shown by Z3 in FIG. 7, the usage environment vector of variable 1 is 0, variable 2 is 1, variable 3 is 0, variable 4 is 2, variable 5 is 30, variable 6 is 80, and the evaluated translation 60 of the variable 7 indicating the appropriateness of the engine 1 is input to the evaluator 1B for the translation engine 1. Similarly, the same use environment vector and 80 of the variable 7 indicating the appropriateness of the evaluated translation engine 2 are input to the evaluator 2B for the translation engine 2, and the same use environment vector and the evaluated translation engine are input. 0 of the variable 7 indicating the appropriateness of 3 is input to the evaluator 3B for the translation engine 3, and the same use environment vector and 60 of the variable 7 indicating the appropriateness of the evaluated translation engine 4 are input to the translation engine 4 To the evaluator 4B. And each evaluator 1B-4B calculates the appropriateness of each translation engine 1-4 based on input data. The evaluators 1B to 4B are updated by machine learning based on such an evaluator's evaluation value.

  When the optimum translation engine selection rule is adopted, when the same use environment vector is input next, as shown by W1 in FIG. 8, the translation engine selection update unit 32 causes the translation engine selection unit 27 to In addition to the use environment vector, the translation engine evaluator 36 inputs the identification of the optimum translation engine indicated by W2 in FIG. 8 attached by the evaluator stored in the evaluation database 26. The translation engine evaluator 36 The optimum translation engine is determined by adding the identification of the optimum translation engine assigned by the evaluator with, for example, a separate weight. For example, as shown by W3 in FIG. 8, the variable 1 indicates that the variable 1 is 0, the variable 2 is 1, the variable 3 is 0, the variable 4 is 2, the variable 5 is 30, and the evaluation is optimal. 2 indicating the 6 translation engines 2 is input to the translation engine evaluator 36. Then, the translation engine evaluator 36 determines an optimal translation engine based on the input data. The translation engine evaluator 36 is updated by machine learning based on such an evaluator's evaluation value.

As described above, according to the present embodiment, when the original text / usage environment receiving unit 21 receives document data in the first language and usage environment information related to the document data from the user terminal 12, the translation management unit 25 Selects the most suitable translation engine from the plurality of translation engines 1 to 4 based on the stored contents of the evaluation database 26 and predetermined rules from the received usage environment information, and the received first language Is translated into the second language, and the translated document data in the second language is returned to the user terminal 12. As a result, the translated text can be received by the user terminal 12.
On the other hand, the translation management unit 25 translates the document data of the first language received by the original text / usage environment reception unit 21 into the second language by each of the plurality of translation engines 1 to 4. However, the document data in the second language is transmitted to the evaluator terminal 14 together with the document data in the first language and the usage environment information that are based on the second language document data. Then, the evaluator comprehensively sees these, evaluates each translation engine 1 to 4, and outputs the evaluation information. This is received by the evaluation information receiving unit 31, and based on this evaluation information, the translation engine selection updating unit 32 updates the stored contents of the evaluation database 26 as necessary. Then, when the document data in the first language and usage environment information related to the document data are received from the user terminal 12, the translation management unit 25 evaluates the evaluation database 26 updated as necessary from the received usage environment information. One translation engine is selected from among the plurality of translation engines 1 to 4 based on the stored contents. In other words, the evaluator's evaluation results are reflected in the selection of the optimal one of the translation engines 1 to 4 at the time of the next translation. It will be possible.

  Further, the evaluation transmitter 30 selects the optimum evaluator terminal 14 based on the usage environment information received by the original / usage environment receiver 21 and transmits the evaluation for evaluation. I can have you. Therefore, the optimum selection of the translation engines 1 to 4 becomes highly reliable.

  The translation engine dictionary updating unit 33 updates the dictionary of a translation engine other than the translation engine using the translation sentence of the translation engine having the highest evaluation by the evaluator among the translation engines 1 to 4. good.

  Further, as shown in FIG. 9, the translation management unit 25 may be provided with a customization database 40 for customization data. In this way, the translation results of all the translation engines 1 to 4 are temporarily stored in the translation management unit 25 and the translation results are not released when the translation is provided to the user side. According to the storage time and the storage capacity, the translation results of all the translation engines 1 to 4 and the translations created by the evaluator can be stored at any time. These translation results are translated when the user obtains the translation result of the next rank after the translation result by the optimum one of the translation engines 1 to 4 is transmitted to the user, and further when the translation result of the next rank is obtained. The result can be provided. Depending on the user's request, the translation engines 1 to 4 and the evaluator can be designated. The stored translation can be released by user operation.

  In addition, the following user customization functions can be added. The parameters of each translation engine 1 to 4, the parameters of the translation selection method, the evaluation data for the translation used by the user are stored in the customized data, and are provided to the user periodically. The user transmits customization data to the translation apparatus 11 when using the translation. The translation engines 1 to 4 are updated from the data, and the translation engine selection unit 27 is generated by the translation environment reflecting the user's translation history. Thereby, both the translation result customized by the user and the general translation result can be acquired.

  Furthermore, the present invention is applicable even when the translation device 11 is built in the user terminal 12. In this case, the original / usage environment receiving unit 21 of the translation apparatus 11 functions as an input unit that receives the operation from the user directly and inputs the original / usage environment information.

It is a block diagram which shows the relationship between the translation apparatus which concerns on one Embodiment of this invention, a user terminal, and an evaluator terminal. It is a block diagram which shows the structure of the translation apparatus which concerns on one Embodiment of this invention. It is a figure which shows the document data of the translation origin received in the translation apparatus which concerns on one Embodiment of this invention, the data of a use environment condition, and an example of vectorization. It is a figure which shows the example of the input and output of the evaluator of each translation engine at the time of employ | adopting all the translation engine evaluation rules in the translation apparatus which concerns on one Embodiment of this invention. It is a figure which shows the example of the input and output of a translation engine evaluator at the time of employ | adopting the optimal translation engine selection rule in the translation apparatus which concerns on one Embodiment of this invention. It is a figure explaining the example of the evaluation data in the translation apparatus concerning one Embodiment of this invention. It is a figure which shows the example of an input to the evaluator of each translation engine at the time of employ | adopting all the translation engine evaluation rules in the translation apparatus which concerns on one Embodiment of this invention. It is a figure which shows the example of an input to a translation engine evaluator at the time of employ | adopting the optimal translation engine selection rule in the translation apparatus which concerns on one Embodiment of this invention. It is a block diagram which shows the structure of the modification of the translation apparatus which concerns on one Embodiment of this invention.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1-4 ... Translation engine, 11 ... Translation apparatus, 12 ... User terminal, 14 ... Evaluator terminal, 21 ... Original text / use environment receiving part (translation source document information input means), 25 ... Dictionary management part (translation control) Means, translation result reply means, evaluation translation control means), 26 ... evaluation database (storage means), 27 ... translation engine selection section (translation control means), 30 ... evaluation transmission section (evaluation transmission means), 31 ... Evaluation information receiver (evaluation information receiver), 32... Translation engine selection updater (updater).

Claims (5)

Translation source document information input means for inputting document data in the first language and usage environment information related to the document data;
A plurality of translation engines capable of translating the first language into the second language;
Storage means for storing an evaluation value corresponding to use environment information for each translation engine;
With reference to the usage environment information input by the translation source document information input means and the storage contents of the storage means, one translation engine is selected from the plurality of translation engines based on a predetermined rule. Translation control means for translating the document data by the translation engine;
Translation apparatus comprising: translation result output means for outputting the document data of the second language translated by the translation engine selected by the translation control means. Translation control means for evaluation for translating the document data of the first language input by the translation source document information input means into the second language by each of the plurality of translation engines;
The first language document data input by the translation source document information input means, the usage environment information related to the document data, and the second language document data obtained by translating the document data by each of the plurality of translation engines are evaluated. Transmitting means for evaluation to be transmitted to the evaluator terminal used by the person,
Evaluation information receiving means for receiving evaluation information from the evaluator terminal;
The translation apparatus according to claim 1, further comprising: an updating unit that updates a storage content of the storage unit based on the evaluation information received by the evaluation information receiving unit. The evaluation transmission means selects an optimal evaluator terminal based on the usage environment information received by the translation source document information input means, and inputs to the evaluator terminal by the translation source document information input means. 3. The second language document data, the usage environment information related to the document data, and the second language document data obtained by translating the document data by the plurality of translation engines, respectively, are transmitted. Translation device. A translation source document information input step of inputting document data in the first language and usage environment information related to the document data from a user terminal used by the user;
A storage step of storing an evaluation value corresponding to use environment information for each translation engine in a storage unit;
Of the plurality of translation engines capable of translating the first language into the second language, the use environment information input in the translation source document information input step and the storage contents of the storage means are referred to and determined in advance. A translation control step of selecting one translation engine from the plurality of translation engines based on the rule and translating the document data by the translation engine;
A translation result output step of outputting the second language document data translated by the translation engine selected in the translation control step. A translation control step for evaluation for translating the document data of the first language input in the translation source document information input step into the second language by each of the plurality of translation engines;
The first language document data input in the translation source document information input step, the usage environment information related to the document data, and the second language document data obtained by translating the document data by each of the plurality of translation engines are evaluated. A transmission step for evaluation to be transmitted to the evaluator terminal used by the person,
An evaluation information receiving step for receiving evaluation information from the evaluator terminal;
The translation method according to claim 4, further comprising an update step of updating the storage content of the storage unit based on the evaluation information received in the evaluation information receiving step.

Images