JP7011124B2 - Machine translation system and machine translation method - Google Patents

Description

The present invention relates to a machine translation system and a machine translation method, in particular, a machine translation system and a machine translation method using international publication data. In addition, in this specification, an international publication means a published patent document (WOXXX / YYYYYY A1 or A2) published by the World Intellectual Property Organization (WIPO).

Currently, various companies provide machine translation systems. However, the accuracy of machine translation is still low. Therefore, it is conceivable to improve the accuracy of machine translation by using AI (Artificial Intelligence). However, it is not easy to improve the accuracy of machine translation even with AI in a mixture of high-quality data and poor-quality data.

Therefore, the present invention is to provide a machine translation system and a machine translation method capable of improving the accuracy of machine translation.

Hereinafter, means for solving the problem will be described using the numbers and codes used in the embodiment of the invention. These numbers and codes are added in parentheses for reference in order to show an example of the correspondence between the description of the claims and the embodiment for carrying out the invention. Therefore, the scope of claims should not be construed in a limited manner by the description in parentheses.

The machine translation system in some embodiments is a machine translation system that translates a first sentence element described in a first language into a second sentence element described in a second language. The machine translation system (1) stores the input device (10) for inputting the first sentence element into the computer (20), the computer (20) for receiving the first sentence element from the input device (10), and the computer (20). It is equipped with a device (40). The storage device (40) has the text data of the international publication written in the second language and the text data of the corresponding publication corresponding to the international publication and described in the first language. I remember. Upon receiving the first sentence element from the input device, the computer (20) determines a specific publication including the first sentence element from a plurality of corresponding publications. The computer (20) determines a specific international publication corresponding to the specific publication from among the plurality of international publications. The computer (20) extracts the second sentence element corresponding to the first sentence element from the specified international publication. The computer (20) outputs the extracted second sentence element as a translation of the first sentence element.

In the translation system, the storage device (40) may store the international publication in association with the nationality data of the international publication. Further, the computer (20) may determine the specific international publication based on the nationality data of the international publication.

In the translation system, the computer (20) may output the second sentence element together with the nationality data of the specified international publication.

In the translation system, when the plurality of the second sentence elements are extracted from the plurality of specific international publications, the computer (20) claims among the plurality of the second sentence elements. The text elements extracted from the description field in the range of may be preferentially output.

The machine translation method in some embodiments is a machine translation method that translates a first sentence element described in a first language into a second sentence element described in a second language. The machine translation method includes (A) a receiving process in which the computer (20) receives the first sentence element from the input device (10), and (B) a plurality of computers (20) described in the first language. The first determination step of determining the specific publication including the first sentence element, and (C) the computer (20) is described in a plurality of international publications in a second language. The second determination step of determining the specific international publication corresponding to the specific public publication, and (D) the computer (20) corresponding to the first sentence element from the specific international publication. It includes an extraction step of extracting a second sentence element, and (E) an output step of the computer (20) outputting the extracted second sentence element as a translation of the first sentence element.

In the machine translation method, in the receiving process, the computer (20) may receive sentence data including a plurality of the first sentence elements. The computer may divide the plurality of first sentence elements into a sentence element including the technical term and a sentence element not including the technical term. Further, the computer may execute the first determination step, the second determination step, and the extraction step only for the sentence element including the technical term.

In the machine translation method, (F) a search step in which the computer (20) searches for the specific publication including the first sentence element from a plurality of publications described in the first language. , (G) The computer (20) divides the first sentence element into a plurality of small elements including the first small element and the second small element, and (H) the computer (20) A step of determining a first specific publication including the first minor element from a plurality of publications described in the first language, and (I) the computer (20) in the first language. A step of determining a second specific publication including the second minor element from the plurality of described publications, and (J) a plurality of computers (20) described in the second language. From the international publications, a step of determining a first specific international publication corresponding to the first specific publication, and (K) the computer (20) having a plurality of the above described in the second language. The step of determining the second specified international publication corresponding to the second specified international publication from the international publication, and (L) the computer (20) from the first specified international publication The step of extracting the third minor element corresponding to the first minor element and described in the second language, and (M) the computer (20) from the second specified international publication, the second. The step of extracting the fourth minor element corresponding to the minor element and described in the second language, and (N) the second computer (20) including the third minor element and the fourth minor element. It may further include a step of outputting the sentence element as a translation of the first sentence element. Further, when the specific publication is found in the search step, the computer (20) may execute the step (B), the step (C), and the step (D). Further, if the specific publication is not found in the search step, the computer (20) may use the step (G), the step (H), the step (I), the step (J), and the step ( K), the step (L), the step (M), and the step (N) may be executed.

INDUSTRIAL APPLICABILITY According to the present invention, it is possible to provide a machine translation system and a machine translation method capable of improving the accuracy of machine translation.

FIG. 1 is a functional block diagram of a machine translation system according to an embodiment or each modification. FIG. 2 is a flowchart showing an outline of the machine translation method in each modification except the embodiment or the sixth modification. FIG. 3 is a diagram showing how the output device outputs the second sentence element. FIG. 4 is a diagram showing how the output device outputs the second sentence element. FIG. 5 is a diagram showing how the output device outputs the second sentence element. FIG. 6 is a diagram showing how the output device outputs the second sentence element. FIG. 7 is a flowchart showing an outline of the machine translation method in the sixth modification. FIG. 8 is a flowchart showing an outline of the machine translation method in the sixth modification. FIG. 9 is a flowchart showing an outline of the machine translation method in the sixth modification.

Hereinafter, the machine translation system and the machine translation method will be described with reference to the attached drawings. In the following description, the members having the same function are given the same drawing number, and the repeated description of the members having the same drawing number is omitted. In the following embodiment, the case where the first language is Japanese and the second language is English will be described. However, the first language is not limited to Japanese. The first language may be, for example, Chinese, French, German, or English. Also, the second language is not limited to English. The second language may be German, French, Chinese, or Japanese. The second language is a language different from the first language.

An outline of the machine translation system according to the embodiment will be described with reference to FIG. FIG. 1 is a functional block diagram of the machine translation system 1 in the embodiment.

The machine translation system 1 in the embodiment is a translation system that translates a first sentence element described in Japanese (first language) into a second sentence element described in English (second language). For example, in some embodiments, the machine translation system translates the first sentence element "extending from to the inner diameter" into the second sentence element "extends radially inwardly from". On the other hand, one of the known machine translation systems translates the first sentence element "extending in the inner diameter direction from" into the second sentence element "extends radiusy". That is, the translation accuracy of a known machine translation system is not high.

The machine translation system 1 includes an input device 10, a computer 20, and a storage device 40. The machine translation system 1 may include an output device 50.

The input device 10 inputs the first sentence element described in Japanese (first language) to the computer 20. The input device 10 may be a keyboard connected to the computer 20 so as to be able to transmit information, may be a storage medium such as a USB memory, may be another computer, and may transfer data to the computer 20. It may be an internet line to be transmitted to. The first sentence element may be a word, a phrase in which a particle or an auxiliary verb is added to the word, or a phrase.

The computer 20 receives the first sentence element from the input device 10. The computer includes a CPU (Central Processing Unit) and an internal memory (storage device). The computer executes a program stored in the internal memory to determine a first determination means 24 for determining a specific publication to be described later, a second determination means 26 for determining a specific international publication to be described later, or a second sentence. It functions as an extraction means 28 for extracting elements.

The storage device 40 is a storage device connected to the computer 20 so as to be able to transmit information. The storage device 40 is, for example, a hard disk, a server, or the like. The storage device 40 stores the text data of the international publication written in the second language and the text data of the corresponding public publication corresponding to the international publication and described in the first language. The text data of the international publication includes, for example, the English text data of the international publication No. 2007/033072, and the corresponding publication data includes, for example, the Japanese text data of the special table 2009-508065. included. Japanese Patent Publication No. 2009-508065 is a patent family document of International Publication No. 2007/033072. That is, in the present specification, the "corresponding publication corresponding to the international publication" means the patent family document of the international publication. Alternatively, the "corresponding publication corresponding to the international publication" means a translated document prepared by translating the specification of the international application corresponding to the international publication, the scope of claims, etc. at the time of domestic transition. .. The corresponding public gazette is, for example, a gazette (literature) issued by a country whose official language is the first language.

The storage device 40 may include a large number, for example, text data of 100,000 or more international publications, or may include text data of 100,000 or more corresponding publications.

The output device 50 is, for example, a display device such as a display, an audio output device, a storage device, or a line such as an Internet line. The computer 20 outputs the second sentence element as a translation result to the output device 50. As a result, when the output device 50 is a display device, the second sentence element is displayed on the display device. Alternatively or additionally, if the output device 50 is a voice output device, the second sentence element is pronounced. Alternatively or additionally, if the output device 50 is a storage device, a file containing the second sentence element is stored in the storage device. Alternatively or additionally, when the output device 50 is a line, the data including the second sentence element is transmitted to another device via the line.

How the machine translation system 1 performs machine translation will be described in the following description of the machine translation method.

The machine translation method in the embodiment will be described. FIG. 2 is a flowchart showing an outline of the machine translation method in the embodiment.

In the first step ST1, the computer 20 receives the first sentence element described in the first language from the input device 10. Here, it is assumed that the first language is Japanese and the first sentence element is the phrase "extending from the inner diameter direction". The input of the first sentence element to the computer 20 may be executed by the operator inputting through the keyboard. Alternatively, the input of the first sentence element to the computer 20 may be executed by the external device transmitting the file containing the first sentence element to the computer 20.

In the second step ST2, the computer 20 determines the specific publication including the first sentence element received in the first step ST1 from the plurality of corresponding publications. The decision is made, for example, using a text search. For example, when the first sentence element is "extending from the inner diameter direction", the computer 20 performs a text search on a plurality of corresponding publications stored in the storage device 40, and as a result, the result is A public gazette containing the phrase "extending in the inner diameter direction" is determined as a specific public gazette. For example, the computer 20 determines the special publication No. 2009-508065 as a specific publication. This is because Japanese Patent Publication No. 2009-508065 includes the phrase "extending in the inner diameter direction from" in the column of claims.

When the first language is Japanese, the above-mentioned "plurality of published publications" corresponds to "plurality of published patent publications".

The second step ST2 described above is executed by the computer 20 functioning as the first determination means 24. In other words, the computer 20 functions as a first determination means 24 for determining a specific publication including the first sentence element from a plurality of corresponding publications by executing a program stored in the internal memory. ..

In the third step ST3, the computer 20 determines the specific international publication corresponding to the specific publication from the plurality of international publications. For example, when the special publication is the special publication 2009-508065, the computer 20 is the international publication corresponding to the special publication 2009-508065, that is, the international publication 2007/033072. The issue will be decided as a specified international publication.

Since the computer 20 determines the specific international publication based on the specific publication, the storage device 40 may store each corresponding publication in association with one international publication. It is assumed that the first association data for associating a plurality of corresponding publications with a plurality of international publications is stored in the storage device 40. In this case, the computer 20 can determine the specific international publication corresponding to the specific publication based on the specific publication determined in the second step ST2 and the first association data. Alternatively, each corresponding publication and one international publication do not necessarily have to be associated. In this case, the computer 20 roughly machine-translates the specified publication determined in the second step ST2 into a second language, and the documents match between the plurality of international publications stored in the storage device 40. Perform degree analysis (matching analysis). Then, the computer 20 determines the document having the highest degree of agreement as a specific international publication.

The above-mentioned third step ST3 is executed by the computer 20 functioning as the second determination means 26. In other words, the computer 20 functions as a second determination means 26 for determining a specific international publication corresponding to the specific publication from a plurality of international publications by executing a program stored in the internal memory. do.

In the fourth step ST4, the computer 20 extracts the second sentence element corresponding to the first sentence element from the specified international publication. For example, when the specified international publication is International Publication No. 2007/033072 and the first sentence element is the phrase "extending from the inner diameter direction", the computer 20 uses "extends radius" as the second sentence element. Extract the phrase "inwardly from".

In addition, in order for the computer 20 to extract the second sentence element corresponding to the first sentence element from the specified international publication, the computer 20 can use a known method. For example, the computer 20 uses a known machine translation means (the known machine translation means does not have to be highly accurate in translation) to the text data of the specified publication including the first sentence element. Convert to language text data. The converted text data is referred to as converted text data here. The computer 20 performs matching of the converted text data including the first sentence element converted into the second language with the text data of the specified international publication, so that the first sentence element converted into the second language can be obtained. , Determine which text element corresponds to the text data of the specified international publication. Then, the computer 20 extracts the determined sentence element as the second sentence element.

Alternatively, in order for the computer 20 to extract the second sentence element corresponding to the first sentence element from the specified international publication, the computer 20 may use the following method. That is, the computer 20 determines in which sentence from the front the first sentence element is present in the specified publication. Here, it is assumed that the 100th sentence from the front contains the first sentence element. Further, the computer 20 extracts a key word (for example, a noun or a verb) from the first sentence element. For example, when the first sentence element is the phrase "extending in the inner diameter direction from", the word "extending" is extracted as a key word. In this case, the computer 20 extracts a plurality of sentences (for example, sentences from the 90th sentence from the front to the 110th sentence from the front) located around the 100th sentence from the front from the specified international publication. Then, the computer 20 uses the words corresponding to the key words (for example, the word "extended") from the extracted plurality of sentences (for example, "extend", "extends", "extending", etc.). To find out. Then, the phrase including the found word (for example, "extends vocabulary inwardly from") is extracted as the second sentence element.

The fourth step ST4 described above is executed by the computer 20 functioning as the extraction means 28. In other words, the computer 20 functions as an extraction means 28 for extracting a second sentence element corresponding to the first sentence element from the specified international publication by executing a program stored in the internal memory.

In the fifth step ST5, the computer 20 outputs the extracted second sentence element to the output device 50 as a translation of the first sentence element. For example, when the output device 50 is a display device, the second sentence element is displayed on the display device.

As described above, in the embodiment, the first sentence element is translated into the second sentence element using the text data of the international publication. When the international application is transferred to each country, the language of the specification of the international application is translated accurately so as to correspond to the official language of each country. That is, the text data of the international publication and the text data of the corresponding publication in each country can be a high-quality translation database for technical terms or technical sentences. In the embodiment, the computer 20 translates the first sentence element into the second sentence element using the high quality translation database. Therefore, the accuracy of machine translation is improved.

(First modification)
The first modification is different from the above-described embodiment in that the computer 20 determines the specific international publication based on the nationality data of the international publication. In other respects, the first modification is similar to the embodiment described above.

Even if the international publication is written in a second language, the official language of the country to which the applicant of the international publication belongs or the official language of the country to which the inventor belongs may be different from the second language. be. For example, even if the applicant is Japanese, the international application may not be made in Japanese (first language) but may be made in English (second language). In this case, although the international publication is written in a second language, the official language of the country to which the applicant belongs is not the second language.

One of the major problems with machine translation systems is that when translating a first sentence element written in a first language into a second sentence element written in a second language, a person whose native language is the first language The point is that the created translation is used. For example, an English sentence created by a Japanese person is less accurate as an English sentence than an English sentence created by a person whose mother tongue is English. Therefore, in the first modification, when translating the first sentence element described in the first language into the second sentence element described in the second language, a sentence created by a person whose mother tongue is the second language. To be used.

In the first modification, the storage device 40 is at least one of the country to which the applicant of the international application belongs, the country to which the inventor of the invention of the international application belongs, or the country where the receiving office of the international application is located. I remember one as the nationality data of the international publication. Further, the storage device 40 stores the second association data in which the international publication and the nationality data are associated with each other. Then, the computer 20 determines the above-mentioned specific international publication based on the nationality data of the international publication.

Hereinafter, a specific example will be described. Here, it is assumed that the first sentence elements are "arranged so as to face each other". By executing the second step ST2, the computer 20 selects the special publication No. 2007-505241 from among the plurality of corresponding publications, and the specific publication including the first sentence element (“arranged facing each other”). Determined as one of them. However, referring to the international publication corresponding to Japanese Patent Publication No. 2007-505241, that is, WO2005 / 026462A1, the country to which the applicant pertaining to the international publication belongs is Japan. The country to which the inventor of the international publication belongs is Japan. The receiving agency of the international publication is the Japan Patent Office. Therefore, although the international publication (WO2005 / 026462A1) is described in a second language (English), the nationality data of the international publication is Japan.

In this case, the second language (English) and the official language (Japanese) corresponding to the nationality data do not match. If the second language and the official language corresponding to the nationality data of the international publication do not match, the computer 20 excludes the international publication from the candidates for the specified international publication. In other words, if the second language and the official language corresponding to the nationality data of the international publication do not match, the computer 20 does not determine the international publication as the specific international publication.

Further, when the first sentence element is "arranged so as to face each other", the computer 20 executes the second step ST2 to obtain the Japanese Patent Publication No. 2005-501784 from among the plurality of corresponding public publications. It is determined as one of the specified publications including the first sentence element (“arranged facing each other”). With reference to the international publication corresponding to Japanese Patent Publication No. 2005-501784, that is, WO2003 / 020594A1, the country to which the applicant pertaining to the international publication belongs is the United States. The country to which the inventor of the international publication belongs is the United States. The receiving agency for the international publication is the United States Patent and Trademark Office. Therefore, the nationality data of the international publication (WO2003 / 020594A1) is the United States.

In this case, the second language (English) and the official language (English) corresponding to the nationality data coincide with each other. When the second language and the official language corresponding to the nationality data of the international publication match each other, the computer 20 includes the international publication as a candidate for the specified international publication. In other words, in the computer 20, the second language and the official language corresponding to the nationality data of the international publication match, and the corresponding publication corresponding to the international publication is in the first language. When the described first sentence element is included, the international publication is determined as a specific international publication.

In the first variant, the computer determines a specific international publication based on nationality data. Therefore, it is highly probable that the second sentence element will be extracted from the English sentence created by a person whose official language is the second language. As a result, the accuracy of machine translation is improved.

The storage device 40 may store the third association data indicating the correspondence between the nationality data and the official language. In the third association data, for example, Japan (nationality data) is associated with Japanese (official language), and Australia (nationality data) is associated with English (official language). In this case, the computer 20 determines whether or not the second language and the official language corresponding to the nationality data of the international publication match, based on the third association data.

In the first modification described above, if the second language and the official language corresponding to the nationality data of the international publication do not match, the computer 20 excludes the international publication from the candidates for the specific international publication. do. Alternatively, if the second language matches the official language corresponding to the nationality data of the international publication, the computer 20 determines the international publication as a priority candidate for the specific international publication, and the second language. If the official language corresponding to the nationality data of the international publication does not match, the international publication may be determined as a non-priority candidate for the specific international publication. In this case, the computer 20 may determine the specific international publication from the priority candidates when there is a priority candidate, and may determine the specific international publication from the non-priority candidates when there is no priority candidate. ..

(Second modification)
The second modification is different from the above-described embodiment or the above-mentioned first modification in that the computer 20 outputs the second sentence element together with the nationality data of the specified international publication. In other respects, the second modification is similar to the embodiment described above or the first modification described above.

In the second modification, the storage device 40 is at least one of the country to which the applicant of the international application belongs, the country to which the inventor of the invention relating to the international application belongs, or the country where the receiving office of the international application is located. I remember one as the nationality data of the international publication. Further, the storage device 40 stores the second association data in which the international publication and the nationality data are associated with each other.

In the second modification, when the computer 20 executes the third step ST3, the computer 20 extracts the nationality data of the specified international publication based on the second association data. For example, when the specified international publication is WO2003 / 020594A1 described above, the nationality data of the specified international publication is "US".

Then, when the computer 20 executes the fifth step ST5, the computer 20 outputs the extracted second sentence element together with the nationality data of the specified international publication. As a result, for example, when the output device 50 is a display device, the second sentence element is displayed together with the nationality data of the specified international publication.

FIG. 3 shows how the output device 50 displays the second sentence element and the nationality data of the specified international publication. In the example shown in FIG. 3, the output device 50 is a display device. A second sentence element (“face each other”) is displayed in the second area 54 of the display device. In addition, nationality data 55 (“US (US)”) is displayed on the display device (more specifically, the second region 54). The nationality data indicates the nationality data of the specified international publication.

As shown in FIG. 3, the first sentence element (“arranged so as to face each other”) may be displayed in the first area 52 of the display device.

Additionally, as shown in FIG. 3, another second text element (eg, "arranged being opposed to each other") may be displayed in the third area 57 of the display device. The display device (more specifically, the third area 57) may display the nationality data 58 (“JP (Japan)” of the specific international publication corresponding to the other second sentence element. ..

In the second modification, the computer outputs the nationality data of the specified international publication. Therefore, the user of the machine translation system can grasp whether or not the second sentence element is extracted from the highly reliable English sentence data.

(Third modification example)
In the third modification, when a plurality of second sentence elements are extracted in the above-mentioned fourth step ST4, the computer 20 describes a claim range from the plurality of the second sentence elements. It differs from the above-described embodiment or the above-mentioned first or second modification in that the second sentence element extracted from the above is preferentially output. In other respects, the third modification is similar to the embodiment described above or the first or second modification described above.

It is highly probable that the translator has taken special care in translating the text in the claims. Therefore, when the second sentence element is extracted from the description column of the claims, it can be said that the accuracy of translation from the first sentence element to the second sentence element is high. Therefore, in the third modification, the second sentence element extracted from the description field of the claims is preferentially output.

A more specific explanation will be given. It is assumed that the first sentence element is "opposed to each other" and the specific international publication determined in the third step ST3 includes the above-mentioned WO2003 / 020594A1 and the above-mentioned WO2005 / 026462A1.

In this case, "face each other" in WO2003 / 020594A1 corresponds to the second sentence element, and the second sentence element corresponds to the description column of the scope of claims of WO2003 / 020594A1 (more specifically, claim 8). It is extracted from the description column). Therefore, the second sentence element is preferentially output by the computer 20.

On the other hand, "arranged being opposed to each other" in WO2005 / 026462A1 corresponds to another second sentence element, and the other second sentence element corresponds to the description column (that is, the claim) of the specification of WO2005 / 026462A1. It is extracted from the part other than the description column of the range). Therefore, the second sentence element is output by the computer 20 in a non-priority manner. In addition, "output with no priority" means that it is not output, or it means that it is output in a less conspicuous manner than the output with priority.

FIG. 4 shows how the output device 50 preferentially displays the second sentence element extracted from the description column of the claims. In the example shown in FIG. 4, the output device 50 is a display device. A second sentence element (“face each other”) is displayed in the second area 54 of the display device. In addition, the display device (more specifically, the second area 54) displays the source data 56 indicating that the extraction of the second sentence element was performed from the description column of the claims. There is.

Further, FIG. 4 shows how the output device 50 displays another second sentence element extracted from the description column of the specification in a non-priority manner. In the third area 57 of the display device, another second sentence element (“arranged being opposed to each other”) is displayed. In addition, the display device (more specifically, the third area 57) displays source data 59 indicating that the extraction of the other second sentence element was performed from the description column of the specification. Has been done. The other second sentence element is displayed below the above-mentioned second sentence element (“face eye other”), that is, in a position less conspicuous than the above-mentioned second sentence element (“face eye other”). There is.

In the third modification, the computer preferentially outputs the second sentence element extracted from the description field of the claims. Therefore, the second sentence element extracted from the highly reliable English sentence data is preferentially used as machine translation.

(Fourth modification)
In the fourth modification, the computer 20 receives the designated technical field data indicating the technical field from the input device 10, and the computer 20 gives priority to the second sentence element extracted from the international publication belonging to the technical field. It is different from the above-described embodiment or any of the above-mentioned first to third modifications. In other respects, the third modification is similar to any of the above-described embodiments or the above-mentioned first to third modifications.

The translation of a term used in the first technical field into a second language may differ from the translation of the term used in the second technical field into a second language. For example, a term that is generally translated as "A" may be translated as "B" in a particular technical field. The fourth modification is a modification corresponding to such a case.

In the fourth modification, the input device 10 inputs the designated technical field data indicating the technical field to the computer 20. The designated technical field data is, for example, subclass data of the International Patent Classification (for example, "A01B" or the like). Further, the storage device 40 stores the fourth association data in which each international publication is associated with the technical field data to which the international publication belongs.

In the third step ST3, when the specific international publication is determined, the computer 20 acquires the technical field data to which the specific international publication belongs based on the above-mentioned fourth association data. Then, the computer 20 determines whether or not the acquired technical field data and the designated technical field data received from the input device 10 match. If they match, the computer 20 preferentially outputs the second sentence element extracted from the specified international publication. On the other hand, if they do not match, the computer 20 outputs the second sentence element extracted from the specific international publication without priority. In addition, "output with non-priority" means not to output, or means to output in a less conspicuous manner than the output with priority.

FIG. 5 shows how the output device 50 preferentially displays the second sentence element extracted from the international publication that belongs to the designated technical field. The designated technical field is a technical field indicated by the designated technical field data received from the input device 10 by the computer 20.

In the example shown in FIG. 5, the output device 50 is a display device. In the first area 52 of the display device, the designated technical field data 521 received by the computer 20 from the input device 10 is displayed. The designated technical field data is data related to the International Patent Classification (for example, subclass data of the International Patent Classification) such as B65D.

Further, in the display device (more specifically, the second region 54), the extraction of the second sentence element (“face each other”) is performed from the document (international publication) to which the technical field belongs. Technical field data 541 indicating that there is is displayed. In the example shown in FIG. 5, the technical field data 541 and the designated technical field data 521 coincide with each other.

Further, FIG. 5 shows how the output device 50 displays the second sentence element extracted from the international publication that does not belong to the designated technical field in a non-priority manner. In the third area 57 of the display device, another second sentence element (“arranged being opposed to each other”) is displayed. In addition, the display device (more specifically, the third area 57) indicates that the extraction of the second sentence element is performed from the document (international publication) to which the technical field belongs. The technical field data 571 shown is displayed. In the example shown in FIG. 5, the technical field data 571 and the designated technical field data 521 are different from each other. Therefore, the other second sentence element is displayed below the above-mentioned second sentence element (“face eye other”), that is, in a position less conspicuous than the above-mentioned second sentence element (“face eye other”). Has been done.

In the fourth modification, the computer preferentially outputs the second sentence element extracted from the international publication that belongs to the designated technical field. Therefore, the second sentence element extracted from the highly reliable English sentence data is preferentially used as machine translation.

In the above-described embodiment or modification, the computer 20 can also receive sentences, paragraphs, etc. from the input device 10. In this case, the computer 20 decomposes the sentence, paragraph, or the like into a plurality of first sentence elements. In this case, the computer 20 performs the above-mentioned second step ST2 to fifth step ST5 for each first sentence element.

Alternatively, the computer 20 performs the above-mentioned second step ST2 to fourth step ST4 for a part of the first sentence elements among the plurality of first sentence elements obtained by the decomposition, and the rest. The first sentence element may be translated using a known machine translation system. For example, when the computer 20 receives from the input device 10 the sentence "fitting portions are provided at both ends of the magnetic shield member, and the magnetic shield member is formed in a substantially I-shaped or Z-shaped shape". Assuming about. In this case, for example, the computer 20 decomposes the sentence into a plurality of first sentence elements. For example, the computer 20 puts the sentence into eight first sentence elements, that is, "of the magnetic shield member", "at both ends", "fitting portions", "the magnetic shield member", and so on. It may be decomposed into "substantially I-shaped", "or", "Z-shaped", and "formed". Next, the computer 20 divides the plurality of first sentence elements into general terms and technical terms. For example, a general term may be a term listed in a predetermined dictionary, and a technical term may be a term not listed in the predetermined dictionary. Here, it is assumed that "a fitting portion is provided" corresponds to a technical term, and the other seven first sentence elements correspond to a general term. In this case, the computer 20 performs the above-mentioned second step ST2 to fourth step ST4 only for the first sentence element including the technical term, and for the remaining first sentence element, a known machine translation system is used. You may perform the translation using. In this way, the computer 20 acquires the second sentence element described in the second language for each first sentence element. That is, the computer 20 acquires a plurality of second sentence elements obtained by translating the plurality of first sentence elements into a second language. Then, the computer 20 outputs the obtained plurality of second sentence elements to the output device 50.

In the above example, the first sentence element that does not include technical terms is translated using a known machine translation system. As a result, the translation time can be shortened. Further, for the first sentence element that does not include technical terms, the translation using a known machine translation system may be more accurate than the translation using the text data of the international publication. Therefore, further improvement in translation accuracy can be expected by properly using a translation system that translates the first sentence element including technical terms and a translation system that translates the first sentence element that does not include technical terms.

(Fifth modification)
In the fifth modification, either the above-described embodiment or the above-mentioned first modification to the fourth modification is performed in that the entire section or sentence including the second sentence element is displayed on the output device 50 (display device). It is different from the modified example. In other respects, the fifth modification is similar to any of the above-described embodiments or the above-mentioned first to fourth modifications.

FIG. 6 is a diagram showing a state in which the entire section including the second sentence element is displayed on the display device which is the output device 50. Here, it is assumed that the first sentence element is "arranged so as to face each other" and the second sentence element is "face each other". In the second area 54 of the display device, the entire section including the second sentence element, that is, "so that a front surface of side first surface and surface of side front phase face" is displayed.

In the example shown in FIG. 6, the display device displays the entire section including the second sentence element. Alternatively, the display device may display the entire sentence including the second sentence element.

In the example shown in FIG. 6, the display device displays the entire section or sentence including the second sentence element. Therefore, the operator can confirm the usage of the second sentence element by looking at the section or sentence including the second sentence element.

(6th modification)
In the second step ST2 described above, the computer 20 may not be able to find the publication containing the first sentence element. The sixth modification is a modification for dealing with such a case.

7 to 9 are flowcharts showing an outline of the machine translation method in the sixth modification. In FIG. 7, the first step ST1, the second-3 step ST2-3, the third step ST3, the fourth step ST4, and the fifth step ST5 are the first step ST1, the second step ST2, and the third step ST3 in FIG. , 4th step ST4 and 5th step ST5 are the same, so the repetition of the 1st step ST1, the 2nd-3rd step ST2-3, the 3rd step ST3, the 4th step ST4, and the 5th step ST5. The explanation of is omitted.

In the second 2-1 step ST2-1, the computer 20 searches for a specific publication including the first sentence element from a plurality of publications described in the first language. If a specific publication containing the first sentence element is found (2-2 step ST2-2: "Yes"), the process proceeds to 2-3 step ST2-3, and the specific publication including the first sentence element is performed. If no publication is found, the process proceeds to the sixth step ST6.

For example, assume that the first sentence element is "the first surface and the second surface are arranged facing each other". In this case, the computer 20 cannot find the public gazette containing the first sentence element (the first surface and the second surface are arranged so as to face each other) (second 2-2 step ST2-2: "No"). ). Therefore, the process proceeds to the sixth step ST6 shown in FIG.

In the sixth step ST6, the computer 20 divides the first sentence element into a plurality of small elements including the first small element and the second small element. When the first sentence element is "the first surface and the second surface are arranged to face each other", the first small element is, for example, "what is the first surface and the second surface", and the second small element. The elements are, for example, "opposed to each other".

In the seventh step ST7, the computer 20 searches for the first specific publication including the first minor element from the plurality of publications described in the first language. If the first specified publication containing the first minor element is found, the process proceeds to the eighth step ST8, and if the first specified publication containing the first minor element is not found, the first minor element is found. Is further divided into a plurality of smaller subelements. Here, it is assumed that the first specific publication including the first minor element has been discovered.

In the eighth step ST8, the computer 20 determines the first specific publication including the first minor element from the plurality of publications (corresponding publications). Here, the computer 20 determines a public gazette including the phrase "what is the first surface and the second surface" as the first specific public gazette. For example, the computer 20 determines the special publication No. 2004-519000 as the first specified publication. This is because Japanese Patent Publication No. 2004-519000 includes the phrase "what are the first and second surfaces" in the column of claims.

In the ninth step ST9, the computer 20 determines the first specific international publication corresponding to the first specific publication from the plurality of international publications. For example, when the first specified publication is Japanese Patent Publication No. 2004-519000, the computer 20 is an international publication corresponding to the first specific publication, Japanese Patent Publication No. 2004-519000, that is, international publication. No. 2002/061481 is determined as the First Specified International Publication. The ninth step ST9 may be executed by the computer 20 functioning as a second determination means 26.

In the tenth step ST10, the computer 20 extracts a third sentence element corresponding to the first minor element and described in the second language from the first specific international publication. For example, when the first specified international publication is International Publication No. 2002/061481 and the first subelement is the phrase "what are the first and second surfaces", the computer 20 has a third sentence element. As, the phrase "the first and second computers" is extracted. The tenth step ST10 may be executed by the computer 20 functioning as the extraction means 28.

In the eleventh step ST11, the computer 20 searches for the second specified publication including the second minor element from the plurality of publications described in the first language. If the second specified publication containing the second minor element is found, the process proceeds to the twelfth step ST12, and if the second specified publication containing the second minor element is not found, the second minor element is found. Is further divided into a number of smaller subelements. Here, it is assumed that the second specified publication containing the second minor element has been discovered.

In the twelfth step ST12, the computer 20 determines the second specific publication including the second minor element from the plurality of publications (corresponding publications). Here, the computer 20 determines the public gazette including the phrase "arranged so as to face each other" as the second specific public gazette. For example, the computer 20 determines the publication No. 2005-501784 as the second specified publication. This is because Japanese Patent Publication No. 2005-501784 includes the phrase "arranged so as to face each other" in the column of claims.

In the thirteenth step ST13, the computer 20 determines the second specified international publication corresponding to the second specified publication from among the plurality of international publications. For example, when the second specified publication is Japanese Patent Publication No. 2005-501784, the computer 20 is an international publication corresponding to the special publication No. 2005-501784, that is, the international publication. No. 2003/20594 is determined as the Second Specified International Publication. The thirteenth step ST13 may be executed by the computer 20 functioning as a second determination means 26.

In the 14th step ST14, the computer 20 extracts the fourth sentence element described in the second language corresponding to the second minor element from the second specified international publication. For example, when the second specified international publication is International Publication No. 2003/20594 and the first subelement is the phrase "arranged so as to face each other", the computer 20 uses "face" as the fourth sentence element. Extract the phrase "each other".

In the fifteenth step ST15, the computer 20 outputs the second sentence element including the third minor element and the fourth minor element as a translation of the first sentence element. Here, the computer 20 combines a third subelement (“the first and second planes”) and a fourth subelement (“face touch another”) to form a second sentence element (“the first and second planes face”). create "each other"). Then, the computer 20 outputs the second sentence element to the output device 50 as a translation of the first sentence element. For example, when the output device 50 is a display device, the second sentence element is displayed on the display device.

In the machine translation system and the machine translation method in the sixth modification, when a first sentence element including a plurality of words is input, first, a text search is executed using the entire first sentence element as a search key. As a result, when a specific publication including the entire first sentence element is found, translation is executed using the specific publication. Therefore, the accuracy of translation is improved. On the other hand, if the specific publication including the entire first sentence element is not found, the first sentence element is divided into a plurality of small elements. Then, a text search is executed using each subelement as a search key. Therefore, the probability of not being able to translate is reduced.

In the above-described embodiment or each modification, an example in which the first sentence element is a sentence element longer than a word has been described. However, the embodiment or each modification is not limited to this example. That is, the first sentence element may be a word. However, if the first sentence element is a phrase, clause, or sentence that is longer than the word, and the phrase, clause, or sentence that is longer than the word is used as the search key, it is better to execute the text search. The accuracy is high. In addition, the phrase includes, for example, a prepositional phrase, a noun phrase, a verb phrase, an adaptive phrase, or an advertial phrase.

Further, in the above-mentioned sixth modification, an example in which the first subelement is a sentence element longer than a word has been described. However, the sixth modification is not limited to this example. That is, the first subelement may be a word. Similarly, in the sixth modification described above, an example in which the second subelement is a sentence element longer than a word has been described. However, the sixth modification is not limited to this example. That is, the second subelement may be a word.

The present invention is not limited to each of the above embodiments, and it is clear that each embodiment can be appropriately modified or modified within the scope of the technical idea of the present invention. In addition, the various techniques used in each embodiment or modification can be applied to other embodiments or modifications as long as there is no technical contradiction.

For example, the nationality data of an international publication may be determined based on the country to which the applicant belongs, may be determined based on the country to which the inventor belongs, or may be determined based on the receiving authority of the international application. It may be decided. Alternatively or additionally, the nationality data of the International Publication may be made by analyzing the specification of the International Publication. For example, if the specification of an international publication uses expressions frequently used by native Americans, the nationality data of the international publication may be determined to be the United States.

1: Machine translation system 10: Input device 20: Computer 24: First determination means 26: Second determination means 28: Extraction means 40: Storage device 50: Output device 52: First area 54: Second area 55: Nationality data 56: Source data 57: Third area 58: Nationality data 59: Source data 521: Designated technical field data 541: Technical field data 571: Technical field data

Claims (7)

A machine translation system that translates a first sentence element described in a first language into a second sentence element described in a second language.
An input device for inputting the first sentence element into a computer,
The computer that receives the first sentence element from the input device,
Equipped with a storage device,
The storage device is
The text data of the international publication written in the second language and
It corresponds to the international publication and stores the text data of the corresponding publication described in the first language.
Upon receiving the first sentence element from the input device, the computer determines a specific publication including the first sentence element from the plurality of corresponding publications.
The computer determines a specific international publication corresponding to the specific publication from a plurality of the international publications.
The computer extracts the second sentence element corresponding to the first sentence element from the specified international publication.
The computer is a machine translation system that outputs the extracted second sentence element as a translation of the first sentence element. The storage device stores the international publication in association with the nationality data of the international publication.
The machine translation system according to claim 1, wherein the computer determines the specific international publication based on the nationality data of the international publication. The machine translation system according to claim 1 or 2, wherein the computer outputs the second sentence element together with the nationality data of the specified international publication. When a plurality of the second sentence elements are extracted from the plurality of specified international publications, the computer is extracted from the description column of the claims among the plurality of the second sentence elements. The machine translation system according to any one of claims 1 to 3, which preferentially outputs the text elements. A machine translation method that translates a first sentence element described in a first language into a second sentence element described in a second language.
(A) A receiving process in which the computer receives the first sentence element from the input device.
(B) A first determination step in which the computer determines a specific publication including the first sentence element from a plurality of publications described in the first language.
(C) A second determination step in which the computer determines a specific international publication corresponding to the specific publication from a plurality of international publications described in the second language.
(D) An extraction step in which the computer extracts the second sentence element corresponding to the first sentence element from the specified international publication.
(E) A machine translation method comprising an output step in which the computer outputs the extracted second sentence element as a translation of the first sentence element. In the receiving process, the computer receives sentence data including the plurality of the first sentence elements, and receives the sentence data.
The computer divides a plurality of the first sentence elements into a sentence element including a technical term and a sentence element not including the technical term.
The machine translation method according to claim 5, wherein the computer executes the step (B), the step (C), and the step (D) only for a sentence element including the technical term. (F) A search step in which the computer searches for the specific publication including the first sentence element from a plurality of publications described in the first language.
(G) A division step in which the computer divides the first sentence element into a plurality of small elements including the first small element and the second small element.
(H) A step of determining a first specific publication including the first subelement from the plurality of publications described in the first language by the computer.
(I) A step in which the computer determines a second specific publication including the second minor element from the plurality of publications described in the first language.
(J) A step in which the computer determines a first specific international publication corresponding to the first specific publication from a plurality of international publications described in the second language.
(K) A step in which the computer determines a second specified international publication corresponding to the second specified publication from a plurality of the international publications described in the second language.
(L) A step in which the computer extracts a third minor element corresponding to the first minor element and described in the second language from the first specified international publication.
(M) A step in which the computer extracts the fourth minor element corresponding to the second minor element and described in the second language from the second specified international publication.
(N) The computer further comprises a step of outputting the second sentence element including the third small element and the fourth small element as a translation of the first sentence element.
When the specified publication is found in the search step, the computer executes the step (B), the step (C), and the step (D).
If the specified publication is not found in the search step, the computer (20) may use the step (G), the step (H), the step (I), the step (J), and the step (K). The machine translation method according to claim 5 or 6, wherein the step (L), the step (M), and the step (N) are performed.

Images