JP2006244236A - Machine translation program, and machine translation device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To substantially reduce a load of a user for obtaining a proper corrected translation in a machine translation program and a machine translation device translating a natural language into another natural language. <P>SOLUTION: A sentence of a first language to be translated, and a translated sentence of a second language are prepared. Each translation in the translated sentence of the second language is coordinated with each word equivalent to the translation in the sentence of the first language to be translated. It is determined whether or not each of the coordinated word of the first language is a compound word. Words determined not to be compound words are compared with each other, words determined to be compound words are compared with each other, and the words determined not to be compound words are compared with the words determined to be compound words. In comparison pairs of the words, words matching at least partially are extracted, and words of the first language by the coordination including the extracted words are displayed along with the translations coordinated with respect to the words. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a machine translation program and a machine translation apparatus for translating a natural language into another natural language, and in particular, a machine translation program and a machine translation suitable for reducing a burden for a user to obtain an appropriate correction translation. Relates to the device.

  With the progress of technology, various documents have been digitized, and accompanying this, large-scale translation dictionaries (parallel translation dictionaries) have been developed and constructed one after another. The number of words that can be used as headwords is increasing every day. The increase in the number of headwords is one of the indicators of the quality of the dictionary, which contributes to the increase. Recently, however, it has been pointed out that the mere number is rather harmful. It was.

  First, since a dictionary is usually compiled by a plurality of people, consistency between related headwords and their translations is not necessarily taken. Thus, the dictionary user needs to make these adjustments. Using multiple types of dictionaries doubles this problem. Secondly, the expression (character string) that was previously given as an example of the main heading is placed in the heading, so that the meaning of the sentence is interpreted with priority over that character string, and the correct interpretation The evil that it was not possible has come out.

  For example, if a sequence of words (word strings) that are interpreted literally in a normal scene has a limited meaning in a specialized field or a specific scene, this word string is used as a headword, and its special meaning If only the translated word is listed, the beginner may try to interpret the sentence using this translated word and end up with no meaning. Similar problems arise when machine translation uses such a large dictionary. However, the problem is more serious because there is no flexible adjustment by humans. The second problem described above is often solved by a person who is familiar with the foreign language to some extent. That is, if it is determined that the meaning of a certain headword does not match the context, it is predicted that the meaning of each word will be combined and a literal interpretation will be attempted. However, in machine translation, the translated word after the heading is unconditionally given priority.

  In addition, if a dictionary constructed by the user is used together, these problems are doubled. This is because the user usually puts an expression that would not normally be a headword as a headword and, on the other hand, has only one translation. However, there is a possibility that this translated word is not a useful translated word except where the user feels the necessity of registration.

  On the other hand, information is frequently acquired on the Internet through the WWW (world wide web). Most of the information available on the Internet is in English, and if English is not your native language and you are not good at it, users have typically tried to do machine translation to understand the meaning of the original text. . In general, the quality of machine translation that translates from foreign language to Japanese language has reached a level that can be tolerated, and it has become normal to output sentences that are syntactically correct.

  However, it may be inferior to human translation in terms of interpretation. One factor is the ambiguity of the target. "Coherence, cohesion" between words in a sentence is maintained not only by directives but also by how the terms are used. The latter specifically refers to a state where, for example, the same term is used repeatedly. However, when this document is translated using a translation dictionary (parallel translation dictionary), even if it is the same term and its surrounding word is listed as a compound word heading, the translated word is applied for the following reasons. , It is not always translated with the same translation as when it occurred alone.

  That is, in machine translation, in order to increase the efficiency of syntax analysis, a dictionary of compound words (including idiomatic phrases) is first drawn (A), and then the remaining part is analyzed and assigned (B). There, the translation of the compound word is uniquely determined. Since the subsequent operation of B is performed independently of A, the relevance of the translation between the portion to be A and the portion to be B is often lost in machine processing.

  If you read only the output of this machine translation without referring to the original text, it will be easy to lose this relevance and hinder interpretation. Therefore, in order to create a translation that is easier to understand, it is necessary to extract a word string that needs to be adjusted that appears in the sentence, perform an exhaustive check, and make adjustments. However, it is a great burden for translators to do all this manually. There is a need for a mechanism that allows translators to devote themselves to the truly intelligent task of determining translations.

  There are certainly attempts to make it easier to review the translation results. However, it is a problem only when the translation of A is different when a plurality of translations are possible for the expression (A) in the first language in question. For example, there is the following Patent Document 1 (Japanese Patent Laid-Open No. 2001-282787). This is so that when a user selects a translation into a part of the displayed translation result sentence, the part where the translation is selected stands out. To display. That is, the relationship between different headwords handled by the present application is not a problem.

  A similar attempt is to display words in list form. This is widely known as a “word book” and is a well-known example in various commercially available electronic dictionaries and machine translation systems. However, in these cases, the candidates for the list are unknown words or referenced words, which are different from the subject in the present application.

In summary, the link between terms that existed in the original text in machine translation has not always been reflected in the translation. Even in manual translation, the consistency of translated words is not always guaranteed unless the translations are reviewed once. In addition, no translation mechanism has been provided, and no mechanism has been provided that can be performed without imposing a heavy burden on the translator. As described above, at present, the user's needs for presenting information necessary for correcting the translation in an easy-to-understand manner are not sufficiently satisfied.
JP 2001-282787 A

  Thus, in conventional machine translation, translation was performed without considering the relationship between words in the first language and the corresponding translations, so the user who became the reader felt restrictions on understanding the translation. . Nor did it provide an appropriate framework for the necessary modifications to improve this.

  The present invention has been made in consideration of the above-described circumstances, and in a machine translation program and machine translation apparatus that translates a natural language into another natural language, the burden on the user to obtain an appropriate corrected translation is greatly reduced. An object of the present invention is to provide a machine translation program and a machine translation apparatus capable of performing the same.

  In order to solve the above problem, a machine translation program according to a first aspect of the present invention includes a first step of preparing a sentence in a first language to be translated and a sentence in a second language to be translated, A second step of associating each translated word in the translated second language sentence with each of the phrases corresponding to the translated word in the first language sentence to be translated; and A third step of determining whether or not each word is a compound word; words and phrases determined not to be compound words by the determination; words and phrases determined to be compound words by the determination; or compounded by the determination A fourth step of comparing a phrase determined not to be a word and a phrase determined to be a compound word by the determination; and a fifth step of extracting a phrase that at least partially matches in the comparison pair of the phrases. And flop, the phrase in the first language by the association which includes the extracted word, characterized in that to execute a sixth step of displaying together with the correlated obtained translation for that word to the computer.

  That is, this machine translation program pays particular attention to compound words with corresponding translations among the first language words to be translated, and is common to the first language sentences including partial words of the compound words. Phrases appearing in are extracted. And the phrase (for example, headword) of the 1st language containing the extracted phrase can be displayed with the matched translation. Therefore, in the first language, it becomes easy to compare common phrases with their corresponding translations, and the user can easily perform operations such as correcting the translations.

  The machine translation program according to the second aspect of the present invention provides a first step of preparing a sentence in a first language to be translated and a sentence in a second language to be translated, and the translated second language. A second step of associating each translated word in the sentence with each of the phrases corresponding to the translated word in the sentence in the first language to be translated, and designation of an arbitrary phrase for the sentence in the first language to be translated A third language prompting the user and a word in the first language by the correspondence including the specified word or phrase, or a utilization type or derivative of the word, are associated with the word in the first language. And causing the computer to execute a fourth step of displaying together with the translated word.

  That is, in this machine translation program, an arbitrary one can be designated as the first language phrase to be translated. Then, it is possible to display a phrase (for example, a headword) in the first language including the designated phrase (further utilization and derivative forms) together with the corresponding translated word. Therefore, in the first language, it becomes easy to compare common phrases with their corresponding translations, and the user can easily perform operations such as correcting the translations.

  The machine translation program according to the third aspect of the present invention includes a first step of preparing a sentence in a first language to be translated and a sentence in a second language to be translated, and the translated second language. A second step of associating each translated word in the sentence with each of the words corresponding to the translated word in the sentence in the first language to be translated, and specifying an arbitrary phrase for the translated sentence in the second language A third step of prompting the user, and the second language phrase by the correspondence including the specified phrase or the utilization form or derivative form of the phrase is associated with the second language phrase And causing the computer to execute a fourth step of displaying together with the phrase of the first language.

  That is, in this machine translation program, an arbitrary one can be designated as the translated second language phrase. Then, it is possible to display a second language phrase (for example, a headword) including the designated phrase (further utilized or derived phrase) together with the associated first language phrase. Therefore, in the second language, it becomes easy to compare the common words and phrases with the corresponding words in the first language, and the user can easily perform operations such as correcting the translation. In other words, this machine translation program is opposite to the above-mentioned “another machine translation program” in that the words that can be specified are before and after translation.

  The machine translation apparatus according to the present invention also includes a first means for preparing a first language sentence to be translated and a translated second language sentence, and each translated word in the translated second language sentence. And a second means for associating each phrase corresponding to the translated word in the sentence of the first language to be translated, and whether or not the phrase of the first language thus correlated is a compound word. A third means for determining, words determined not to be compound words by the determination, words determined to be compound words by the determination, or words determined to be not compound words by the determination and the determination A fourth means for comparing a phrase determined to be a compound word; a fifth means for extracting a phrase that matches at least partially in the comparison pair of the phrases; and the correspondence including the extracted phrase 1st language Phrases and characterized by comprising a sixth means for displaying together with the correlated obtained translation for that word.

  Another machine translation device according to the present invention provides a first means for preparing a sentence in a first language to be translated and a sentence in a second language to be translated, and a sentence in the translated second language. A second means for associating each translated word with each phrase corresponding to the translated word in the sentence in the first language to be translated, and prompting the user to specify an arbitrary phrase for the sentence in the first language to be translated The translation of the third language and the translation of the first language word including the designated word or phrase, or the first language phrase by the correspondence including the utilization form or derivative form of the word, with the word of the first language. And a fourth means for displaying together.

  Still another machine translation device according to the present invention provides a first means for preparing a first language sentence to be translated and a translated second language sentence, and the translated second language sentence. A second means for associating each translated word in the sentence with each of the words corresponding to the translated word in the sentence in the first language to be translated, and designation of an arbitrary word or phrase for the translated sentence in the second language to the user A third means for urging and the second language phrase by the association including the specified word or phrase, or a phrase of the inflected form or derivative of the word, associated with the second language phrase And a fourth means for displaying together with the phrase of the first language.

  These machine translation devices also have the same operations and effects as the above machine translation programs. Each unit included in the machine translation device may be configured by hardware such as a microprocessor, a memory, and a hard disk, and basic software such as an operating system operating on the hardware, an application program, and the like. Is possible.

  According to the present invention, in a machine translation program and a machine translation apparatus that translate a natural language into another natural language, a burden for a user to obtain an appropriate corrected translation can be greatly reduced.

  As an embodiment of the machine translation program according to the first aspect of the present invention, for the word / phrase determined to be a compound word by the determination, a step of dividing the compound word, each of the divided words and the compound word not determined by the determination And converting the obtained word / phrase into an original word / phrase to obtain an original word string, and causing the computer to execute the step, wherein the fourth step converts the obtained original word string into a compound word according to the determination. A word in the first language by the correspondence, in which the sixth step includes a phrase in the utilization form of the extracted phrase, which is used instead of a phrase determined to be present or a phrase which is not a compound word by the determination Can be displayed together with the corresponding translation for the word. This makes it easy to compare the common phrases and their corresponding translations except for the utilization forms in the first language. Note that the “word string” includes a case where there is only one word.

  Further, as an embodiment in the same case, for the word / phrase determined to be a compound word by the determination, a step of dividing the compound word, and each of the divided words and the word / phrase determined not to be a compound word by the determination are derived. Converting the previous word / phrase to obtain a word string before derivation, further causing the computer to execute the word / phrase, wherein the fourth step determines that the obtained word string before derivation is a compound word by the determination. Alternatively, it is used instead of a word that is determined not to be a compound word according to the judgment, and the sixth step uses a word in the first language by the association including a word derived from the extracted word as the corresponding word. It is possible to display it together with the associated translated words. This makes it easy to compare the common phrases and their corresponding translations except for the derivative forms in the first language. Here, the “word string” includes the case of only one word.

  Further, as an embodiment in a similar case, for each translation in the translated second language sentence, the computer further executes a step of acquiring a translation type, and the third step is the association. Only when the translation of the second language corresponding to the phrase of the first language is a specific translation type in the acquired translation type, it is determined whether or not it is a compound word. It can be. According to this, in the first language, it is possible to automatically select and present to the user when there is a particularly high necessity for comparison between common phrases and their corresponding translations.

  Here, the specific translation type may be a transliteration, a one-to-one translation, a partial translation, or an explanatory note.

  Further, as an embodiment in the same case, the sixth step includes a sentence in which the phrase of the first language is used together with the phrase of the first language and the translation corresponding to the phrase of the first language. Can be displayed. By displaying the sentence, the display becomes easier to understand for the user.

  Here, the display of the sentence in which the phrase of the first language is used can include an indication that the phrase of the first language is a keyword. According to such a display, the display becomes easier to understand for the user.

  As an embodiment of the program according to the second aspect of the present invention, the fourth step may not function when the specified word / phrase is a specific translation in the translated word. This is a mode for canceling display in advance when no comparison is necessary.

  As an embodiment of the program according to the third aspect of the present invention, the fourth step does not function when the designated phrase is a specific phrase in association with the first language. It is good also as. This is also a mode for canceling the display when the comparison is not necessary.

Based on the above, embodiments of the present invention will be described below with reference to the drawings.
(First embodiment)
FIG. 1 is a block diagram showing a configuration of a machine translation apparatus according to an embodiment of the present invention. This machine translation apparatus is provided with a machine translation program, whereby, as shown in the figure, an input unit 11, a system control unit 12, an editing unit 13, a translation unit 14, a translation dictionary unit 15, a display unit 16, a document storage unit Each part of 17 functions. The relationship between the machine translation program and the machine translation apparatus is the same in the following embodiments. In the following description, including other embodiments, the first language is described as English and the second language is described as Japanese.

  The input unit 11 inputs sentences and sentences in the first language to be translated through communication means such as the Internet, a keyboard, and the like. Also, a command for instructing an operation as a machine translation device is input. The input sentence or command is sent to the system control unit 12.

  The system control unit 12 controls the entire machine translation apparatus in an integrated manner. For example, a sentence or sentence obtained from the input unit 11 is sent to the display unit 16 for display on the display unit 16, sent to the editing unit 13 for editing, sent to the translation unit 14 for translation, or a document storage unit for storage. 17 and so on. Further, the operations of the editing unit 13, the translation unit 14, the translation dictionary unit 15, the document storage unit 17, and the display unit 16 are controlled in accordance with commands from the input unit 11. For example, data in the translation dictionary unit 15 is taken out and sent to each unit, a sentence to be translated is selected from a sentence input according to a command, or a dictionary to be worked out is selected from the dictionary of the translation dictionary unit 15.

  In addition to editing the sentence in the first language selected by the system control unit 12, the editing unit 13 displays the target editing displayed on the display unit 16, that is, in response to an instruction such as a command input from the input unit 11. It controls the movement of the cursor above and the display of the translated text obtained by the translation processing in the translation unit 14 on the display unit 16 via the system control unit 12.

  The translation unit 14 performs translation processing on the sentence in the first language selected by the system control unit 12 while referring to the dictionary of the translation dictionary unit 15 selected by the system control unit 12. The translation dictionary unit 15 stores and holds various dictionaries used when the translation unit 14 performs translation processing. The contents of various dictionaries included in the translation dictionary unit 15 will be described later.

  The display unit 16 displays, under the control of the system control unit 12, a first language sentence (original sentence) input from the input unit 11, a second language sentence (translation sentence) translated by the translation unit 14, and the like. Is. The document storage unit 17 stores and holds the first language sentence and the second language sentence independently under the control of the system control unit 12, and also stores the first language and the second language as a result of the translation. Data is stored and held in association with a predetermined data.

  As illustrated, the translation dictionary unit 15 includes a utilization change dictionary 15a, an analysis grammar dictionary 15b, a word / idiom dictionary 15c, a conversion grammar dictionary 15d, a generation grammar dictionary 15e, and a morpheme generation grammar dictionary 15f. The utilization change dictionary 15a is a dictionary for converting words and idioms whose endings and the like are changed into original forms of the first language. The analysis grammar dictionary 15b is a dictionary in which a grammar for analyzing the first language is stored. The word / idiom dictionary 15c is a dictionary in which translated words in the second language corresponding to the words / idioms in the first language are stored together with the part of speech information. The conversion grammar dictionary 15d is a dictionary in which structure conversion rules between the first language and the second language are stored. The generation grammar dictionary 15e is a dictionary for determining the structure of the second language sentence. The morpheme generation grammar dictionary 15f is a dictionary for completing a translated sentence by changing the word form such as the end of the second language.

  Next, the operation of the machine translation apparatus shown in FIG. 1 will be described with reference to FIG. FIG. 2 is a flowchart showing an operation flow of the machine translation apparatus shown in FIG. First, an original sentence (a sentence in a first language to be translated) is taken into this machine translation apparatus using the input unit 11 (step 21). The captured text is displayed on the display unit 16 by the system control unit 12 and stored / held in the text storage unit 17.

  When the first language sentence is displayed on the display unit 16, the cursor is then positioned on the sentence to be translated (step 22). The first is usually the first sentence. Next, the user inputs a command for instructing execution of an event from the input unit 11 (step 23). If the input command is not “end” or “translation”, the processing corresponding to the command is executed (step 29), and the process returns to step 23 where the command can be input again. When the input command is “end” (Y in step 24), the process proceeds to an operation for creating a “translation list” described later (from step 30 to step 33). If the input command is “translation” (Y in step 25), the process proceeds to a translation process (step 26).

  The movement of the cursor that designates the translation target is automatically moved to the next sentence when the translation of the sentence is completed in order from the first sentence. In addition, the movement of the cursor can be specified by a command. In this case, the cursor moves by executing the process of step 29. Internally, the editing unit 13 moves the cursor on the display unit 16 via the system control unit 12 according to the command. In this embodiment, a sentence in which the cursor is positioned (for example, the cursor is positioned at the head thereof) is a translation target when a translation command is instructed. This is because the movement of the cursor corresponds to, for example, when it is desired to translate not only the sentence at the current cursor position but also the sentence before and after the original sentence.

  The translation processing (step 26) is performed by the system control unit 12 in an integrated manner. A known process can be applied to the translation process itself. However, along with the translation process, a correspondence relationship between the phrase of the first language sentence and the translated phrase of the second language sentence is derived, and this correspondence relationship is also stored and stored in the document storage unit 17. When these processes are completed, the translation process result is displayed on the display unit 16 (step 27). Further, the cursor for designating the translation target is moved to the next sentence (step 28). Then, the process returns to step 23 where the command can be input again. Thereafter, the translation process can proceed in the same flow.

  A specific example will be described with reference to FIGS. FIG. 3 is a diagram showing a display example of the original sentence and the translated sentence on the machine translation apparatus shown in FIG. FIG. 4 is a diagram showing the correspondence between the original sentence and the translated sentence in the original sentence and the translated sentence shown in FIG. In the display of the original sentence (first language) shown in FIG. 3A, in the state where the first sentence has been translated, “For | the | stability | of | light-weight | composite low | centrifugal force | requires | a | large | diameter | bearing |. ”is divided into 14 elements, excluding punctuation. Here, “composite blades” and “centrifugal force” consisting of multiple words are treated as one unit each because they are registered as compound words in the translation dictionary headwords. is there.

  Next, of the 14 elements divided here, with respect to content words that are not function words such as prepositions and articles, what kind of dictionary headwords are applied and what translations are applied are part of speech information. In addition, the data is generated as a pair of data. At this time, information indicating whether or not the content word is a compound word (composite word) is further given. This can be judged as a compound word in English due to the presence of spaces and hyphens.

  In the above example, 4 words “for, the, of, a” are function words, and 8 words “stability, light-weight, composite blades, low, centrifugal force, requires, large, diameter, bearing” are content words. . Among these, “light-weight, composite blades, centrifugal force” are compound words, and among these content words, the two headwords “composite blades” and “requires” have different headwords (the original form is different). It also remembers that it exists. That is, since the former is a plural form of a noun, “composite blade” is referred to, and since the latter is the third person singular of the verb, the original form “require” is referred to.

  In addition, the compound word here is used in a broader meaning than what is generally understood as a compound word, and indicates that the headword consists of a plurality of words. Therefore, it includes idioms and idioms such as “full of beans”. This is because, in general, idioms and proper nouns may be interpreted as meanings as idioms or idioms, but in some cases they must be interpreted literally. In the previous example, if a “bean” alone occurs in a document, the possibility is high. Also, both its unique meaning and literal meaning may be required. This is a kind of word play (pun), a phenomenon seen in literature.

  By the processing as described above, a correspondence list as shown in FIG. 4A can be generated. This list is sent to the document storage unit 17 together with the sentence number and stored / saved. When the above processing is completed for the first sentence shown in FIG. 3A, the translation result of the sentence such as the first sentence in FIG. 3B is displayed on the display unit 16 (step 27). Next, the cursor is moved to the next sentence (step 28), and translation processing for this sentence is performed.

  Here, as in the following, parsing is performed as follows: “Thus |, | strict | quality control | is maintained | on | the | Of these, the five content words are “thus, strict, quality control, is maintained, bearing geometry”. Furthermore, since “is maintained” is a passive voice of the verb, it is converted to its original form and the translation is obtained from the one word “maintain”. Therefore, only two words “quality control” and “bearing geometry” are determined to be compound words.

  The association list obtained in this way is as shown in FIG. 4B, for example. This list is stored and saved in the document storage unit 17 as in the list of FIG. When the translation of the second sentence shown in FIG. 3A is completed, a translation result of the sentence such as the second sentence shown in FIG. 3B is displayed. In the case of this specific example, since there is no sentence after this, an end command is input in step 24.

  When the above processing is completed, this translation apparatus asks the user whether or not it is necessary to create a “translation list” (you may automatically shift to the next processing without asking). The “translation list” is created as follows. This will be described with reference to FIG. 2 again. First, a compound word (however, the original form) is taken out from the correspondence list accumulated in the document storage unit 17 during the translation process, and divided into words (step 30).

  In the example shown in FIG. 3 and FIG. 4, the five words “light-weight, composite blade, centrifugal force, quality control, bearing geometry” are targeted, and “light-weight” is “light” and “weight”. “Composite blade” is two words “composite” and “blade”, “centrifugal force” is two words “centrifugal” and “force”, “quality control” is two words “quality” and “control” “Bearing geometry” is divided into two words “bearing” and “geometry”. This gives a total of 10 words: “light, weight, composite, blade, centrifugal, force, quality, control, bearing, geometry”.

  Next, a word other than the compound word, that is, a phrase (however, a prototype) composed of one word is taken out (step 31). Here, nine words “stability, low, require, large, diameter, bearing, thus, strict, maintain” are targeted. Then, any known pattern matching technique is used to determine whether the individual words obtained in step 31 are present in the set extracted in step 30, that is, whether there are at least partially matching words. Search and extract using (step 32).

  In the example shown in FIGS. 3 and 4, “bearing” is included in both. Therefore, “bearing” is extracted, and “bearing feature” of the second sentence having “bearing” of the first sentence and “bearing” as a component is created as a “translation list” as shown in FIG. It is displayed (step 33). As a result, the user can understand that the word “bearing” is translated into “bearing” and “bearing” in the same document. By viewing this display, the user can determine whether the translation needs to be adjusted. In addition to the display shown in FIG. 5, the sentence in which it occurs may be displayed simultaneously. This is because it is often easier to adjust the translation while looking at the context.

  In the above example, the comparison pair is a set of one word and a set of compound words in the first language. However, it is also meaningful to compare one word and compound words. First, comparison within a set of one word will be described using the example sentence shown in FIG. FIG. 6 is a diagram showing another example of the display of the original text and the translation text prepared on the machine translation apparatus shown in FIG. As a background explanation, even if the same one-word heading is drawn in machine translation, the same translation is not given because the translation has various translation rules for determining the translation with reference to the other words that co-occurred for that headword. Sometimes.

  The original text shown in Fig. 6 (a) is “Gears | are mounted | on | a | shaft |. | This | .. ”and a correspondence list as shown in FIG. 7 is stored and stored in the document storage unit 17.

  Therefore, in step 30, the compound word (however, the original form) is divided into word units, and in step 31, after extracting a word (however, original form) consisting of one word, in step 32, a set of one word (“gear, shaft, machine, “contain, arrange, drive, large, gear”) are examined, and those having a frequency of 2 or more are extracted. In this example, “gear” corresponds to that. Then, the correspondence list shown in FIG. 7 is referred to, and when different translations are assigned to these, the list is set as a display target (step 33). “Gear” in FIG. 6A is a display target because two translations of “gear” and “gear” are used as shown in FIGS. 7 and 6B. In this case, the “translation list” is displayed as shown in FIG. 8, for example. FIG. 8 is a diagram showing a display example of a bilingual list (by comparing one word) obtained from the correspondence shown in FIG.

  Next, comparison between compound words will be described. As a specific example, the word strings “capacity increase” and “standard capacity” are included in the same document, which match the dictionary headwords, and are assigned the translations of “capacity increase” and “standard capacity”, respectively. An example will be described. In this case, when divided into one word in step 30, four words “capacity, increase, standard, capacity” are obtained. Next, one having a frequency of 2 or more for one word is extracted. Here, “capacity” corresponds to this. Therefore, a compound word including the word “capacity”, that is, “capacity increase” and “standard capacity” are displayed.

  Furthermore, it is assumed that a sentence continues after these two sentences and "capacity increase" appears again. If the translated words there are the same, for example, as shown in FIG. 9A, if they are different, for example, as shown in FIG. 9B, a “translation list” is displayed. FIG. 9 is a diagram showing a display example of a bilingual list (by comparison between compound words) obtained from the association list shown in FIG. In FIG. 9B, the numeral 2 is listed at the end of the headword “capacity increase”, which indicates that there are two translated words “capacity increase” and “capacity increase”.

  The user can easily determine whether the translation needs to be adjusted by using the “translation list” as shown in FIGS.

(Second Embodiment)
Next, a second embodiment of the present invention will be described. As already described in the first embodiment, the extraction object can be expanded to the utilization form by making the extraction object a complete match of the original word. An example of extraction including such words is first supplemented and described as one of the second embodiments. In order to perform this processing, a utilization change dictionary 15a is particularly required for converting a word / idiom having a change in the end of the translation dictionary unit 15 into its original form.

  As an example, the word “communications service” and “communication speed” are included in the same document, and these match the dictionary headwords, and the translations “communication service” and “communication speed” are associated respectively. Suppose that At this time, if words are divided in step 30, four words “communications, service, communication, speed” are obtained. Further, among these, those that are not in the original form are converted into the original form with reference to the utilization change dictionary 15a. In this example, “communications” is a noun plural, so it is converted to the singular “communication”.

  Therefore, since the search target is four words “communication, service, communication, speed”, those having a frequency of 2 or more are extracted and words are extracted. Here, “communication” is applicable. Therefore, a compound word using “communication”, that is, “communications service” and “communication speed” are displayed. FIG. 10 shows an example of the display. FIG. 10 is a diagram illustrating a display example of a “translation list” obtained by comparing phrases before use. In addition to the plural and singular nouns, inflection words include possession forms of nouns (eg, Avogadro number and Avogadro ’s number), conjugation forms of verbs such as present participle and past participle.

  Next, a description will be given of an embodiment in which a derivative word based on the same concept as in the case of a usage word is included and extracted. For this process, a word source dictionary 15g (not shown) in which word words (etymology, derivation) of words in the first language are newly written is necessary.

  For example, the word “managerial technique” and “management approach” are included in the same document, and these match the dictionary headwords, and the translations of “management technology” and “management approach” are associated respectively. Suppose that At this time, if words are divided in step 30, four words “managerial, technique, management, approach” are obtained. Next, when the etymology of each phrase is examined using the etymology dictionary 15g, “managerial” is an adjective of the noun “manager” derived from the verb “manage”, and “management” is the verb “manage”. Information such as a nounization can be obtained. Thus, it is understood that “managerial” and “management” are words having the same etymology. Therefore, “managerial technique” and “management approach” are displayed on the “translation list” with “manage” as an intermediary.

  Another example is similar processing within one word. For example, it is assumed that two words “establish, reestablish” appear in the same document, both of which are listed as dictionary headwords, and translated into “establish” and “rebuild”, respectively. Here, using the etymology dictionary 15g, “reestablish” indicates that the verb “establish” is prefixed with “re” and is derived. As a result, it is understood that “establish” is a word from which “reestablish” is derived, so these are the display objects to be listed in the “translation list”.

  There are various types of derivatives other than those described here. Even if it takes a single suffix, it is a derivative of the verb and has a particularly strong coining power. -Able, -ible (deserved to be passive ... ”(Which means“ ... easy to do ”), and the suffix-er (which means“ ... person ”,“ ... thing ”,“ ... thing ”), etc.

  Also, different notations can be included as derivatives. This includes English spelling vs. American spelling (eg color and colour; organize and organize), capital letters vs small letters (Babbitt metal and babbitt metal, social security number and Social Security number), There are things such as “all initial capital letters”, “all initial capital letters in lower case” vs. “a part of initial words in capital letters”. By considering in this way, the present embodiment can be applied without limiting the types of utilization / derivation to be targeted.

(Third embodiment)
Next, third and fourth embodiments of the present invention will be described. The configuration of the machine translation apparatus according to the third and fourth embodiments is the same as that of the first embodiment shown in FIG. 1 except for one point. This difference is that, as shown in FIG. 11, there is a bilingual list creation availability control unit (as part of the “editing unit + translation list creation availability control unit” 13A) that controls whether or not a bilingual list should be created. . FIG. 11 is a block diagram showing a configuration of a machine translation apparatus according to another embodiment of the present invention. In the third and fourth embodiments, translation is performed in the same manner as in the first embodiment, and the “translation” described in the first embodiment is applied to the character string that the parallel translation list creation availability control unit 13A sets “permitted”. Create a "list".

  There are roughly two possibilities for this realization. One is to control whether or not the bilingual list is created based on the result of the determination performed by the machine translation apparatus, and the other is to perform this control by an instruction from the outside. Hereinafter, the former will be described as a third embodiment, and the latter will be described as a fourth embodiment.

  The third embodiment uses the second language utilization change dictionary and the analysis grammar dictionary to determine how the first language headwords correspond to the translations of the first language, and It is characterized by analyzing the type and determining whether or not to create a “translation list” based on this.

  The operation of the machine translation apparatus according to this embodiment will be described with reference to FIGS. 12 is a flowchart showing an operation flow of the machine translation apparatus shown in FIG. 11, FIG. 13 is a flowchart showing a detailed flow of step 300 shown in FIG. 12, and FIG. 14 is a step shown in FIG. 3 is a flowchart showing a detailed flow of 350. In FIG. 12, the steps already described in FIG. 2 are denoted by the same reference numerals, and the description of those steps is omitted because it is duplicated.

  Although the method of operation differs depending on the language pair, here, an example will be described in which one word (original word) and its compound word appear in English-Japanese translation of a noun (phrase). As a specific example, focusing on the single word “cylinder”, the headword “cylinder rod” and the translated word “cylinder rod”, the headword “cylinder” and the translated word “cylinder”, the headword “brass cylinder assembly” And the translation "brass cylinder assembly" pair, the entry "graduated cylinder" and the translation "cylindrical cylindrical liquid volume measuring instrument" pair, the entry "gas cylinder" and the translation "bomb" are the same document Consider the case of appearing inside.

  In the analysis in step 300, first, a plausible katakana transliteration is obtained from the English spelling of the original word in a state where the translation of the English phrase in “” is obtained from the word / idiom dictionary 15c (step 301). In addition, since the collation technique of an original language and a katakana character string is already established, it will be used. In this example, “cylinder” to “cylinder” or “cylinder”, “rod” to “rod”, “brass” to “brass”, “assembly” to “assembly” or “assembly”, “graduated” to “grass” “Gas” is derived from “duated” and “gas”, respectively.

  Next, the translated word and the character string after the katakana replacement are collated, and if they match (Y in step 302), it is determined that the transliteration. In this example, a pair of “cylinder rod” and “cylinder rod” corresponds to this, and as will be described later, the constituent elements have a one-to-one correspondence such as “cylinder” and “cylinder”, and “rod” and “rod”. ing. On the other hand, if no match is found (N in step 302), first, using the utilization change dictionary 15a and the analysis grammar dictionary 15b in which the grammar for analyzing the first language is stored, It is determined whether or not can be divided (step 304).

  When the division is completed, the word / idiom dictionary 15c is searched for each word and a possible translation is checked (step 306). Next, the translated words of the word before division are analyzed using a target language (Japanese) utilization change dictionary (not shown) and an analysis grammar dictionary (not shown) to extract content words such as nouns and verbs (step 307). ). A comparison pair of these content words and the previously extracted possible translation words is created, and the degree of coincidence is observed (step 308).

  If they match without excess or deficiency, it is determined that the translation is a one-to-one correspondence (step 309). In the above, “brass”, “cylinder”, “assembly” and “brass cylinder assembly” and “brass cylinder assembly” pairs that fall into this category fall into this category. If it partially matches, it is determined as a partially corresponding translation (step 310), and the matching location is stored and saved. Here is the translation of “graduated cylinder”. When “graduated cylinder” is disassembled, the translation of “graduated” coincides with “graduated” because there is a translation of “graduated, graded, progressive” and cylinder is “cylinder”. This is because there is a portion that does not match, such as “liquid volume measuring device”.

  If they do not match at all, it is determined as a non-corresponding translated word (step 311), and the fact that the matching degree is 0 is stored and stored. This is typically the case where the number of morphemes (words) in the original word exceeds the number of morphemes (words) in the translated word, such as a “gas cylinder” and a “bomb” that cannot be further divided.

  If the division performed in step 304 is not successful, the type of translated word cannot be determined (step 305). In this case, as described later, the “translation list” is displayed.

  When the type classification of the translation word is completed as described above, the result of the type classification is given to the translation word (step 312). Note that the above type classification process (step 300) can be omitted if the word / idiom dictionary 15c has already been given the type of translation (for example, classification such as transliteration, explanatory note, and one-to-one correspondence). .

  In the state where the above processing is performed, when “end” is input in step 24, the translation apparatus asks the user whether or not it is necessary to create a “translation list” (the next automatically without asking The process proceeds to step 350 for creating a “translation list”.

  In step 350, as shown in FIG. 14, first, it is determined whether the translation includes a transliteration, a one-to-one correspondence translation, and a partial correspondence translation (step 351). If these types of translations exist, the processes of “divide compound words into words” (step 352) and “take out words consisting of one word” (step 353) are performed. These processes are the same as steps 30 and 31 in FIG.

  Next, each word of the compound word or a word consisting of one word is compared, and an at least partially matching word is extracted as a “word of interest” (step 355). In the specific example, “cylinder” is extracted. As a result, “cylinder” is obtained internally from “cylinder rod”, “cylinder” from “cylinder”, and “cylinder” from “brass cylinder assembly” and “graduated cylinder”. That is, it can be seen that there are three types of translations of “cylinder”, “cylinder”, and “cylinder” for the same “cylinder”. If there is only one translation (Y in Step 355), the “translation list” is not created in this embodiment (Step 357), and the user is notified that there is no target word.

  On the other hand, as described above, when there are a plurality of translated words (N in Step 355), or indeterminate words or the like (N in Step 351 to Y in Step 358), in this embodiment, the “translation list” Is made. If there are no undecidable words and non-corresponding translated words in step 358 (N in step 358), the “translation list” is not created (step 360).

  A “translation list” created as a specific example is as shown in FIG. FIG. 15 is a view showing a display example of a parallel translation list obtained by the machine translation apparatus shown in FIG. In this display example, as shown in the figure, it is possible to devise such that the correspondence can be extracted, for example, underlined or easy to see. In addition, if there is a compound word that includes “cylinder”, the translation equivalent of “cylinder” exists in that translation, and if it is the same as “cylinder”, which is the translation when it is alone, the “translation list” is displayed. You may make it remove from the object of. In addition, even when an undecidable word exists (Y in Step 358), in this embodiment, it is a target for creating a “translation list” (Step 359).

  The user can easily make necessary corrections to the translated text while viewing the “translation list” as shown in FIG. For example, if “cylinder” alone and “brass cylinder assembly” appear in or near the same sentence, this “cylinder” is very likely to refer to the same thing, so the former translation is not “cylinder” You can see that it is easier to understand if you use the same translation, such as changing to "Cylinder".

  To see how “cylinder” is translated in the same document for “brass cylinder assembly” as in this specific example, what is the “brass cylinder assembly” listed in the dictionary heading? It helps users think about new and different translations. Furthermore, the presence of different notations such as “cylinder” and “cylinder” can be easily detected, and it would be desirable to unify these unless there is a specific reason. In Japanese, in addition to the long clef symbols, there are also midpoints and variations in katakana notation. Similarly, in English, for compound words, whether to use one word (eg “broadband”), two words (connect with a hyphen (eg broad-band), or leave a space (eg: The current situation is that notation is not necessarily unified in the same dictionary in both the commercial dictionary and the machine translation dictionary, so adjustment of the notation in the document after applying the dictionary is inevitable. Yes.

  Furthermore, the translation of one-word cylinder headwords in the document of this specific example may include the ones that should be added to the dictionary as headings of compound words by adding the preceding and following elements. There is sex. In order to facilitate the dictionary registration work for this purpose, it is also possible to provide the following functions. For this purpose, first, the “used sentence” item is displayed as KWIC (keyword in context) as shown in FIG. Further, from the phrase structure obtained in the translation processing process, the range of phrases (here, noun phrases) is clearly shown as shown by the underline in FIG. The translation of this part of the noun phrase is a default translation, which is not always desirable, but the current translation is displayed in the vicinity for reference. While referring to this result, the user can smoothly perform dictionary registration as necessary.

(Fourth embodiment)
Next, a fourth embodiment of the present invention will be described with reference to FIG. FIG. 16 is a block diagram showing a configuration of a machine translation apparatus according to still another embodiment of the present invention. As shown in the figure, this machine translation apparatus has a pre-editing unit (as part of “editing unit + pre-editing unit” 13B) in addition to the components of the machine translation device shown in FIG. Yes. This pre-editing unit is for an editing process that the user performs on the document before translation in order to bring the translation result closer to a desired shape. For example, one of them is rewriting such as dividing a long sentence or making a passive sentence into an active sentence, or specifying a phrase to be displayed as it is in a translated sentence. Furthermore, in this embodiment, this pre-editing unit is also used for designating a phrase for which a “translation list” is to be created. This aspect will be described below.

  First, as a pre-translation preparation, the user designates a word / phrase to be given a “translation list” using the input unit 11 through the pre-editing unit. Here, as a specific example, it is assumed that such a designation is made for the word “cylinder” used in the description of the third embodiment. For the designation, for example, a bilingual list creation target designation screen as shown in FIG. 17 may be used to instruct the bilingual list target and format. FIG. 17 is a diagram illustrating an example of a screen for prompting the user to specify an arbitrary phrase that is displayed on the display unit by the function of “editing unit + pre-editing unit” illustrated in FIG. 16. When instructed in this way, the machine translation device executes a process of creating a “translation list” based on the instruction.

  The options shown in FIG. Can be specified from among three types: (a) -1), only a single word ((a) -2), or whether all are included in the target ((a) -3) . In a specific example, if ((a) -1) is selected, “cylinder” alone is excluded, and ((a) -2) is conversely only “cylinder” alone, ((a) -3 ) Is for both.

  In addition, 2. As a result, in addition to “cylinder”, “cylindrical” (adjective), “cylindrically” (adverb), “cylindricality” (noun), “cylinder-like” ( Adjectives) are also searched, and in addition, because “cylinder” is a verb derived from a noun, “cylindering” and “cylindered” can also be searched. In addition, when only a search for a part of the derivative words and the use words is desired, a framework for limiting the search target by designating the part of speech or designating the word itself is provided. This is useful when dealing with all derivatives and the number of output is very large.

  The above is the limitation on the original language side. In this embodiment, control on the translated word side is also possible. If a translated word is designated here, it is possible to remove from the display object the corresponding translated word. For example, when “equipment” is a search word, it is possible to specify that the translated word is, for example, “device”. Then, when “semiconductor processing equipment”, “semiconductor processing equipment”, “electronic equipment”, “electronic equipment”, “airborne equipment”, “aircraft equipment” are included in the document, the former two are excluded. This is effective when applied to a document in which it is known in advance that a specific word repeatedly appears, such as “apparatus” in a patent document.

  4. The related information requested by the user can be selected. For example, items such as the original language, the part of speech of the translation, the type of translation (transliteration, one-to-one correspondence translation, partial correspondence translation, explanatory tone, etc.), and the sentence that occurred are included.

  The above description refers to the original text and specifies whether or not the user wishes to create a “translation list”, but allows the user to input whether or not he wishes to create a “translation list” while referring to the translation. May be. In other words, the above example is a case where the translator determines the necessity of the “translation list”, but this example can also be regarded as a case where the reader of the translation document determines the necessity of the “translation list”. The operation in such a case will be described next.

  In such a case, the “pre-editing unit” in the embodiment described above is functionally a “post-editing unit”. First, after the translation is completed, the user uses the input unit 11 through the post-editing unit to specify a word / phrase to which the “translation list” is to be assigned. For example, it is assumed here that “heat radiation” corresponding to the original word “heat” is designated. For such designation, a screen for designating a translation list creation format as shown in FIG. 17 can be used. However, “single, compound word” and “derivative word, utilization word” in FIG. 17 indicate those on the translation side, and “translation word to be removed from display object” is “original word to be removed from display object”.

  In this example, for example, in the case where “single and compound words”, “derivative words, and usage words” are not limited and there is no restriction on the original language, the compound word “heat dissipation circuit” other than “heat dissipation” is used. In addition to “Heat dissipation source” and “Heat dissipation characteristics”, you can also search for usage words and related terms “Heat dissipation” and “Heat dissipation possible” (provided that these are used in the document) To do.) After the processing is completed, for example, a “translation list” as shown in FIG. 18 is created and displayed. By doing so, the dictionary registration work in the reverse direction (in this case, Japanese-English) can be reduced by sorting the translated words as keys.

  Note that the embodiment described above can be implemented with various modifications as follows. For example, in the above embodiment, a part of the processing necessary for creating the “translation list” is performed in parallel with the translation of the text, but at that stage, the processing necessary for creating the “translation list” is not performed at all. It is also possible to go through the process of creating a “translation list” after performing machine translation.

  In addition, the designation of creating the “translation list” is collectively performed before or after the translation, but it may be instructed that the translation list is desired for each word as interactive translation. Moreover, although the said embodiment uses the output of machine translation as a translation, you may use what digitized what was translated manually. However, there are significant differences in this case. That is, in the case of machine translation, there is a distinction between a single word and a compound word, but not in the latter. Not only terms in the same sentence but also sentence units do not always correspond one-to-one.

  Therefore, in this case, the processing as in the above-described embodiment is performed for expressions that have a corresponding relationship in the machine processing. The method for automatically extracting the correspondence relationship of bilingual documents is described in the literature: “Yamato Takahashi, Satoshi Shirai, Yoshifumi Oyama“ Automatic Generation of Corresponding Corpus of Japanese-English Newspapers ”The 3rd Annual Conference of the Language Processing Society, pp. 127-130, 1997 ”, literature:“ Kenji Matsumoto, Hideki Sasaoka, Hideki Tanaka “Construction of Japanese-English Bilingual Corpus Using Field-Specific Information”, Information Processing Society of Japan 63rd National Convention, vol. 2, pp. Since various devices such as “251-252, 2001” or Japanese Patent Application Laid-Open No. 2004-102481 have been devised, these techniques can be used.

  Even if the correspondence is obtained only partially, the application of the present embodiment to human translation is useful for detecting a phenomenon that is unlikely to occur in machine translation but is likely to occur in human translation. In other words, in the case of manual translation, there are cases where different translations are applied to different places depending on places due to carelessness and memory differences with respect to the original language expressions that should be translated in the same way. According to this embodiment, it is possible to comprehensively detect such a possibility. Although the above description has been described by taking English-Japanese translation as an example, it can be applied to not only English-Japanese but also Japanese-English.

  As described above, according to each embodiment, a different combination of a related word and its translation in the first language is extracted and displayed, and the translation is corrected while referring to the display result. Then, it becomes easy to correct or add to the translation dictionary (parallel translation dictionary) based on the extraction result.

  More specifically, four main effects can be expected. First, it is possible to find out the translated expression in which the object to be pointed is unclear and to make an efficient correction. That is, it is a problem that different translations are given to indicate the same thing. This is a particularly serious problem if the translations are not synonyms in the second language. Second, the quality of the dictionary can be improved by unifying expressions in the current machine dictionary and adding translation candidates.

  Third, it is possible to mitigate the negative effects of a user dictionary created by a user who is not a dictionary editor. This problem is that only a limited sentence is seen, words that should not be used as headwords are used as dictionary headlines, and their character strings and translations are prioritized more than necessary. Fourth, because the search target includes derivative terms, expressions that are close in meaning, such as “radiation heat, radiative heat, radiant heat” but have been apt to be treated separately in the past, are related. become.

1 is a block diagram showing a configuration of a machine translation apparatus according to an embodiment of the present invention. The flowchart which shows the operation | movement flow of the machine translation apparatus shown in FIG. The figure which shows the example of a display of the original sentence and translation which were prepared on the machine translation apparatus shown in FIG. The figure which shows a response | compatibility with the phrase of an original sentence, and the phrase of a translation sentence in the original sentence and translation which are shown in FIG. The figure which shows the example of a display of the bilingual list obtained from the matching shown in FIG. The figure which shows another example of the display of the original sentence and translation which were prepared on the machine translation apparatus shown in FIG. The figure which shows a response | compatibility with the phrase of an original sentence, and the phrase of a translated sentence in the original sentence and translation which are shown in FIG. The figure which shows the example of a display of the bilingual list (by comparison of one word) obtained from the correspondence shown in FIG. The figure which shows the example of a display of the bilingual list (by the comparison of compound words) obtained from the correspondence shown in FIG. The figure which shows the example of a display of the bilingual list obtained by the comparison of the words before utilization. The block diagram which shows the structure of the machine translation apparatus which concerns on another embodiment of this invention. 12 is a flowchart showing an operation flow of the machine translation apparatus shown in FIG. 11. The flowchart which shows the detailed flow of step 300 shown in FIG. The flowchart which shows the detailed flow of step 350 shown in FIG. The figure which shows the example of a display of the bilingual list obtained with the machine translation apparatus shown in FIG. The block diagram which shows the structure of the machine translation apparatus which concerns on another embodiment of this invention. FIG. 17 is a diagram showing an example of a screen for prompting the user to specify an arbitrary phrase that is displayed on the display unit by the function of “editing unit + pre-editing unit” shown in FIG. 16. The figure which shows the example of a display of the bilingual list obtained with the machine translation apparatus shown in FIG.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 11 ... Input part, 12 ... System control part, 13 ... Editing part, 13A ... Editing part + Parallel list creation availability control part, 13B ... Editing part + Pre-editing part, 14 ... Translation part, 15 ... Translation dictionary part, 15a ... Utilization change dictionary (first language), 15b ... analysis grammar dictionary (first language), 15c ... word / idiom dictionary, 15d ... conversion grammar dictionary, 15e ... generated grammar dictionary (second language), 15f ... morpheme generation grammar dictionary (Second language), 16... Display unit, 17... Document storage unit.

Claims (22)

Preparing a first language sentence to be translated and a translated second language sentence;
A second step of associating each translated word in the translated second language sentence with each phrase corresponding to the translated word in the first language sentence to be translated;
A third step of determining whether or not the words in the first language associated with each other are compound words;
Phrases determined not to be compound words by the determination, phrases determined to be compound words by the determination, or phrases determined not to be compound words by the determination and phrases determined to be compound words by the determination A fourth step of comparing
A fifth step of extracting at least partially matching phrases in the phrase comparison pair;
A machine translation program for causing a computer to execute a sixth step of displaying a phrase in the first language by the association including the extracted phrase together with the translated word corresponding to the phrase. Dividing the compound word with respect to the phrase determined to be a compound word by the determination;
Further converting each of the divided words and the words determined to be non-compound words by the determination into original words to obtain an original word string;
The fourth step uses the obtained original word string in place of a word that is determined to be a compound word by the determination or a word that is determined not to be a compound word by the determination;
The sixth step is to display a phrase in the first language by the association including a phrase in the utilization form of the extracted phrase together with the translated word corresponding to the phrase. The machine translation program according to claim 1. Dividing the compound word with respect to the phrase determined to be a compound word by the determination;
Further converting each of the divided words and the phrase determined not to be a compound word by the determination to a word before derivation to obtain a word string before derivation,
The fourth step uses the obtained pre-derivation word string in place of the phrase determined to be a compound word by the determination or the phrase determined not to be a compound word by the determination,
The sixth step is to display a phrase in the first language by the association including a phrase derived from the extracted phrase together with the associated translation for the phrase. The machine translation program according to claim 1. For each translation in the translated second language sentence, further causing the computer to execute a step of obtaining a translation type,
The third step is a compound word only when the second language translation corresponding to the associated first language phrase is a specific translation type in the acquired translation type. The machine translation program according to claim 1, wherein the machine translation program determines whether or not.   5. The machine translation program according to claim 4, wherein the type of the specific translation is a transliteration, a one-to-one correspondence translation, a partial correspondence translation, or an explanatory note.   The sixth step is to display a sentence in which the phrase of the first language is used together with the phrase of the first language and the translation corresponding to the phrase of the first language. The machine translation program according to claim 1.   The machine translation program according to claim 6, wherein the display of the sentence in which the phrase in the first language is used includes an indication that the phrase in the first language is a keyword. Preparing a first language sentence to be translated and a translated second language sentence;
A second step of associating each translated word in the translated second language sentence with each phrase corresponding to the translated word in the first language sentence to be translated;
A third step for prompting the user to specify an arbitrary phrase for the sentence in the first language to be translated;
Displaying a word in the first language by the correspondence including the designated word or phrase, or a phrase in a utilization form or a derivative of the word, together with the translated word associated with the word in the first language A machine translation program for causing a computer to execute steps and.   9. The machine translation program according to claim 8, wherein the fourth step does not function when the designated phrase is a specific translation in the translated word. Preparing a first language sentence to be translated and a translated second language sentence;
A second step of associating each translated word in the translated second language sentence with each phrase corresponding to the translated word in the first language sentence to be translated;
A third step for prompting the user to specify an arbitrary phrase for the translated second language sentence;
Displaying the second language phrase by the correspondence including the specified phrase or the utilization form or derivative form of the phrase together with the first language phrase associated with the second language phrase A machine translation program for causing a computer to execute the fourth step.   The machine translation program according to claim 10, wherein the fourth step does not function when the designated phrase is a specific phrase in association with the first language. First means for preparing a first language sentence to be translated and a translated second language sentence;
A second means for associating each translated word in the translated second language sentence with each phrase corresponding to the translated word in the first language sentence to be translated;
A third means for determining whether or not the phrase of the first language associated with each other is a compound word;
Phrases determined not to be compound words by the determination, phrases determined to be compound words by the determination, or phrases determined not to be compound words by the determination and phrases determined to be compound words by the determination A fourth means for comparing
Fifth means for extracting at least partially matching phrases in the phrase comparison pair;
A machine translation apparatus comprising: a sixth means for displaying a phrase in the first language by the association including the extracted phrase together with the translated word corresponding to the phrase. Means for dividing the compound word with respect to the phrase determined to be a compound word by the determination;
Means for converting each of the divided phrases and the phrase determined not to be a compound word by the determination into an original word phrase to obtain an original word string;
The fourth means uses the obtained original word string in place of the phrase determined to be a compound word by the determination or the phrase determined not to be a compound word by the determination,
The sixth means is characterized in that the phrase in the first language by the correspondence including the phrase of the utilization type of the extracted phrase is displayed together with the corresponding translated word for the phrase. The machine translation apparatus according to claim 12. Means for dividing a compound word with respect to a phrase determined to be a compound word by the determination;
Means for converting each of the divided words and a word that is determined not to be a compound word by the determination into a word before derivation to obtain a word string before derivation;
The fourth means uses the obtained pre-derivation word string in place of the phrase determined to be a compound word by the determination or the phrase determined not to be a compound word by the determination,
The sixth means is characterized by displaying a phrase in the first language by the association including a phrase derived from the extracted phrase together with the associated translation for the phrase. The machine translation apparatus according to claim 12. For each translation in the translated second language sentence, further comprising means for obtaining a translation type,
Each of the third means is a compound word only when the translation of the second language corresponding to the word / phrase of the associated first language is a specific translation type in the acquired translation type. The machine translation device according to claim 12, wherein each of the machine translation devices determines whether or not.   16. The machine translation apparatus according to claim 15, wherein the type of the specific translation word is a transliteration, a one-to-one correspondence translation, a partial correspondence translation, or an explanatory tone.   The sixth means displays a sentence in which the phrase of the first language is used together with the phrase of the first language and the associated translation for the phrase of the first language. The machine translation apparatus according to claim 12.   18. The machine translation device according to claim 17, wherein the display of the sentence using the phrase in the first language includes an indication that the phrase in the first language is a keyword. First means for preparing a first language sentence to be translated and a translated second language sentence;
A second means for associating each translated word in the translated second language sentence with each phrase corresponding to the translated word in the first language sentence to be translated;
Third means for prompting the user to specify an arbitrary phrase for the sentence in the first language to be translated;
Displaying a word in the first language by the correspondence including the designated word or phrase, or a phrase in a utilization form or a derivative of the word, together with the translated word associated with the word in the first language And a machine translation device.   20. The machine translation apparatus according to claim 19, wherein the fourth means does not function when the designated phrase is a specific translation in the translated word. First means for preparing a first language sentence to be translated and a translated second language sentence;
A second means for associating each translated word in the translated second language sentence with each phrase corresponding to the translated word in the first language sentence to be translated;
A third means for prompting the user to specify an arbitrary phrase for the translated second language sentence;
Displaying the second language phrase by the correspondence including the specified word or phrase utilizing or deriving from the phrase together with the first language phrase associated with the second language phrase A machine translation apparatus comprising: a fourth means configured to:   The machine translation apparatus according to claim 21, wherein the fourth means does not function when the designated word / phrase is a specific word / phrase associated with the first language.

Images