JPH10105555A - Translation-with-original example sentence retrieving device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To more preferentially perform the retrieval and display of an example sentence required for a user by weighting keywords while storing the important degrees of keywords for each part of speech, and performing retrieval while considering the important degrees of keywords. SOLUTION: A morpheme analysis part 2 analyzes a retrieval request sentence into morphemes while using a morpheme analytic dictionary 6, and a keyword extracting part 3 extracts the keyword from the words of the morpheme analyzed results while using a keyword dictionary 7. In this case, the part of speech is provided from the keyword dictionary 7 for keyword weighting as well. At a keyword weighting part 11, the keyword is weighted corresponding to the part of speech. A retrieval part 4 finds the example sentence including all the provided keywords or one part of them out of a translation-with-original example sentence data base 8. A keyword weighting processing part 12 determines the order of outputting while considering the weight provided by the keyword weighting part 11. Thus, when preparing the sentence in the other language, the more effective example sentence is more preferentially retrieved and displayed and sentence preparation work is facilitated.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a bilingual translation for displaying a sentence close to a desired sentence as an example sentence when creating a sentence in a language other than the native language with reference to a bilingual example sentence in a native language and another language. The present invention relates to a bilingual example sentence search device that searches for example sentences.

[0002]

2. Description of the Related Art When a document in another language is created, usually, a translated word is looked up in a dictionary, and an example sentence described therein is referred to. Conventionally, there is a method of easily searching for such example sentences by accumulating the example sentences in a memory so as to be searchable.

For example, if you want to create an English sentence corresponding to "I like English."
Several example sentences including the translation corresponding to "like" are displayed, and the English sentence can be created by referring to the example sentences. In some cases, a plurality of words are input and an example sentence including a translation corresponding to the plurality of words is displayed.

[0004] Further, instead of inputting a word as an input, there is a type in which a sentence is input, a keyword is automatically extracted from the sentence, and a search is performed from the extracted keyword.

In Japanese Patent Laid-Open Publication No. 57-27367, a keyword is input and an example containing the keyword is searched. Japanese Patent Laid-Open Publication No. Hei 3-276369 proposes an example-driven machine translation system that searches for similar bilingual translation examples, performs some processing on them, and translates them.

[0006]

However, a machine translation system that is generally commercialized or a machine translation system of the above-mentioned example-driven type requires a large amount of processing and currently cannot expect much translation accuracy. Therefore, consider a parallel translation example sentence search apparatus that merely presents a similar translation example and allows the user to independently compose in another language with reference to the similar translation example.

Conventionally, when writing a sentence in another language, an outline is often written in advance in its own language. Then, for some of the words in the native language sentence, a dictionary is searched for translations and examples, and a foreign language sentence is created with reference to the translated words and examples. A system that supports this would be a system that automatically morphologically analyzes the sentence in its own language, searches for the necessary keywords from the extracted words, and searches for examples using the keywords or different notations or synonyms of the keywords. Can be It is difficult for ordinary people to use a keyword or a keyword and its logical expression as in a so-called search system.

[0008] Furthermore, when creating a foreign language, it is necessary to give the highest priority to a bilingual example sentence of a word such as a verb or an auxiliary verb that determines the style of a foreign language sentence. .

For example, consider the creation of English native to Japanese. "I will send you a letter of appreciation." Examples of "send", such as whether "send" can take two objects. Also, I want to rely on example sentences for what "thank you" is. However, the example sentences of "I" and "you" are unnecessary first.

Therefore, in the invention according to the first aspect, by changing the weight of the search keyword depending on the part of speech, it is possible to search and display the example sentence required by the user with higher priority.

In the invention according to claim 2, even if the same part of speech is used,
To deal with the case where you want to change the weight for each word,
It is possible to set words that you do not want to be prioritized.

[0012] For example, even in the case of a noun,
eciation) wants an example sentence, but a "desk" is unnecessary first. By setting these in advance, the example sentence desired by the user can be searched and displayed with higher priority.

In addition, it is possible to individually set these words that are not required much. Since the settings can be freely changed between a user who has language skills and a user who does not have language skills, a search suitable for the user level can be performed.

In the invention according to claim 3, it is possible to consider the distance between keywords and the order of appearance. For example,
Enter "He raised his hand."
I washed my hands. "A raised his hand."
It is desirable to give a higher priority to two keywords having the same order of “hand” and “raise”.

If the user inputs "turn on the power.", "Turn on" and "turn on" appear at a position closer to the remote location than "the power supply unit has been purchased and turned on the machine.""Please turn on the power immediately." In the invention according to claim 3, these can be considered.

Next, consider a case where a search request sentence having a plurality of parts to be assisted, that is, a request sentence including a keyword for which a plurality of parallel translation example sentences are desired to be viewed is input.

For example, "Study science, society and arithmetic"
Is input, and if there are 30 or more example sentences containing "Science" and "Society" in the system that displays 30 bilingual example sentences, even if there is one example sentence containing "Arithmetic" in the database, Do not show. Therefore, in the invention according to claim 4,
With these considerations in mind, example sentences for all keywords can be displayed within the output range as long as they are in the database.

In the invention according to claim 5, when the searched example sentence is displayed in a different color for each keyword so that it can be seen at a glance, or when the same sentence is matched by a synonym. Can be selected so as to be distinguished at a glance from the case where they completely match.

[0019]

According to the first aspect of the present invention, a bilingual example sentence database having a bilingual example sentence in a first language and a second language, a keyword translation dictionary, and a search in the first language Search request sentence input means for inputting a request sentence,
Means for morphologically analyzing the search request sentence using a morphological dictionary, keyword extracting means for extracting a keyword from a morphological analysis result of the search request sentence using the keyword dictionary, and a search request sentence from the bilingual example sentence database. In a bilingual example sentence search device having a search means for extracting an example sentence containing a keyword or a synonym thereof and a display means for displaying the extracted parallel sentence, the importance of the keyword is stored for each part of speech, The above object is achieved by providing a keyword weighting means for performing weighting, and a weight processing means for performing a search in the search means in consideration of the importance of a keyword by the keyword weighting means.

According to the second aspect of the present invention, the plain word storage means for storing plain words, the plain word editing means for modifying the plain word storage means, and the plain word stored in the plain word storage means are provided. The above object is achieved by providing a keyword weighting means for lowering the weight of a word and a weight processing means for performing a search in the search means in consideration of the importance of a keyword by the keyword weighting means.

According to claim 3 of the present invention, keyword position detecting means for recognizing the distance between keywords and the order of appearance, keyword weighting means for performing search weighting based on the result of the keyword position detecting means, The above object is achieved by providing a weighting means for performing a search in consideration of the importance of a keyword by the keyword weighting means.

According to the fourth aspect of the present invention, the above problem is solved by providing an output adjusting means for outputting, when displaying a search result, a bilingual example sentence for all keywords in an output display area. I do.

According to the fifth aspect of the present invention, the keyword in the search request sentence and the corresponding keyword in the bilingual example sentence are displayed correspondingly, or the keyword in the bilingual example sentence is completely matched or distinguished according to the synonymous match. The above object is attained by providing a feature display selecting means for selecting whether or not to do so, and a flag for storing the selected mode, and displaying the feature of the keyword based on the flag.

[0024]

DESCRIPTION OF THE PREFERRED EMBODIMENTS A document retrieval apparatus according to the present invention will be described in detail with reference to the drawings. FIG. 1 shows an example of the configuration of an example sentence search device according to the present invention. 1 is a search request input unit, 2 is a morphological analysis unit, 3 is a search request keyword extraction unit, 4 is a search unit, 5 is a display unit, 6 is a morphological dictionary, 7 is a keyword dictionary, 8 is a bilingual example sentence database, and 9 is a keyword. Position detection unit, 10 is an output adjustment unit, 11 is a keyword weighting unit, 12 is a keyword weight processing unit, 13 is a plain word storage unit, 14 is a plain word editing unit, 15 is a feature display selection unit, and 16 is a feature display mode flag. It is. These components are basically configured by a computer or the like, and a CPU (control unit) (not shown) controls each component. Each configuration will be described later.

The operation of the apparatus according to the present invention will be specifically described with reference to the flowchart of FIG. In this example, the case where the native language is Japanese and the other language is English will be described.

In step S1 (hereinafter simply referred to as S1), a selection is made as to whether to search for a bilingual example sentence, to edit plain word registration or deletion, or to end the processing.
If a bilingual example sentence search is to be performed, the process proceeds to S2. If a plain word is edited, the process proceeds to S14. If so, shut down the system.

In step S2, when the search result is displayed, the color is changed for each keyword, the type of underline is changed, and the search result is displayed (FIG. 6). The feature display selection unit 15 selects whether the notation is matched or not (FIG. 7), or whether such display is not performed at all, and stores it in the feature display mode flag 16.

In S3, a search request is input in the search request input unit 1. As the input means, any means such as kana-kanji conversion and character recognition may be used.

In S4, the morphological analysis unit 2 recognizes a word from the words registered in the morphological analysis dictionary 6. In particular, no algorithm is specified. A general morphological analysis unit used conventionally is used.

In S5, the keyword extraction unit 3 extracts a keyword from the words of the morphological analysis result in S4 of the search request sentence, using the keyword dictionary 7. Since the keywords of the example sentences registered in the bilingual example sentence database 8 also consist of one or more words registered in the keyword dictionary 7, the extracted words are keywords. S8
Is also obtained from the keyword dictionary 7 for the purpose of weighting keywords.

In step S6, the keyword position detection unit 9 detects the appearance position of the keyword. This is explained in the example sentence above.

When "He raised his hand" was input,
In the morphological analyzer, it is analyzed as "he", "ga", "hand", "wo", "raised", pronoun, particle, noun, particle, verb. Of these, "he", "hand" and "raised" pronoun noun verb is registered in the keyword dictionary 7 and is extracted as a keyword. It means that the keyword appeared in this order, that there was one word "ga" between "he" and "hand", and that one word "wa" between "hand" and "raised" Is sent to the search unit 4.

Similarly, "turn on the power."
“Power” and “turn on” are keywords, and the fact that they appear in this order and the fact that there is one word “を” between “power” and “turn on” are sent to the search unit 4.

At S7, the keyword weighting unit 11 checks the plain word storage unit 13 and uses it for weighting at S8. Only when the English word (shown in FIG. 3) of the plain word storage unit 13 appears on the English side of the bilingual example sentence, the word is processed as a plain word.

For example, in the case of "dog" shown in FIG. 3, since the English part is "dog", the Japanese keyword is "dog".
Even if it is used in the meaning of "spy" and "spy" appears anywhere in the English sentence, it is not included. That is, they are not treated as plain words. This does not correspond to Japanese words and English words ("dog" and "spy" in the example), but this is sufficient. However,
A response may be taken in advance.

At S8, the keyword weighting unit 11
Weights keywords by part of speech. For example, a verb, an adjective, or the like, a sentence-end modal expression such as an auxiliary verb has a priority "A rank", a noun or the like has "B rank", and a pronoun or the like has a "C rank". Those in the plain word storage unit 13 confirmed in S7 are unconditionally set to “C rank”. FIG. 9 shows an example of storing in correspondence with FIG.

In S9, the search unit 4 searches the bilingual example sentence database 8 for an example sentence including all or a part of the keyword of the search request obtained in S5. An example sentence in which all keywords match, and thereafter, an example sentence in which more keywords match are obtained.

Further, a synonym table as shown in FIG. 11 is provided in the keyword dictionary 7, and words that are not completely the same but are considered to be almost synonymous to be extracted as a keyword are stored. Search by. Here, a flag indicating that the word is searched for a synonym is added.

At S10, the output order is determined by the keyword weighting unit 12 in consideration of the weight obtained at S8 with respect to the result at S9. For example, the total score is calculated by assuming that the word of the rank A is 1, the score is 10 when the word is included, the score is 5 for the rank B, and the score is 1 for the rank C.

Calculating in the above example sentence, if the request sentence is "I will send a letter of thanks to you", "I bought a letter to you." Since "you" is a pronoun, "letter" has 5 nouns, one point at a time, and a total of 7 points. In "Send a postcard to her", "Send" is a verb with 10 points, and priority is given to a small number of matching keywords. That is, "She sent a postcard to her" is output first.

Further, the keyword appearance position obtained in S6 is also taken into account. When he inputs "He raised his hand.", "He lifted his hand and washed his hand." And "A raised his hand." The latter, which has the same score for the two keywords but the same order of appearance of the words, is prioritized (FIG. 10A).

When "Power on." Is input, "The power supply unit has been purchased and the power supply unit has been turned on."
And "Turn on the power immediately", the score is the same for the two keywords "power" and "turn on", but the request sentence is one word between "" and " I bought it and put it in the machine. "
These are seven words, "", "purchase", "ta", "so", "machine", and "ni". On the other hand, "Please turn on the power immediately."
Since there are two words "" and "immediately", "Please turn on the power immediately" is determined to be closer to the request sentence and has priority (FIG. 10B).

In S11, the output adjusting unit 10 sets S1
Before the result of 0 is displayed on the display unit 5, if the translation example sentence corresponding to all the keywords of the search request sentence cannot be displayed in the output display area, the example sentence with the same keyword that is frequently output is deleted. Either a representative of the example sentence of the keyword not appearing is displayed (FIG. 4), or the representative is displayed so that the entire sentence can be viewed as needed (FIG. 5).

In FIGS. 4 and 5, for example, if the number of outputs in the output display area is 30 bilingual, and if there are 30 or more example sentences that match two keywords of “book” and “read”, the output display area is changed. In this example, 28 examples of "book" and "read" are used, and example sentences of "teacher" and "school" are displayed. In FIG. 5, when "+" is displayed, all sentences of "teacher" and "school" are displayed in a separate window.

In step S12, the feature display mode flag 16 is checked, and the keywords of the search request sentence are distinguished and displayed for each keyword (using color and underline). A feature corresponding to the keyword is displayed (FIG. 6).

Alternatively, a feature is displayed (using a color or an underline) in a portion of the search result where the keyword of the search request sentence is included, and a portion where a synonym or a different expression of the keyword of the search request sentence is included Display a different feature (using a different color or underline).

In FIG. 7, “book” is a keyword of the search request sentence, and is underlined, and “teacher” is a synonym of the keyword “teacher” of the search request sentence, so a wavy underline is added. . Alternatively, the above-mentioned display is not performed, or one of them is selected by the feature display selection unit 15 and sent to S13.

In S 13, the sentence data of the translated example sentence is displayed on the display unit 5. At this time, the bilingual example sentence database 8
8 may be displayed so that the source possessed by the user can be understood. In FIG. 8, the source of the example sentence of the cursor line is distinguished and displayed according to the source (in this case, the underline of the number is changed), and the source is displayed in detail at the bottom.

Further, display is performed in the mode selected in S12. The user creates a necessary English sentence by referring to the displayed English sentence of the bilingual example sentence, making a part or change, and returns to S1.

In S14, the contents of the plain word storage unit 13 as exemplified in FIG.
Figure 3 shows that the native language keyword is "dog" and its translation is "do
In the case of "g", the weight of the keyword is reduced because it is simple. When the native language keyword is "child" and the translated word is "child" or "kid", the weight of the keyword is reduced because it is plain. Furthermore, if the native language keyword is "television", it means that the weight of the keyword is reduced because the translation is easy, whatever the translation.

For editing, a screen as shown in FIG. 3 may be displayed so that the user can add or delete the data. Alternatively, as shown in FIG. You may do so. In S15, S
The data corrected in 14 is stored in the plain word storage unit 13,
Returning to S1, the process is repeated. Through the above-described processing, it becomes possible to search for and display an example sentence requested by the user.

[0052]

As is clear from the above description,
According to the present invention, a bilingual example sentence is more preferentially searched for and displayed when a foreign language sentence is created, thereby facilitating composition work. Also, since this can be set for each user, it can be used according to the language level of the user.

Further, the order of appearance of keywords and the distance between keywords are determined without a full-scale syntax analysis or semantic analysis requiring a large amount of processing, and a more useful similar example is output. As long as it is in the database, all keyword examples are
Since it can be displayed in the output display area, all the desired translation examples can be seen by one search operation, so that the operation time is reduced.

Further, it is possible to know at a glance where in the searched example sentence the match was found because of the match, and it is also possible to see at a glance whether the match was made with a synonym and the case where the match was completely made, so that the composition sentence is referred to. This is convenient when searching for sentences from search results.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention.

FIG. 2 is a flowchart showing the operation of one embodiment of the present invention.

FIG. 3 is a diagram showing an example of an easy word list editing screen.

FIG. 4 is a diagram showing an example of a screen displaying an example corresponding to all keywords.

FIG. 5 is a diagram showing an example of a screen displaying an example corresponding to all keywords.

FIG. 6 is a diagram showing an example of a screen for characteristically displaying a corresponding keyword.

FIG. 7 is a diagram showing an example of a screen for characteristically displaying a keyword corresponding to a synonym.

FIG. 8 is a diagram illustrating an example of a screen displaying a source of an example.

FIG. 9 is a diagram illustrating a storage example of keyword weighting data based on part of speech.

FIG. 10 is a diagram illustrating a calculation example of keyword weighting according to the present invention.

FIG. 11 is a diagram illustrating an example of a synonym table according to the present invention.

[Description of Signs] 1 search request input unit 2 morphological analysis unit 3 search request keyword extraction unit 4 search unit 5 display unit 6 morphological dictionary 7 keyword dictionary 8 bilingual example sentence database 9 keyword position detection unit 10 output adjustment unit 11 keyword weighting unit 12 Keyword weight processing unit 13 Simple word storage unit 14 Simple word editing unit 15 Feature display selection unit 16 Feature display mode flag

Claims (5)

[Claims] 1. A bilingual example sentence database having a bilingual example sentence in a first language and a second language, a bilingual example sentence database having a keyword dictionary, search request sentence input means for inputting a search request sentence in a first language, and the search Means for morphologically analyzing the request sentence using a morphological dictionary; keyword extracting means for extracting a keyword from a morphological analysis result of the search request sentence using the keyword dictionary; In a bilingual example sentence search device having a search means for extracting an example sentence containing the synonym and a display means for displaying the extracted translated sentence, the importance of the keyword is stored for each part of speech, and the weighting of the keyword is performed. Keyword weighting means for performing the following, and the search means considers the importance of the keyword by the keyword weighting means. Search translated example sentence search apparatus characterized by comprising a weight processing unit for performing. 2. An easy word storage means for storing an easy word, an easy word editing means for modifying the easy word storage means, and a keyword weighting means for reducing a weight of a word registered in the easy word storage means. 2. The bilingual example sentence search device according to claim 1, wherein the search means comprises: a weight processing means for performing a search in consideration of the importance of the keyword by the keyword weighting means. 3. A keyword position detecting means for recognizing a distance between keywords and an order of appearance; a keyword weighting means for performing a search weighting based on a result of the keyword position detecting means; 3. The bilingual example sentence search device according to claim 1, further comprising a weight processing unit for performing a search in consideration of the importance of the keyword. 4. The apparatus according to claim 1, further comprising output adjustment means for outputting, when displaying a search result, a bilingual example sentence for all keywords in an output display area. A bilingual example sentence search device. 5. A feature display selection for selecting whether the keyword in the search request sentence and the corresponding keyword in the bilingual example sentence are displayed in correspondence, or whether the keyword in the bilingual example sentence is completely matched or distinguished according to a synonymous match. 5. The bilingual example sentence search device according to claim 1, further comprising: means for storing a selected mode, and a feature display of a keyword based on the flag.