JP3343941B2 - Example sentence search system - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a system for searching a sentence similar to a search sentence from example sentences stored in advance.

[0002]

2. Description of the Related Art A conventional technique for searching for example sentences is disclosed in Japanese Unexamined Patent Publication No. Hei.
190972. There, we added information that grammatically constrains the search target using a text database that is categorized and stored in grammatical information such as part of speech, notation, part of speech, inflection type, meaning, etc. for each word constituting the example sentence. A system for searching for an example sentence by a search keyword is described.

[0003]

The prior art uses a one-word search keyword added with information that grammatically restricts the target in order to flexibly perform an example sentence search when searching for an example sentence. However, in a similarity search in which a plurality of words form a group and make a certain meaning, it is not always possible to accurately search for a similar example sentence for a search keyword by using only a thesaurus of one word or to narrow down the search. There is a problem.

[0004] An object of the present invention is to use, as a search key, a main word (a subject, a verb, and an object are collectively defined as a main word) extracted by syntactic analysis in a search sentence, and use these search keys or similar search keys. It is to search for an example sentence including a search key.

[0005]

The object of the present invention is that, when the main words form a group and have a certain meaning, a similar example sentence cannot be searched for even if the keyword is followed by a thesaurus in the thesaurus. This is achieved by replacing the main word of the search sentence with a semantically similar main word and searching for an example sentence containing the similar main word.

More specifically, a main word is extracted from a search sentence by syntactic analysis, and the extracted main word is associated with a semantically similar main word and stored in association with a main word related dictionary. Search for. The example sentence is searched using the main words from the search sentence as keywords. If the search of the example sentence fails, the main words that are semantically similar to the main words of the search sentence from the main word related dictionary (hereinafter referred to as similar main words)
Is searched, and an example sentence is searched from the example sentence database using the searched main word. An example sentence search procedure is shown below.

(1) An example sentence is searched from an example sentence database using a main word extracted from the search sentence as a search key. If the search is successful, the corresponding example sentence is displayed on the display device.
When the search fails, a main word similar to the main word used as the search key is searched for from the main word related dictionary, and an example sentence is searched from the example sentence database using the searched similar main word as a new search key. If there is no example sentence containing a similar principal word in the example sentence database, an example sentence similar to the search sentence is retrieved from the example sentence database by connecting the principal words having different character faces using the principal word similarity dictionary. The principal word similarity dictionary is a file storing information that one of the principal words serving as a search key is related to one of the other principal words by a thesaurus. In addition to the similar key words that serve as search keys, related information, such as the source field of the example sentence in the example sentence database and the extended information indicating the semantic connection with other example sentences, etc. A similar example sentence is searched from the example sentence database.

[0008] When a new principal word is added to the principal word related dictionary, a representative word, which is a representative principal word in the principal word related dictionary, is displayed to the user, so that which representative word is similar to the representative word is displayed. When the user makes a selection, it is registered in the principal word-related dictionary.

(2) When a compound word is included in a main word serving as a search key, information for decomposing the compound word is stored in order to decompose the compound word and replace it with a simple expression. By providing a compound expression decomposition dictionary, compound words in the main words are replaced with the main words of a decomposed simple expression. Next, an example sentence similar to the search sentence is searched from the example sentence database using the main word related dictionary based on the main words replaced with the simple expressions.

[0010] (3) If a key word that can be a search key does not exist in the search sentence, the user searches the source field or the extended information, which is related information of the example sentence database, to obtain an example sentence similar to the search sentence. Search using the example sentence database.

If an example sentence similar to the search sentence cannot be searched from the example sentence database, a message indicating that the search has failed is displayed.

[0012]

According to the present invention, the read sentence is parsed,
The subject, verb, and object (which are collectively defined as the subject in this specification) are extracted. However, as a result of parsing,
Even if any of the subject, verb, or object cannot be extracted, the search can be performed based on the extracted subject. In other words, the main word extracted from the search sentence is set as a search key, and a main word that matches the main word serving as the search key or a main word similar thereto is selected using the main word related dictionary and becomes the search key. By replacing the main words, an example sentence including a search key is searched from the example sentence database. If an example sentence cannot be searched for using the search key, the example sentence including the search key is searched from the example sentence database by associating the main word having a different face with the main word serving as the search key using the similar word dictionary. . As a result, it is possible not only to search for a similar expression for each word of the search sentence, but also to search for a main word similar to the meaning represented by a combination of plural words of the main word using the main word related dictionary, Example sentences that are semantically similar to the search sentence can be searched. In addition, when a compound word is included in the main words in the search sentence, the compound word is replaced with a simple expression using the compound expression decomposition dictionary, and is used as a search key. Example sentences can be searched from the example sentence database without greatly increasing the number of words. If the key words similar to the key words in the search sentence could not be retrieved from the key word-related dictionary, one of the key words was linked to the other key words by the thesaurus. Similar example sentences can be searched even if they do not match. Also, in addition to the search key, the user adds, as relevant information of the search sentence, a source field or extended information indicating a semantic connection with another example sentence, the source field stored in the example sentence database, The corresponding example sentence can be searched from the extended information. This allows
Even if the main word is missing in the search key, an example sentence semantically similar to the search sentence can be searched.

If the example sentence cannot be retrieved by the above example sentence search, the fact that the search failed is displayed to clearly show the user that the search failed.

[0014]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A first embodiment of the present invention will be described below with reference to FIG. In the following embodiment, an example sentence search based on a main word extracted from a search sentence will be mainly described.

In step 10, a search sentence to be searched is read from the input device 110 shown in FIG. In step 20, the read search sentence is parsed to extract main words (subject, verb, object), and the search sentence is stored in the search sentence column 1502 of the main word extraction table 160 shown in FIG. Subject column 1504 and verb column 150 in column 1503
5. The extracted main words are stored in the object field 1506. Depending on the storage status, the subject, verb,
It is determined whether each object exists. The syntax analysis technique used here is a well-known technique, and details thereof are described in, for example, the syntax analysis in Chapter 5 of Shin Nagao and Japanese Information Processing (IEICE).
Here, the description is omitted. However, here, syntactic analysis is used in a broad sense including morphological analysis and multiple part-of-speech resolution.
If the principal word exists in the search sentence in step 20, a search key expressed by a combination of the principal words extracted in step 30 is set. The details will be described later with reference to FIG. In step 40, a search is performed using the search key represented by the main word set in step 30 and a main word related dictionary 1710 storing main words similar to the search key. The detailed configuration of the main word-related dictionary 1710 will be described later with reference to FIG. 13, and the details of step 40 will be described later with reference to FIG. Next, in step 50, a main word serving as a search key is set based on the main word searched in step 40, and an example sentence similar to the search sentence is searched from the example sentence database 180 in which the example sentences are stored in advance. Details of step 50 will be described later with reference to FIG. Next, in step 60, it is determined whether or not an example sentence similar to the search sentence has been retrieved from the example sentence database up to step 50. If the result of the determination in step 60 is that the search was not successful, it is determined in step 70 whether the search key used in the search in step 40 is one word. If the search key is one word, step 9
In order to confirm to the user that the search failed with 0,
This is displayed on the output device 130 shown in FIG. On the other hand, in step 70, the search key is not one word,
If the number of words is equal to or more than the word, a process of reducing one word in the search key is performed in step 80. Details of steps 70 to 90 will be described later with reference to FIG. Step 10
0, the result searched in step 50 is output to the output device 130.
(FIG. 7). Details of step 100 will be described later with reference to FIG.

Next, step 30 will be described in detail.
As shown in FIG. 2, first, in step 301,
Key word as search key key word search key storage table 1
The main word storage table 1610 in FIG. 60 (FIG. 9) (FIG. 1)
The subject field 1613 and the verb field 16 of the subject field 1612 of 1)
14. Stored in the object field 1615. Next, step 3
02, as information for searching, step 3
The user is inquired whether to add new information separately from the main word stored in 01. The inquiry method is
This is a normal product of the output device 130 and the input device 110, and the description is omitted. If the user does not add the information, the main word column 161 of the main word storage table 1610
In step 40 and step 50, an example sentence similar to the search sentence is searched from the example sentence database 180 based on the main word stored in the second sentence. When the user adds information in step 302, the source field information is stored in the source field column 1623 of the related information storage table 1620 of the key word search key storage table 160 as related information of the search sentence in step 303. In step 304, if the user further adds information related to the search sentence to the information stored in the key word search key storage table in steps 301 and 303, the key word search key storage table 1
The example sentence database 180 searches the example sentence database 180 for an example sentence similar to the search sentence based on the information in the source field column 1623 of the main word storage table 1610 and the related information storage table 1620. If the user adds further information related to the search sentence in step 304, in step 305,
As related information of the search sentence, extended information indicating the semantic connection between the search sentence and another search sentence is stored in the extended information column 1624 of the related information storage table 1620 of the key word search key storage table 160. Therefore, an example sentence similar to the search sentence is searched from the example sentence database 180 based on the related information stored in the key word search key storage table 160.

The details of the main word related dictionary 1710 will be described. As shown in FIG. 13, the principal word related dictionary 1710
It is composed of a main word number column 1711, a related reference word column 1712, a representative word column 1713, a related word column 1714, and an example sentence number column 1715. Here, the related reference word column 1712 stores a reference word for linking related main words, and the representative word column 1713 stores a representative word among the relevant main words. The related word column 1714 stores principal words related to the principal words stored in the representative word column 1713. The example sentence number column 1715 stores an example sentence number in which an example sentence including the main word stored in the upper part of the example sentence number column 1715 is stored.

Next, the details of step 40 will be described with reference to FIG. In step 401, the information in the key word search key storage table 160 is stored in the key word storage table 1610.
It is determined whether there is only information stored in. If the result of the determination is that there is only the key word storage table information, in step 402, the key words that match the key words extracted from the search sentence stored in the key word storage table 1610 of the key word search key storage table 160 Into the related reference term section 1712 and the representative term section 1713 of the principal term related dictionary 1710,
Alternatively, the search is performed from the main words stored in the related reference word column 1712 and the related word column 1714. Step 401
As a result of the judgment, the key word search key storage table 16
In addition to the main words stored in the main word storage table 1610, the related information storage table 1620
If the source field information and the extended information, which are information related to the search sentence, are stored in step S403, the process proceeds to step 403. In step 403, the key word search key storage table 160
Of the search sentence stored in the related information storage table 1620 and the source field column 182 or the extended information column 18 of the example sentence database 180.
3 is searched for an example sentence whose extension information matches. In step 404, the main words that match the main words extracted from the search sentence in step 402 are stored in the main word related dictionary 1
It is determined whether or not it exists at 710. As a result of the determination, when the matching main word is in the main word related dictionary 1710,
Proceed to step 405. On the other hand, if there is no matching main word in the main word related dictionary 1710, the process proceeds to step 406. In step 406, the main words stored in the main word column 1713 of the main word related dictionary 1710 are displayed, and the user specifies a main word that is thought to be related to the main word extracted from the main sentence, and Is registered in the related word column 1714 of the main word related dictionary 1710 as a similar main word to the designated representative word. In step 405, it is determined whether or not an example sentence including the main word extracted from the search sentence exists in the example sentence database 180. As a result of the determination, when an example sentence including the extracted main word is present in the example sentence column 184 of the example sentence database 180, the process proceeds to step 50, and an example sentence including the main word serving as a search key is searched from the example sentence database 180. On the other hand, if the example sentence including the extracted main word does not exist in the example sentence database 180, the process proceeds to step 407. Step 4
At 07, a main word similar to the main word extracted from the search sentence is searched using the main word related dictionary 1710. That is, a word that matches one of the main words extracted from the search sentence is searched from the words stored in the related reference word column 1712 of the main word related dictionary 1710. Further, a word that matches one of the extracted main words is searched from the words stored in the representative word column 1713 and the related word column 1714. as a result,
In the next step 408, in the same column as the main word searched from the representative word column 1713 and the related word column 1714,
All of the similar principal words whose example sentence numbers are stored in the example sentence number column 1715 are the principal word storage tables 161 of the principal word search key storage table 160 as the main words of the retrieval sentence.
0 is stored. Regarding similar main words in storage, the search number column 1611 of the main word storage table 1610
, The first search number is common, and the subsequent search numbers are sequentially numbered from 1 and stored. Using this search key, in step 50, an example sentence including a main word serving as a search key is searched from the example sentence database 180.

Next, step 50 will be described in detail. As shown in FIG. 4, first, in step 501, the main word storage table 1 of the main word search key storage table 160 is set.
Among the search numbers stored in the search number column 1611 of 610, it is determined whether or not the main words are stored in ascending order of the later numbers among the common numbers of the first number. If it is stored, the example sentence number column 171 of the principal word related dictionary 1710 is stored based on the stored principal word.
Using the example sentence number stored in No. 5, an example sentence including the main word is searched from the example sentence database 180. Next, for the similar key word next to the key word searched earlier in the key word search key storage table 1610, step 5 is repeated.
The process is repeated from 01. If it is determined in step 501 that there is no main word different from the main word that has been processed so far, the process of step 50 ends, and the process goes to step 60.

Next, steps 70 to 90 will be described in detail with reference to FIG. First, in step 70,
It is determined whether the search key stored in the main word storage table 1610 of the main word search key storage table 160 is one word, that is, whether the main word serving as the search key is composed of one word. If the result of the determination is that the search key is two or more words, the procedure proceeds to step 801. In step 801, main words (subject, verb, and subject) stored in the main word storage table 1610 of the main word search key storage table 160 are stored.
The user selects one word to be reduced from (object words).
Next, in step 802, it is displayed whether or not one word selected in step 801 is correct, and the user is urged to confirm whether or not the word is correct. If step 70
In step 1, if the selected one word is incorrect, the process returns to step 701 again, and the main words (subject,
The user again selects one word to be deleted from the verbs and object words). If there is no error in the selected one word in step 802, the procedure goes to step 803. Step 803
Then, a principal word from which one of the principal words serving as the retrieval key selected in step 801 is deleted is set as a retrieval key again and stored in the principal word storage table 1610 of the principal word retrieval key storage table 160. Next, in step 804, an example sentence including the search key is searched from the example sentence database 180 using the principal word related dictionary 1710 using the search key set in step 803. As a result, when the search is successful, the processing of steps 70 to 90 is ended. When the search is not successful, the procedure goes to step 805.
In step 805, it is determined whether any of the key words (subject, verb, object) serving as the search key has been deleted. If it is determined that a word which has not been deleted exists in the main word, step 80
Go to 7. In step 807, step 805
The one word deleted so far is restored again, and the same processing as in step 801 is performed for words in the main words which have not been deleted yet. On the other hand, if it is determined in step 805 that a word which has not been deleted does not exist in the main word, the process proceeds to step 806. In step 806, the principal word obtained by deleting one word specified by the user in step 801 from the search key is stored again as a search key in the principal word storage table 1610 of the principal word search key storage table 160, and the processing is repeated from step 70. repeat. In the first step 70, if there is only one word (subject, verb, object) constituting the main word as a search key, the process goes to step 90. In step 90, a process is performed to display to the user that the search failed because a similar example sentence from the search sentence could not be searched from the example sentence database 180.

The detailed description of steps 70 to 90 has been completed, and step 100 will now be described in detail. As shown in FIG. 6, first, in step 1001, the example sentence number of the searched example sentence including the main word is searched from the example sentence number stored in the example sentence number column 1715 of the main word related dictionary 1710. Next, step 1002
In step, an example sentence that matches the example sentence number searched in step 1001 is searched from the main related number column 181 and the example sentence column 184 of the example sentence database 180 and called. Next, in step 1003, the example sentence called in step 1001 is displayed on the output device 130 of FIG.

Next, in order to specifically explain the above processing, the following input sentence (search sentence) will be taken as an example.
"Mr. A may fall in the mayoral election to be held next month." First, a parsing process is performed on the input sentence to extract key words. In the case of the above search sentence, the subject is "Mr. A", the verb is "fall", and the object is "election".
Becomes The extracted main words are stored in the search sentence column 1502 of the main word extraction table 150 and the main column 150 of the main word column 1504.
4, stored in the verb column 1505 and the object column 1506, respectively. If it is the first search sentence, “1” is stored in the search number column 1501, and the subsequent search sentences are sequentially incremented in number. Next, the key word extraction table 150
Is set as a search key, and the main word storage table 161 of the main word search key storage table 160 is set.
“Mr. A” is stored in the subject field 1613 of the 0 subject field 1612, “fall” in the verb field 1614, and “election” in the object field 1615. Next, it is determined whether the user further adds related information of the search sentence. If the user decides to search for an example sentence similar to the search sentence using only the main word, the relevant criterion of the main word related dictionary 1710 that matches the main word stored in the main word storage table 1610 of the main word search key storage table 160 The related reference word stored in the word column 1712 is searched. As a result of the search, “election”, which is a related reference word that matches the extracted main word, is found in main word number 1 of the main word related dictionary 1710, and is searched. Next, a key word search key storage table 16 from the representative word related to the related reference word and the related word information is stored.
A search is made for a word that matches the key word stored in the 0 key word storage table 1610. In the case of the above-described search sentence, one of the main words “fall” is present in the related word column 1714 of the main word related dictionary 1710, and therefore this is searched first. Next, an example sentence number corresponding to “fall” which is related word information of the searched related word section 1714 is searched from the example sentence number section 1715. As a result of the search, when an example sentence number is stored in the example sentence number column 1715, a search sentence corresponding to the example sentence number is searched from the example sentence database 180. The example sentence number corresponding to the related word "fall"
Since it is not stored in the example sentence number column 1715, the example sentence cannot be searched as it is. Accordingly, the representative word column 1713 and the related word column 1713 of the principal word related dictionary 1710 stored as expressions similar to the related word “fall”.
Of the information stored in No. 14, a search is performed for all of the information in which the corresponding example sentence number is stored in the example sentence number column 1715, and the search result is stored in the main word search key storage table 160 in the main word. The data is sequentially stored in the table 1610. In the case of the above search sentence, the related word "fall"
As the main words similar to the above, and the example sentence number information is stored, there are “defeated” stored in the representative word column 1713 and “defeat” stored in the same related word column 1714. . These two pieces of information and “election” stored in the related reference word information column 1712 are stored in the keyword storage table 16 of the keyword search key storage table 160.
10 is stored. At this time, since the main words extracted from the search sentence are unnecessary, the main words extracted from the search sentence already stored in the main word column 1612 of the main word search table 1610 are deleted, and the main word related dictionary 1710 is searched again. Stored main words. The key word storage table 161
In the search number column 1611 of 0, the first number is assigned a common number, and the numbers after the hyphen are stored with search number information assigned in the search order. Based on the main words set as described above, an example sentence similar to the above-described search sentence is searched from the example sentence database 180 using the example sentence numbers stored in the main word related dictionary 1710. That is, since the example sentence numbers stored in the example sentence number column 1715 of the principal word related dictionary 1710 are 1-1-A and 1-1-C, respectively.
When the example sentence number stored in the main related number column 181 is compared with the example sentence database 180, the example sentence “He loses the election.”
May lose to city council election. Is called. The called example sentence is output device 1 in FIG.
30 and displayed and output as an example sentence similar to the search sentence described above. The above is the processing content for searching for an example sentence similar to the search sentence when the search sentence is actually input.

Next, FIG. 7 shows a hardware configuration for realizing this embodiment. The input sentence (search sentence) is input from the input device 110, and the example sentence search is executed by the CPU 120. Next, the example sentence search result is displayed by the output device 130. The storage device 140 is roughly divided into a key word extraction table 150 and a key word search key storage table 16.
0, a main word dictionary 170, an example sentence database 180, and an area for storing a search program 190. Further, the key word search key storage table 160 includes a key word storage table 1610 and a related information storage table 1620. The principal word dictionary 170 is composed of a principal word related dictionary 1710, a compound expression decomposition dictionary 1720, and a principal word similar dictionary 1730.

The compound expression decomposition dictionary 1720 and the principal word similarity dictionary 1730 will be described in detail in the second and subsequent embodiments of the present invention.

Next, a second embodiment of the present invention will be described with reference to FIG.

Steps 10 'to 30' are the same as steps 10 to 30 in the first embodiment, and a description thereof will be omitted. Step 30 '
Next, go to step 3000. In step 3000, if a compound word is included in the main word serving as a search key, the compound word is decomposed by the compound expression decomposition dictionary 1720 and replaced with a basic expression. The main word in the main word search key storage table 160 in which the main word is stored is reset. Therefore,
Step 3000 will be described in detail with reference to FIG. First, in step 3001, it is determined whether or not a compound word is included in a main word serving as a search key. If the result of the determination is that a compound word is included in the main word serving as a search key, the process proceeds to step 3002. In step 3002, the compound words in the main words serving as search keys are replaced with basic expressions using the compound expression decomposition dictionary 1720. Here, the compound expression decomposition dictionary 1720 will be described in detail. The compound expression decomposition dictionary 1720 is mainly composed of a main word column 1712 containing compound words and a main word column 1722 of a basic expression. The subject columns 1723 and 1726, the verb columns 1724 and 1727, and the object column are respectively provided. 1725 and 1728. Next, in step 3003, the principal word replaced with the basic expression in step 3002 is entered in the principal word column 1 of the principal word storage table 1610 of the principal word search key storage table 160.
612 subject field 1613, verb field 1614, object field 1
615. This concludes the detailed description of step 3000. Next step 40 'to step 100'
About steps 40 to 100 of the first embodiment.
Since the processing is the same as described above, the description is omitted here.

Next, in order to specifically explain the processing of the second embodiment, the following input sentence (search sentence) will be taken as an example. "He will be in the city council election next week,
I will lose. First, a syntax analysis process is performed on the input sentence to extract a main word. In this case, the subject is "he", the verb is "rejected", and the object is "city council election". Next, the input search sentence is stored in the search sentence column 1502 of the main word extraction table 150, and the main words are stored in the subject column 1504, the verb column 1505, and the object column 1506 of the main word column 1503, respectively. Next, it is determined whether or not a compound word is included in the extracted main words. As a result of the judgment, the compound word "rejected" is included in the main words
Exists, the compound word is replaced with the basic expression using the compound expression decomposition dictionary 150. In other words, from “delete” stored in the verb column 1724 of the principal word column 1721 containing the compound word, “fall” stored in the verb column 1727 of the principal word column 1722 of the basic expression and the object column Reference is made to “Elections” stored at 1728. Then, the compound word “rejected” is replaced with “fall” and “election”. Furthermore, the key words are extracted from the search sentence and stored in the key word storage table 1610 of the key word search key storage table 160, and the compound words are replaced with the key words obtained by replacing the compound words with the basic expressions. To be stored. By the above processing, the search key storage table 160
Based on the main words stored in the main word storage table 1610, an example sentence similar to the search sentence is searched using the main word related dictionary 1710. In the case where an example sentence is searched from the example sentence database 180 using the main words, the first embodiment is described.
Since this is the same as the example of the input sentence (search sentence) shown by, the description is omitted here. Example sentence database 180
As a result of the search, an example sentence similar to the search sentence is displayed in an example sentence column 184.
An example sentence "He loses the election." And an example sentence "Ms. B may lose the city council election." The above example sentences are displayed on the output device 130 in FIG.

Next, a third embodiment of the present invention will be described with reference to FIG.

Steps 10 "to 40" are the same as steps 10 to 40 of the first embodiment, and a description thereof will be omitted. Next, in step 4100, it is determined whether or not the main term serving as the search key set in step 40 ″ is stored in the main term related dictionary 1710. If the result of the determination is that the search was not successful, the procedure goes to step 4200. Step 420
At 0, a similar word of the search key is searched using the principal word similarity dictionary 1730. Here, the principal word similar dictionary 173
“0” is stored in the reference word column 1731 with the word that comes into the higher concept in the thesaurus as the reference word, and similar words belonging to the reference word are stored in the similar word column 1732. That is, in step 4200, a word that matches one of the main words serving as a search key is searched from the similar words stored in the similar word column 1732, and a word that is a superordinate concept of the searched similar word is determined as a reference. The search is performed from the reference words stored in the word field 1731. Further, the stored main words serving as the search key, the words searched using the main word similar dictionary 1730 are replaced with the searched reference words. next,
In step 4300, it is determined in step 4200 whether the reference word has been retrieved from the similar words. If the result of the determination is that a reference word that eventually becomes a superordinate concept could not be retrieved, the procedure goes to step 4400. Step 4
In 400, one of the main words is stored in the related reference word column 1712 and the representative word column 1713 of the main word related dictionary 1710. Based on the principal word related dictionary 1730 updated in step 4400, step 40 '' is executed again.
Is repeated. On the other hand, if the result of the determination in step 4300 is that the search was successful, the process returns to step 40 ′ based on the replaced key words as search keys.
Repeat the processing from '. By the way, step 4100
In, when a main word similar to the main word serving as a search key can be searched from the main word related dictionary 1710 without using the main word similar dictionary 1730, the process goes to step 50 ''. Steps 50 "to 100" are the same as steps 50 to 100 of the first embodiment, and thus description thereof is omitted.

Next, in order to specifically explain the processing of the third embodiment, the following input sentence (search sentence) will be taken as an example. "Mr. C will lose to the city election." First, the input sentence is subjected to a parsing process to extract key words. In this case, the subject is "Mr. C", the verb is "rejected", and the object is "city election". These search sentences and the main words are stored in the search sentence column 1 of the main word extraction table 150.
The input search sentence is stored in 502 and the main word field 150 is stored.
3 subject field 1504, verb field 1505, object field 150
6 respectively. Next, the main word related dictionary 1710
Search for similar key words using. In the case of the above-mentioned search sentence, since there is no match among the related reference words stored in the related reference word column 1712 of the principal word related dictionary 1710, the search key is searched using the similar word similar dictionary 1730. A similar word of one of the main words is searched. In the case of the above-described search sentence, a reference term, which is a superordinate concept, is searched for, using the principal word similarity dictionary 1730, for the object "city election" among the key words as the search key. As a result of the search, a word that is a superordinate concept of the object “city council election” is entered in the reference word field 173.
Since “election” is stored in No. 1, the object “city election” is replaced with “election” among the key words serving as search keys. Next, the replaced main word is stored in the main word storage table 1610 of the main word search key storage table 160, and an example sentence similar to the search sentence is searched using the main word related dictionary 1710. The case where the example sentence is searched from the example sentence database 180 using the main words is the same as the case of the example of the input sentence (search sentence) shown in the first embodiment.
Here, the description is omitted. As a result of a search from the example sentence database 180, an example sentence similar to the search sentence in the example sentence column 184 is an example sentence "He is defeated in the election" and an example sentence "Ms. B may lose the city councilor election." Become. The above example sentences are displayed on the output device 130 in FIG.

Next, a fourth embodiment of the present invention will be described with reference to FIG.

Steps 10 '''to100'''
′, Steps 10 to 10 of the first embodiment
Since the process is the same as 0, the description is omitted here. However, unlike step 30 of the first embodiment, step 30 ″ ″ sets a search key only by a combination of the main words. On the other hand, if it is determined in step 20 ′ ″ that neither the subject, the verb, nor the object, which is the principal word, exists in the search sentence, the process proceeds to step 5100. In step 5100, in the related information storage table 1620 of the search key storage table 160, the source field information as the related information of the search sentence and the extended information indicating the semantic connection with another example sentence are stored in the source field of the related information column 1622. Column 162
No. 3 field 1625, detailed information field 1626, and extended information field 1624. Next, in step 5200, the source field information and the extended information, which are the related information of the search sentence stored in the related information storage table 1620, are called. Next, in step 5300, step 520
Using the related information called at 0 as a search key, an example sentence having the same related information as the search key in the source field field 182 and the extended information field 183 of the example sentence database 180 is searched from the example sentence field 184. As a result, step 5400
When the example sentence search is successful,
At 0 ''', an example sentence search result is displayed, and when the search fails, display of search failure is performed at step 90'''.

Next, in order to specifically explain the processing of the fourth embodiment, the following input sentence (search sentence) will be taken as an example. "Hello." First of all, performs the syntax analysis processing with respect to the input sentence, to extract the main language. However, in the case of the search sentence, since it is impossible to extract the main words,
The subject, the verb, and the object, which are the main words, cannot be stored in the main word extraction table 150 in the search sentence. That is, there is no principal word in the search sentence. Therefore, the source field and extended information, which are information related to the search sentence, are stored in the field field 1625 of the source field field 1623 of the related information storage table 1620 of the key word search key storage table 160 by “greeting”. , Detailed information column 16
For example, “when visiting a person's house” is stored in “26”, and “expression used during daytime” is stored in the extended information column 1624, respectively. Next, an example sentence related to the search sentence is searched using the example sentence database 180 using the stored related information as a search key. As a result, the source field information column 182 of the example sentence database 180 stores the same “when visiting a person's house” as the search key, and thus “sorry.” Stored in the example sentence column 184 is searched. Is called. The searched example sentence is displayed on the output device 130 in FIG.

[0034]

According to the present invention, a similar term is searched for by using a principal term related dictionary to search for a similar term similar to the principal term (subject, verb, object) extracted from the retrieval sentence. Example sentences including the search can be searched. That is, a main word is extracted from the search sentence, and an example sentence including the main word is searched using the example sentence database. When the example sentence containing the extracted main word does not exist in the example sentence database, the main word similar to the main word extracted from the search sentence is searched using the main word related dictionary to include the similar main word. The example sentence semantically similar to the search sentence can be searched using the example sentence database. In addition, when the main words cannot be extracted from the search sentence, using the example sentence database from the related information of the search sentence,
It is also possible to search for example sentences related to the search sentence.

[Brief description of the drawings]

FIG. 1 is a flowchart showing a first embodiment.

FIG. 2 is a detailed diagram of step 30.

FIG. 3 is a detailed diagram of step 40.

FIG. 4 is a detailed diagram of step 50.

FIG. 5 is a detailed diagram of steps 70 to 90.

FIG. 6 is a detailed diagram of step 100.

FIG. 7 is a hardware configuration diagram of an embodiment.

FIG. 8 is a configuration of a main word extraction table.

FIG. 9 is a key word search key storage table.

FIG. 10 is a main word dictionary.

FIG. 11 shows a configuration of a main word storage table.

FIG. 12 is a configuration of a related information storage table.

FIG. 13 is a configuration of a main word related dictionary.

FIG. 14 shows the configuration of a compound expression decomposition dictionary.

FIG. 15 shows the configuration of a principal word similar dictionary.

FIG. 16 shows a configuration of an example sentence database.

FIG. 17 is a flowchart showing a second embodiment.

FIG. 18 is a detailed diagram of step 3000.

FIG. 19 is a flowchart showing a third embodiment.

FIG. 20 is a flowchart showing a fourth embodiment.

[Explanation of symbols]

30: a process of setting a search key based on the relevant information of the main word and the search sentence; 40: a process of searching a main word related dictionary for a main word similar to the main word to be the search key; 50: a main word to be the search key A process of searching for an example sentence included in the example sentence database, 110 ... input device, 120 ... CPU, 130
... output device, 140 ... storage device, 150 ... key word search table, 160 ... key word search key storage table, 170
… Key word dictionary, 1710… Key word related dictionary, 1720…
Compound expression decomposition dictionary, 1730...
… Example sentence database, 190… search program, 300
0: a process of converting a compound word into a basic expression using a compound expression decomposition dictionary, 4200: a process of performing a thesaurus search for similar words of a main word from a similar word dictionary, 5100: setting related information of a search sentence as a search key Process 5300: A process of searching an example sentence database for an example sentence related to a search sentence from related information serving as a search key.

Continuation of the front page (56) References JP-A-3-148765 (JP, A) JP-A-2-253474 (JP, A) Yukinori Terahama, Example sentence retrieval method in English composition support system, Information Processing Society of Japan Proceedings of the 45th National Convention (3), October 11, 1994, p. 145-146 (58) Field surveyed (Int. Cl. ⁷ , DB name) G06F 17/30

Claims (3)

(57) [Claims] (1)Search sentence from example sentences stored in advance
Example sentence search system that searches for similar example sentences similar to
hand, Said An example sentence database storing example sentences,Said Search sentenceFrom, Subject, verb and objectIncluding key words
Extraction means for extracting One of the subject, verb and object, and the subject, verb
The other one of the lyric and object and the other one of the words
Including consecutive words Principal language related dictionary file,Includes the subject, verb and object of the extracted subject
If the example sentence is the same as the similar example sentence,
First search means for searching from  The similar example sentence can be searched by the first search means.
If not found, the subject, verb and
Key words containing words that match one of the words
Extract a dictionary file and extract the extracted key word related dictionary
Extract the relevant words from a file, and
One of the subject, verb and object of the subject and extracted
An example sentence that includes the related word
Second search means for searching from the example sentence database
When,The similar example sentence can be searched by the second search means.
If not found, the subject, verb and
And reducing means for reducing one of the object words, The first search means and the second search means are configured to
In place of the spoken subject, replace the subject with one less word
To retrieve the similar example sentence from the example sentence database.
Search Example sentence search system. 2. The example sentence retrieval system according to claim 1 , wherein the compound, a subject, a verb, and an object obtained by decomposing the compound.
A compound word decomposition dictionary file containing words and, when the extracted main words include compound words,
Using the compound word decomposition dictionary file, the extracted
The compound word that is included in the main language, subject, verb you and purpose
Decomposition means for decomposing into words, wherein the first retrieval means and the second retrieval means
Instead of the subject, verb and object of the subject
Using the subject, verb and object interpreted,
An example sentence search system for searching for a similar example sentence from the example sentence database . 3. The example sentence retrieval system according to claim 1, wherein one of the subject, the verb and the object,
One of the subject, verb and object was superordinated
A principal word similar dictionary file including a reference word, and the second search unit outputs a subject of the extracted principal word,
A word including a word that matches one of a verb and an object word
If the main language related dictionary file cannot be extracted,
One of the subject, verb, and object of the subject issued
The reference word, which is a high-level concept of
Extraction means for extracting from the file, the first search means and the second search means
Instead of the subject, verb and object of the subject
Using the reference word issued, the similar example sentence
An example sentence search system that searches from a sentence database .