JPH09128405A - Method and device for retrieving document - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve the retrieving speed by reducing the data quantity read out from indexes arranged on an external storage and to improve the retrieving speed with respect to a retrieving key which often consists of the characters of the same kind by changing the arrangement of character position information within the indexes. SOLUTION: This device is provided with a max. and min. ID extraction part 15 picking the min. ID number and the max. ID number of a character string constituting plural retrieval key word from a retrieval expression inputted at the time of retrieval, and an index reading part 16 reading position information for the number of IDs between the min, and max. ID numbers of the character string from the index. Then, the min. and max. ID numbers of the character string constituting the plural key retrieval key words from the retrieving expression inputted at the time of retrieval are retrieved to retrieve position information necessary for retrieval by reading not position information of all the character string but that between the min. and max. ID numbers from the index.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a document retrieval apparatus for retrieving a document registered in a document database or the like based on a keyword, and more particularly to a full-text search type document retrieval apparatus in which the entire text of a document is retrieved. The present invention relates to a document search method used in this device.

[0002]

2. Description of the Related Art Conventionally, in a full-text search in which the entire text of a document is searched, the original document is not searched as it is, but a search-specific index created in advance is searched. was there. Since this index has a large capacity, it is often placed in an external storage device such as a hard disk device instead of a memory, and a search using an index with such a format
A lot of time was spent reading the index on the external storage device.

Further, the entered search keys are of the same character type,
That is, it is often composed of all katakana, all kanji, etc., and an effective search focusing on this information has not been made.

[0004]

As described above, in the conventional document retrieval, since the index has a large capacity, it is often arranged not in the memory but in the external storage device such as the hard disk device. In a search using an index having a different format, much time was spent reading the index on the external storage device.
Further, the input search key is often composed of the same character type, that is, all katakana, all kanji, etc., and an effective search focusing on this information has not been performed.

The present invention has been made in consideration of the above circumstances. By reducing the amount of data read from an index arranged on an external storage device, the search speed is improved and the characters in the index are also improved. An object of the present invention is to provide a document search device and a document search method that change the arrangement of position information to improve the search speed for search keys that are often composed of the same type of characters.

[0006]

In order to achieve the above object, the present invention assigns a unique ID number to each character string such as a character or a word and assigns a unique ID number to each character string appearing in one document.
A document search method for performing a document search based on an index for all documents that stores appearance position information of character strings in a format that can be retrieved in the order of D numbers or for each ID number. The minimum ID number and the maximum ID number of the character strings that form the search keyword are extracted, and the positional information required for the search is not the positional information of all the character strings, but the positional information between the minimum ID number and the maximum ID number. The feature is that retrieval is performed by reading from the index.

Further, in order to achieve the above object, the present invention provides the above-mentioned document retrieval method, in which a unique ID number is assigned to each character string by limiting the character string to a character and allocating a symbol / alphanumeric character. A continuous character ID is assigned to each character type, which is a classification of hiragana, katakana, kanji, etc., and it is possible to search when all keywords input at the time of search are composed of the same type of characters.

Further, the present invention has the following objects to attain the above objects.
A unique ID number is assigned to each character string such as a character or word, and all documents storing appearance position information of the character strings that appear in one document can be taken out in the order of the ID numbers or for each ID number. Is a document search device for performing high-speed document search based on the index of, and an input means for inputting a search expression or a processing instruction, and a character string forming a plurality of search keywords from the search expression input at the time of search Minimum ID
Number and maximum ID number, maximum / minimum ID extraction means, index reading means for reading position information for the number of IDs between the minimum ID number and maximum ID number of the character string from the index, a search means, and a processing result output And an output means for performing the same.

Further, in order to achieve the above object, the present invention provides
In the above document retrieval apparatus, a unique ID number is assigned to each character string because the character string is limited to characters and consecutive character IDs are assigned to each character type.

According to the above construction, a unique ID number is assigned to each character string of characters or words, all the appearance positions of the characters are extracted in the order of the ID numbers of the characters appearing in one document, and each ID number is extracted. The index for all documents in a format in which the appearance position information can be acquired is arranged for each document and the character string position information of the document is arranged in the order of the character string ID number. Minimum ID of the character string that constitutes multiple search keywords described in the search formula
The number and the maximum ID number are extracted, and the position information for the number of IDs between the minimum ID number and the maximum ID number of the character string is read from the index, so that the number of all the IDs in the document is read, as compared with the case where it is read. The index information can be reduced and the search speed can be improved.

Further, in the above construction, by assigning consecutive ID numbers to the characters of the same character type, the index has a structure in which the positional information of the characters is collected by the same character type. As a result, the difference between the minimum ID number and the maximum ID number of the search keyword that is often composed of the same character type becomes small, and as a result, the amount of character position information read from the index becomes smaller and the search speed can be improved. Becomes

[0012]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The outline of the present invention is as follows. That is, a unique ID number is assigned to each character string of characters or words, all the appearance positions of the character strings are extracted in the order of the ID numbers of the character strings appearing in one document, and the appearance position information is obtained for each ID number. A document search method for creating an index for all documents in a format that can be acquired and performing a search using this index, in which the indexes on the external storage device are arranged for each document and are arranged in the document in the order of ID numbers of character strings. Paying attention to the fact that the position information of existing character strings are arranged side by side, the minimum ID number and the maximum ID of the character strings that make up a plurality of search keywords described in the search formula input at the time of search
By extracting the number and reading the position information of the number of IDs between the minimum ID number and the maximum ID number of the character string from the index, the amount of position information is reduced compared to reading the position information of all IDs. This is a method of reading from the index.

Further, the character string to which the unique ID number is assigned is limited to characters, all the appearance positions of the characters are extracted in the order of the ID numbers of the characters appearing in one document, and the appearance position information is acquired for each ID number. This is a document search method for creating an index for all documents in possible formats and performing a search using this index. When assigning a unique ID number to a character, an index that assigns consecutive ID numbers to the same character type is used. By creating, the minimum ID number and maximum ID of search keywords that are often composed of the same character type
Character ID characterized by reducing the number difference and reducing the amount of position information read from the index
There is a method of assigning numbers.

Further, a unique ID for each character or character string
ID of the character string that appears in one document by assigning a number
Extract all the appearance positions of the character string in the order of numbers, and
A document search device for creating an index for all documents in a format in which appearance position information can be acquired for each D number, and performing a search using this index, including an input means for inputting a search expression and a processing instruction, and a search Maximum / minimum ID extraction means for extracting the minimum ID number and the maximum ID number of the character strings forming the plurality of search keywords from the search expression that is input at some time, and the number of IDs between the minimum ID number and the maximum ID number of the character string. A document retrieval apparatus comprising: an index reading unit that reads out position information from an index; a retrieval unit; and an output unit that outputs a processing result.

Further, a character string to which a unique ID number is assigned is limited to characters, consecutive ID numbers are assigned to characters of the same character type, and all the appearance positions of the characters appear in the order of the ID numbers of the characters appearing in one document. Is a document search device that creates an index for all documents in a format in which appearance position information can be acquired for each ID number, and uses this index to perform a search, and inputs a search expression and a processing instruction. Means, a maximum / minimum ID extracting means for extracting the minimum ID number and the maximum ID number of the characters constituting the plurality of search keywords from the search expression input at the time of the search, and the number of IDs between the minimum and maximum ID numbers of the characters A document detecting device comprising: an index reading unit for reading the position information of the minutes from the index; a searching unit; and an output unit for outputting the processing result. Apparatus is in.

In the above arrangement, a unique ID number is assigned to each character string of characters and words, all the appearance positions of the characters are extracted in the order of the ID numbers of the characters appearing in one document, and A search is performed by using the index for all documents in a format in which the appearance position information can be acquired for each ID number, arranged for each document, and the character string position information of the document arranged in the order of the character string ID number. The minimum ID number and the maximum ID number of the character string that compose the plurality of search keywords described in the search formula that is input at some time are extracted, and the number of IDs between the minimum ID number and the maximum ID number of the character string is extracted from the index. By reading the position information, all I's in the document
The index information to be read can be reduced and the search speed can be improved as compared with the case where the number of D is read.

Further, in the above structure, by assigning consecutive ID numbers to the characters of the same character type, the index has a structure in which the positional information of the characters is fixed in the same character type. As a result, the difference between the minimum ID number and the maximum ID number of the search keyword that is often composed of the same character type becomes small, and as a result, the amount of character position information read from the index becomes smaller and the search speed can be improved. Becomes

An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing a schematic configuration of a document search device. Reference numeral 1 is a control device composed of a CPU and a memory for searching a document and controlling other blocks, and 2 is a search keyword and a processing instruction. An input device 3 including a keyboard and the like, an output device 3 including a display for displaying a processing result, and an external storage device 4 such as a hard disk drive (HDD) for storing a search index.

FIG. 2 is a block diagram showing a detailed example of the control device 1 shown in FIG. The controller 1 is the initialization unit 1
1, input unit 12, output unit 13, control unit 14, maximum minimum I
D extraction unit 15, index reading unit 16, search unit 1
7, a control system of the index creation unit 18, a storage system of a search expression storage buffer 21, a maximum / minimum ID storage buffer 22, a character position information storage buffer 23, and words stored in an external storage device or a memory as shown in FIG. Species ID table 3
1, APT index table 3 as shown in FIG.
2, APT index 33 as shown in FIG. 10, FIG.
The index system of APT 34 as shown in FIG. Details of the method of creating indexes, formats, and document retrieval using these indexes are described in Japanese Patent Application No. 6-322068.

In FIG. 2, the initialization section 11 initializes each buffer of the storage system. The input unit 12 inputs information on a search formula from the input device 2. The output unit 13 outputs information such as a list of titles of the relevant document, which is the result of the search, to the display device 3.

The control unit 14 controls the entire control system to comprehensively control each process in the search. The maximum / minimum ID extraction unit 15 uses the input unit 12 to retrieve the search expression storage buffer 21.
Using the word type ID table 31, the search keywords extracted from the search formula stored in
The maximum and minimum ID numbers are extracted and stored in the maximum / minimum ID storage buffer 22.

The index reading unit 16 is a control unit 14.
The information of the APT index 33 of the search target document is obtained from the APT index table 32 based on the number of the search target document given from the above and the maximum and minimum values of the ID numbers obtained from the maximum / minimum ID storage buffer 22. And further A
Addresses for obtaining position information from the minimum ID to the maximum ID are obtained from the PT index 33, and the position information is read from the APT 34.

The search unit 17 determines whether or not the search target document is satisfied with respect to the search formula. First, when it is found from the position information that the search keyword ID string appears consecutively, it is confirmed that the search keyword exists in the search target document, and the search expression of the search target document is added to the search expression according to the logic of the search expression. Judge whether it is established or not.

The index creating unit 18 exists for pre-processing for retrieval, decomposes the input document into character strings of words and characters, and extracts the I of the character strings from the word type ID table 31.
Get the D number. If there is no character string in the word type ID table 31, it is newly registered and a new ID number is acquired. The positions where the character strings of the ID numbers appear in the order of the ID numbers appearing in the document are extracted and registered in the APT 34.
The start address and end address of the position information on 4 are registered in the APT index 33. After registering the position information in the APT 34 for all the ID numbers appearing in the input document, the start address and the end address of the data of the set of the start address and the end address corresponding to the ID numbers stored in the APT index 33 are set as AP.
It is stored in the T index table 32. This series of processes is repeated for the number of search target documents to create a search index.

Next, the operation and the flow of processing of the document retrieval apparatus according to the above embodiment will be described with reference to the flowcharts shown in FIGS. 3 and 4. First, the creation of the index will be described.

In the index preparation which is the pre-processing of the search, the input section 12, the control section 14 and the index preparation section 18
Performs the operation, and the word type ID table 31, the APT index table 32, the APT index 33, the APT
34 various indexes are created.

In the flowchart shown in FIG. 3, the input unit 12 is activated in step 301 to input a document to be searched. In steps 302 to 308 of FIG. 3, the control unit 1
4 operates.

In step 302, the document input from the input unit 11 is divided into sentences using Japanese analysis or the like and converted into a character string (word type) such as a word or a character. In step 303, if there is a word type that has not been registered in the word type ID table 31 among the word types that have appeared in the document, it is newly registered, and an ID number that corresponds uniquely is obtained.

In step 304, the appearance position is obtained from the word with the smallest ID number of the word that appears in the document,
Register with APT34. In step 305, the APT
The start address and end address of the data stored in 34 are stored in the APT index 33.

In step 306, APT 34 and APT
It is determined whether the storage in the index 33 has been performed for all the word types appearing in the input document, and if NO, the control is transferred to step 305. If YES, go to the next process.

In step 307, the APT index table 32 stores the start address and end address of the entire data of the set of start address and end address indicating the position information of the word type appearing in the input document stored in the APT index.

In step 308, it is judged whether or not index creation has been performed for all documents to be searched, and N
In the case of O, the control is transferred to step 301, and in the case of YES, the index creation is completed.

Next, the flow of the search processing is shown in the flowchart of FIG. First, the initialization unit 11 executes step 401.
Start with and initialize each buffer. Next, in step 402, the search expression is input from the input unit 12 and the control unit 1
4, the control unit 14 stores the search expression in the search expression storage buffer 21 as shown in FIG.

In step 403, the maximum / minimum ID extraction unit 15 is activated, a search keyword is extracted from the search expression stored in the search expression storage buffer 21, and an ID number is acquired using the word type ID table 31. The maximum and minimum values of this ID number are stored in the maximum / minimum ID storage buffer 22 as shown in FIG.

At step 404, the control unit 14 is activated,
Steps 405 to 411 until there are no documents to be searched
repeat. In steps 405 to 407, the index reading unit is activated.

Of these, in step 405, the control unit 14
According to the document number received from the APT index table 32
Read from.

In step 406, the data of the APT index 33 corresponding to the document number is read from the read data of the APT index table 32. In step 407, the maximum ID number and the minimum ID are read from the data of the APT index 33 and the maximum / minimum ID storage buffer 22.
APT34 star in which the character position information of the number is stored
get t address and end address, and start with the smallest ID
Data between the t address and the end address of the maximum ID is read from the APT 34 and stored in the character position information storage buffer 23 as shown in FIG.

In steps 408 to 410, the search unit is activated. Of these, in step 408, the search keyword is extracted from the search expression storage buffer 21, and the ID numbers are converted using the word type ID table 31.

In step 409, the position information of the corresponding ID number is read from the character position information storage buffer 23, and the position of the character string of the ID number is the ID of the search keyword.
It is determined that the search keyword exists in the document when the number is the same as the sequence.

The flow of this judgment is shown in FIG. In step 410, it is determined whether or not each search keyword of the search expression is present in the document in step 409 and whether or not the search expression is satisfied with respect to the search target document based on the logical operation between the search keywords. To the control unit 14.

In step 411, the output unit is activated to output according to the search result passed from the control unit 14, and control is passed to step 404. The present invention is not limited to the above-described embodiments, and the following embodiments may be considered, for example.

The difference from the above embodiment is the data of the word type ID table and its arrangement. In the word type ID table in the above embodiment, character strings of words and characters newly obtained at the time of creating the index are sequentially registered. In this registration method, character types such as letters, numbers, symbols, hiragana, katakana, and kanji are mixed.

On the other hand, in the word type ID table in another embodiment of the present invention, only characters are registered as shown in FIG. Furthermore, hiragana, letters, numbers, symbols, and katakana are I-to-I with the letters before indexing.
A D number is given and registered in the word type ID table. The character type of the kanji is registered in the word type ID table when it newly appears when the index is created.

[0044]

As described above in detail, according to the present invention, a search keyword in a document is used in an index in a format in which position information of a character string can be obtained from an external storage device for each ID number associated with the character. In the process of determining the presence / absence of the search keyword, the minimum and maximum ID numbers of the character strings of the search keyword are acquired, so that the I of all characters appearing in the document
It is possible to retrieve less position information from the external storage device than to acquire position information corresponding to the D number, and as a result, it is possible to improve the search speed.

Further, by classifying the ID numbers associated with the characters for each character type and giving consecutive ID numbers to the same character type, it is further necessary when a search keyword often composed of the same character type is input. It is possible to reduce the amount of position information to be displayed, which has an excellent effect of increasing the search speed.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a schematic configuration of an apparatus according to an embodiment of the present invention.

FIG. 2 is a block diagram showing a detailed configuration of a control device 1 in the document search device according to the embodiment.

FIG. 3 is a flowchart showing an outline of an index creating operation in the embodiment.

FIG. 4 is a flowchart showing an outline of a document search operation according to the first embodiment.

FIG. 5 is a search expression storage buffer 21 according to the same embodiment.
FIG.

FIG. 6 is a diagram showing a storage example of a maximum / minimum ID storage buffer 22 in the same embodiment.

FIG. 7 is a diagram showing a storage example of a character position information storage buffer 23 in the same embodiment.

FIG. 8 is a word type ID table 31 in the embodiment.
FIG.

FIG. 9 is a diagram showing a storage example of an APT index table 32 according to the same embodiment.

FIG. 10 is an APT index 3 according to the embodiment.
3 is a diagram showing a storage example of FIG.

FIG. 11 is a diagram showing an example of storage of an APT in the same embodiment.

FIG. 12 is a diagram showing the flow of processing for determining the presence / absence of a search keyword in a document from character position information according to the first embodiment.

FIG. 13 is a diagram showing a storage example of a word type ID table 31 according to another embodiment of the present invention.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Control device, 2 ... Input device, 3 ... Output device, 4 ... External storage device, 11 ... Initialization part, 12 ... Input part, 13 ... Output part, 14 ... Control part, 15 ... Maximum / minimum ID extraction part , 16
... Index reading unit, 17 ... Search unit, 18 ... Index creating unit.

Claims (4)

[Claims] 1. A unique ID number is assigned to each character string such as a character or a word, and I is assigned to a character string appearing in one document.
A method for searching a document based on an index for all documents that stores appearance position information of character strings in the order of D numbers or for each ID number. The minimum ID number and maximum ID number of the character strings that form the search keyword are extracted, and the positional information required for the search is not the positional information of all the character strings, but the positional information between the minimum ID number and the maximum ID number. A document search method characterized by searching by reading from an index. 2. A unique ID number is assigned to each character string by limiting the character string to characters and allocating consecutive characters for each character type that is a classification of symbols / alphanumeric characters, hiragana, katakana, kanji, etc. 2. The document search method according to claim 1, wherein an ID is assigned, and when the keywords input at the time of search are all composed of the same type of characters, the search is possible. 3. A unique ID number is assigned to each character string such as characters and words, and I is assigned to a character string appearing in one document.
A document search device that searches documents at high speed based on an index for all documents that stores appearance position information of character strings in the order of D numbers or for each ID number. Input means for inputting, and maximum / minimum ID extraction means for extracting the minimum ID number and maximum ID number of a character string constituting a plurality of search keywords from the search expression input at the time of search, the minimum ID number and maximum ID of the character string. A document retrieval apparatus comprising: an index reading unit that reads out position information corresponding to the number of IDs between numbers from an index; and an output unit that outputs a processing result. 4. The unique ID number is assigned to each character string by limiting the character string to characters and assigning consecutive character IDs to each character type. Document retrieval device.