JPH03260868A - Information processor - Google Patents

Abstract

PURPOSE:To retrieve a large quantity of data at a high speed to plural retrieving inputs by allocating the record corresponding to a key word to a retrieving file and taking the register information common to the retrieving inputs out of the record corresponding to a key word having the lowest register frequency. CONSTITUTION:A CPU 1 is provided to perform various types of arithmetic processing or decision processing together with a memory 2, a keyboard 4, a display 5, an input/output part 3, an external storage control part 6, and a common bus 8. The register frequency of a key word to be registered is counted for production of a register frequency-key word table. Furthermore a retrieving file is produced to store the registered key words having the register frequencies higher than that of the preceding key word and the register information in pairs. Then the key word having the lowest register frequency is taken out of the key word table, and the register information common to plural inputted retrieving key words is taken out. Thus a large quantity of information can be retrieved at a high speed with use of plural retrieving inputs.

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Application Field] The present invention relates to information retrieval using a plurality of keywords in an information processing device, and is particularly suitable for searching a large amount of information at high speed.

〔overview〕

The present invention provides an information processing apparatus that performs one or more keyword searches on information having a plurality of keywords, performs a comparison, and performs a search process for the information to be sought. For each keyword included in the registered information, all keywords with a registration frequency higher than or equal to the keyword corresponding to the record are stored in the record, and the registered information common to the search input is selected from the records corresponding to the keyword with the lowest registration frequency. By extracting the data, it is possible to quickly search large amounts of data based on multiple search inputs.

[Conventional technology]

2. Description of the Related Art Conventionally, information searches in information processing apparatuses are often performed using a complete transposition method for one or more arbitrary search inputs. This complete transposition method prepares a transposed file for each attribute of a keyword, extracts a record corresponding to the search input from the transposed file that corresponds to the attribute of the search input, and uses information common to the search input from this extracted record. The search process is performed by extracting the .

[Problem to be solved by the invention]

However, searching using the transposed file method requires access time from the transposed file in proportion to the number of search inputs for multiple search inputs, and when searching for large amounts of information, it takes time to retrieve common information between records. Therefore, there was a problem that it was not suitable for high-speed searching of large amounts of information.

This example will be explained.

When constructing a database system for book management in a library, the books stored have multiple keywords such as author, publisher, subject, etc., and when searching, use any keyword from among these keywords. . At this time, if using the transposed file method, the author name,
If three search inputs, publisher and subject, are used, it is necessary to retrieve records corresponding to each of the three search inputs. Therefore, a search time is required in proportion to the number of search inputs, and there is a problem in that the search time becomes long when performing an information search using a plurality of search inputs.

Furthermore, when performing a search process for a large amount of information, the amount of data for each retrieved record becomes enormous, so there is a problem in that it takes time to retrieve common information between records.

The present invention provides high speed when searching with only one record for multiple search inputs, and also because it takes a lot of time to process the first two records in the process of retrieving common information between multiple records. This was done based on the fact that searching can be performed at high speed by extracting common information between the first two records of the search process at the time of registration. An object of the present invention is to provide an information processing device capable of performing search processing and an information search method.

[Means to solve the problem]

The present invention includes a storage device having an area for storing search keywords and a search processing device that executes a search process according to a search input, wherein the search keywords extracted from registered information to be searched are registered. , in an information processing device that performs a search by comparing the registered keyword with a search input, a first means for counting the registration frequency of the registered keyword, and a first means for counting the registration frequency of the registered keyword and the keyword. a second means of creating a matched keyword family, and a registration process that is equal to or higher than the registration frequency of this keyword for each registered keyword by referring to the first means and second means, targeting all keywords in the registered information to which this keyword belongs. The present invention is characterized by comprising a third means for creating a search file consisting of records in which keywords and registered information are stored in pairs.

Further, in the present invention, the search processing device extracts the least frequently registered keyword from the keyword group from among the plurality of input search keywords, and uses this keyword as a directory to search the corresponding record column of the search file. The present invention is characterized in that it includes means for extracting registered information common to the plurality of input search keywords from the record field as a search result.

[Effect]

The information processing apparatus of the present invention has two operations: a keyword registration process and a search process by checking a match with a registered keyword according to a search input.

In the keyword registration process, first, the keywords are extracted from the registered information of the search target, and the keywords that are paired with this registered information are counted and their frequency of registration is calculated. A keyword family is created in which identification codes and registration frequencies are associated with each other.

Next, a search file is created, consisting of a record column created for each keyword to be registered, in which the keyword identification code and registration information are stored in pairs. In this record field, pairs of identification codes and registration information of keywords having a registration frequency higher than or equal to the keywords in this record field are registered.

When using a keyword as a search input to match registered keywords, search for the least frequently registered keyword among the multiple input keywords, and search for records in the search file for this least frequently registered keyword. , extracts pairs of keyword identification codes and registered information related to all search-input keywords stored in that record. By extracting the registered information common to multiple search inputs, the desired search target can be extracted.

In this manner, the present invention enables high-speed searching by searching for a record of a search file that corresponds to the keyword with the least registration frequency in response to multiple search inputs.

〔Example〕

The present invention will be described in detail below with reference to the drawings.

FIG. 1 shows the configuration of an information processing apparatus in an embodiment of the present invention.

The information processing apparatus of this embodiment includes a CPU 1 that performs various arithmetic processing or judgment processing, a human output unit 3 that connects programs such as search processing and keyword registration processing, a memory 2 that stores keywords, etc., a keyboard 4, and a display 5. , an external storage device control unit 6 that connects an external storage device 7 in which various information such as search files are stored, a CPU 1, a memory 2,
A common bus 8 is provided to connect the human output section 3 and the external storage device control section 6.

The processing of the information processing apparatus in this embodiment is divided into two: registration processing related to keyword registration and creation of a search file for use in search processing, and search processing for search input. In this embodiment, a case will be explained in which a search display for browsing an electronic library is performed.

First, the keyword registration and search file creation and registration processing will be explained.

FIG. 2 shows a keyword family created in this keyword registration.

This keyword family is made up of keyword areas as many as the types of keywords to be registered, and each keyword area is made up of a keyword field 9, a registration number field IO, and a registration frequency field 11. The keyword column 9 is a 4-byte field that stores the keyword of the book to be registered, the registration number column 10 is a 4-byte field that stores the registration number indicating the order of keyword registration, and the registration frequency column 11 is a 4-byte field that stores the keyword of the book to be registered. This is a 4-byte field that stores the registration frequency of keywords used for. Here, the keyword is a character string input at the time of registration, and both the registration number and registration frequency are positive integers.

FIG. 3 shows a hash value frequency table, which is composed of a number of hash value frequency columns 12 set by the setter. For example, if the number of hash values is 1000, 1000 fields are secured in the hash value frequency table from the Nα1 hash value frequency column to the NQ and 1000 hash value frequency columns. Each hash value frequency column 12 stores a calculation frequency indicating the number of keywords having the same hash value calculated from a hash function. In other words, the hash value frequency column 12
The calculation frequency also indicates the number of consecutively arranged keywords having the same hash value in the keyword family shown in FIG.

For example, if the calculation frequency in the No. 1 hash value frequency column is "5"
”, if the calculation frequency in the NO12 hash value frequency column is “3”, N11 from the Nα1 keyword of the keyword family
The five keywords up to 5 correspond to the calculation frequency "5" of the calculation frequency in the Nα1 hash value frequency column, and the N of the keyword family
Three keywords from the α6 keyword to the Nα8 keyword correspond to the calculation frequency “3” in the Nα2 hashin value frequency column.

Figure 4 shows the structure of the hash value starting address table.
Hash value frequency column 12 of the hash value frequency table shown in FIG.
It consists of a hash value start address field 13 which is a 4-byte field with the same number as . This hash value starting address column 13 stores the keyword area starting address of the registered keyword stored in the keyword family of FIG. 2. That is, the hash value starting address table corresponds to the keyword calculation frequency table shown in FIG. indicates the first address.

FIG. 5 is a diagram showing the structure of a search file.

This search file is composed of record areas as many as the keyword areas shown in the keyword family, and each record area is composed of a header section 14 and a data section 15.

FIG. 6 shows the structure of records in this search file. The header section 14 is composed of a record number column 16 and an accommodation number column 17, and the record number column 16 is composed of 4
It consists of a byte field and stores the keyword registration number as the record number that identifies the record. The capacity column 17 is composed of a 4-byte field and stores a record capacity indicating the number of keyword areas stored in the data section 15.

The data section 15 is composed of a number of keyword areas indicated by the record capacity in the storage number column 17, and each keyword area is composed of a registration number column 18 and an information number column 19. This registration number column 18 is composed of a 4-byte field and stores the registration number of the keyword. The information number column 19 is composed of a 4-byte field, and stores the information number of the book that is the search target and corresponds to the keyword in the registration number column 18.

FIG. 7 shows the structure of the record capacity table, which consists of the same number of record capacity columns 20 as the keyword areas of the keyword group shown in FIG.
The record capacity column is also associated in ascending order starting with "1". The record capacity column 20 consists of a 4-byte field that stores a numerical value indicating the capacity of the data section of each record area of the search file shown in FIG.

FIG. 8 is a record start address table showing the storage address of each record, which is composed of n record start address columns 21, the same number as the record capacity table, and corresponds to each record capacity column 20. This record starting address column 21 is
It stores the record start address of each record stored in the above search file, and corresponds to the record capacity column 20 of the record capacity table in FIG. 7. For example, the kl record start address is Nα1, and the record corresponding to the code capacity column This indicates the starting address in the search file.

Next, the flowchart of the keyword registration processing operation is shown in Section 9.
This will be explained with reference to the diagram.

The operations of this keyword registration process can be broadly divided into categorizing keywords and counting their frequency of registration, and creating a search file by storing pairs of registration numbers and information numbers in records for each keyword in the search file. The first step is to create a hash head address table based on the hash value of the keyword, the second step is to create a keyword family, the third step is to create the start address of the search file, and the fourth step is to create a hash head address table using the hash value of the keyword. The step consists of the operation of creating a search file.

First, in the first step, as pre-processing for creating a keyword table, keywords are classified by hash value, and the number of keywords classified by hash value and the starting address of a group of keywords classified by hash value in the keyword table are calculated.

This operation will be explained.

First, extract bibliographic information from the registered file 3101
), all keywords are extracted from the bibliographic information. The hash values of these keywords are calculated using a hash function, and "1" is added to the corresponding hash value frequency column 12 of the hash value frequency table in FIG. 3 (S102). The same process is performed for all registered bibliographic information.

Next, create a hash head address table. First, "0" is stored in the Nαl hash value starting address field (S104>). Then, the process moves to the Nα2 hash value starting address field (3105).
, the value obtained by adding the calculation frequency in the NCL 1 hash value frequency field of the hash value frequency table by 3 and the NIIL 1 hash value leading number "0" is stored as the leading address in the Nα2 hash value leading address field (3106). . It is checked whether the first address has been stored in the same number of hashish starting address fields as the halash value frequency field of the hashish value frequency table (3107), and similarly, the calculated frequency of the NcL2 hashish value frequency field of the hashish value frequency table is tripled. The value obtained by adding the numerical value in the Nα2 hash value starting address field to the &3 hash value starting address field is stored as the starting address in the &3 hash value starting address field. This is to store (NcL(n-1) base value frequency field) x 3+(Nα(n-1) base value first address field) at the No. n hash value first address. Similar processing is performed for all the hash start address fields in the hash start address table.

The second step is to create a keyword family.

After setting the registration number to "1", extract the bibliographic information from the registration file (3110.111), extract the keyword from the bibliographic information, and calculate the hash value of this keyword using the hash function (S112). , the leading address of the hashish leading address column corresponding to this hash value is extracted from the hashish leading address table (sl13).

In the keyword column of the empty area after the keyword area of the keyword family corresponding to this first address, put the keyword extracted from the bibliographic information in the registration number column, and put the current value of the registration number that will be the registration number of this keyword in the registration frequency column. , the calculated frequency in the corresponding hash value frequency column of the hash value frequency table is stored (S114). Then, before processing the next keyword, "l" is added to the registration number (S115). Similar processing is performed for all keywords in the extracted bibliographic information. Similar processing is performed for all bibliographic information.

In the third step, as pre-processing for creating a search file, the capacity of each record in the search file and the starting address in the search file are calculated.

First, create a record capacity table. Bibliographic information for one book is extracted from the registered file (3118), a keyword is extracted from the bibliographic information, and a hash value of this keyword is calculated using a hash function (S119). From the hash head address table, extract the head address of the hash head head address field that corresponds to this hash value (3120);
Extracts the number of keyword areas indicated by the calculation frequency in the hash value frequency field corresponding to this first address (5121, 12
2). From the extracted keyword areas, extract the registration number and registration frequency of keyword areas that store keywords that match the keyword being registered (S1
23). Similar processing is performed on the remaining keywords of the extracted bibliographic information.

Among these retrieved registration number and registration frequency pairs, store the number of retrieved registration number and registration frequency pairs in the record capacity column of the record capacity table that corresponds to the registration number with the lowest registration frequency ( S 125).

From these pairs of registration numbers and registration frequencies, delete the pair with the lowest registration frequency (S126), and then delete the record capacity table corresponding to the registration number with the lowest registration frequency among the remaining pairs of registration numbers and registration frequencies. Store the number of pairs of registration number and registration frequency in the record capacity field, and delete the pair of minimum registration frequency (
S 127). Similar processing is repeated until there are no more pairs of registration numbers and registration frequencies.

Next, create a record start address table.

First, store “0” in the Nα code first address field (
S 130). Add "1" as the capacity of the header section of the record shown in Figure 6 to the capacity in the kl record capacity column of the record capacity table, and add "O" in the NaI start address column to the value doubled. Then, with this value as the starting address, N(
Store it in the first address column of the L2 record. Furthermore, the value obtained by adding "1" to the capacity in the Nα2 record capacity field of the record capacity table and the value in the Nα2 record start address field is set as the start address, and the value of the Nα3 record start address is set as the start address. Store in address.

This process is performed at the first address of the No, n hash value (NCL(
Tl-1) Record capacity column +1) x2+ (N(L (
n-1) Record start address field) is stored (S132). Similar processing is performed for all record capacity columns of the record capacity table (S133-135).

In the fourth step, a search file is created.

Extract bibliographic information for one book from the registered file S 1
40) A keyword is extracted from the bibliographic information, and a hash value of this keyword is calculated using a hash function (S141). Then, from the hashish starting address table, extract the starting address of the hashish starting address field that corresponds to this hashish value. The keyword area indicated by the calculated frequency in the value frequency column is extracted (S143 and 144).

From the retrieved keyword area, the registration number and registration frequency storing the keyword that matches the keyword of the retrieved bibliographic information is retrieved (S145). Similar processing is performed on the remaining keywords of the extracted bibliographic information.

Among these retrieved pairs of registration numbers and registration frequencies,
Store the registration number of the minimum registration frequency in the record number column of the search file indicated by the start address of the record start address column of the record start address table corresponding to the registration number of the minimum registration frequency,
Add the number of pairs of registration numbers and registration frequencies that are currently being registered in the capacity column, and store all the registration numbers and registration frequencies that are being retrieved and registered in the free area of the data section (S
147.148〉.

Furthermore, from among these pairs of registration numbers and registration frequencies, the pair with the lowest registration frequency is deleted (S149), and among the remaining pairs of registration numbers and registration frequencies, the first address of the record corresponding to the registration number with the lowest registration frequency is deleted. Store the registration address with the minimum registration frequency in the record address of the search file that corresponds to the start address in the record start address column of the table, add the number of pairs of registration number and registration frequency remaining in the storage capacity column, and Store the remaining registration number and registration frequency in the free space, and delete the pair with the lowest registration frequency. Similar processing is performed until there are no more pairs of registration numbers and registration frequencies (3150, 147-149). Similar processing is performed for all registered bibliographic information of the registered file.

FIG. 1O is a flow diagram illustrating the search operation.

First, a keyword is extracted from the search input, and the hash value of this keyword is calculated using a hash function (S
201). The first address of the hashish starting address column corresponding to this hashish value is extracted from the hashish starting address table (S 202), and the calculation frequency corresponding to the hashish value calculated from the hashish value frequency table is extracted (S 203>).

Then, the keyword area indicated by the calculated frequency extracted from the hash value frequency table is extracted from the keyword area of the keyword family corresponding to the first address (S204). The registration number and registration frequency of the keyword area in which the keyword field of the retrieved keyword area stores a keyword that matches the keyword under search processing retrieved from the search input is retrieved (S205). Similar processing is performed on the remaining keywords entered in the search, and the registration number and registration frequency of the keywords corresponding to the search input are narrowed down.

Among these narrowed-down pairs of registration numbers and registration frequencies, extract the first address in the record start address column of the record start address table that corresponds to the minimum registration frequency and the capacity indicated by the record capacity that corresponds to the registration number with the minimum registration frequency. , extracts the record of the search file indicated by this start address and record capacity (3207-209).

Then, for the data section 16 of this record, an information number common to all narrowed-down registration numbers is extracted and set as a search target (3210).

For example, in the record in Figure 6, if the registration numbers corresponding to the keywords entered for search are "11J", "50", and "88", the information number common to these is rloOJ, which becomes the information number of the searched book. .

This will be explained by giving a more specific example.

For example, the bibliographic number r registered in the registration file of Toshoyari
Assume that 100 J has rISDNJ, "communication", and rO3IJ as keywords. Looking at the registration numbers and frequency of registration for these three keywords, r I S DNJ has a registration number of "50" and frequency of registration is rloo J, and "communication" has a registration number of "11".
Assume that the registration frequency is r350 J, the registration number of "03IJ" is "88", and the registration frequency is "50".

Since the registration frequency is in the relationship "rO3I"<rlsDN"< r communication", the record of rO3Ij in the search file contains rO3IJ, "l5DN", and "communication" with its registration number and information number (bibliographic number r100 J). The rlsDNJ record contains "l5DN" and "
"Communication" is stored together with its registration number and information number, and in the "Communication" record, "Communication" is stored together with its registration number and information number to create a search file.

When performing a search process, enter "O3■", r as a search input.
When the keywords ``lsDN'' and ``communication'' are input, the registration frequency of these three keywords is checked, and the record of rO3IJ with the lowest registration frequency is retrieved. This rO3IJ
From the data section of the record, “○SIJ, rlsDNJ
, three registration numbers for "communications": "88", "50", "1"
1'' is extracted, and the bibliographic information, that is, the book, with the common information number r100J is extracted as a search result.

In this way, it is possible to search for a plurality of search inputs simply by searching for the search record corresponding to the keyword that is least frequently registered, and the search can be speeded up.

〔Effect of the invention〕

As explained above, the present invention enables searching for multiple search inputs using only the Lou code, and extracts common information between the first two records of the search process at the time of registration. This has the excellent effect of speeding up the search process when searching for multiple search inputs.

[Brief explanation of drawings]

FIG. 1 shows an example of the configuration of the information processing apparatus according to the embodiment. FIG. 2 is a keyword table of the example. FIG. 3 is a hash value frequency table of the example. FIG. 4 is an example hash head address table. FIG. 5 is an example search file. FIG. 6 is a diagram showing the record structure of the search file. FIG. 7 is a record capacity table of the embodiment. FIG. 8 is a record start address table of the embodiment. FIG. 9 is a diagram illustrating the keyword registration processing operation of the embodiment. FIG. 10 is a diagram illustrating the search processing operation of the embodiment. 1...CPU, 2...Memory, 3...Person output section,
4...Keyboard, 5...Display, 6...
External storage device control unit, 7...external storage device, 7...common bus.

Claims (1)

[Claims] 1. A storage device including a storage device having an area for storing search keywords and a search processing device that executes search processing according to a search input; In an information processing device in which a keyword is registered and a search is performed by comparing the registered keyword with a search input, a first means for counting the registration frequency of the registered keyword; and a first means for counting the registration frequency of the registered keyword. a second means of creating a keyword table in which keywords are associated with keywords; An information processing apparatus comprising: a third means for creating a search file consisting of records in which registered keywords and registered information having a registration frequency or higher are stored in pairs. 2. Equipped with a storage device having an area for storing search keywords and a search processing device that executes search processing according to the search input, the search keywords extracted from the registered information to be searched are registered, and this In an information processing device that performs a search by comparing registered keywords and search input, a first means for counting the registration frequency of registered keywords and a correspondence between the registration frequency of the registered keywords and the keywords are provided. a second means of creating a keyword table with reference to the first means and second means, and for each registered keyword, for all keywords in the registered information to which this keyword belongs, registered keywords with a registration frequency equal to or higher than that of this keyword; and a third means for creating a search file consisting of records stored in pairs with registered information. A means is provided for extracting registered information common to the plurality of search keywords inputted above from the above-mentioned keyword table, searching for a corresponding record in the above-mentioned search file using this keyword as a directory, and extracting registered information common to the plurality of search keywords inputted above from this record as a search result. An information processing device characterized by: