JPH0574859B2 - - Google Patents

Description

[Detailed Description of the Invention] Industrial Application Field The present invention relates to an information retrieval device composed of a data processing device, an internal storage device, and an external storage device. After being stored as a search index in a storage device, it is used by an information search device that performs search processing.

BACKGROUND TECHNOLOGY Information retrieval devices used in OA systems, etc. have internal memory retrieval, in which all search information in an external storage device is stored in the internal storage device, and then search processing is performed, and search information in an external storage device. There is an external memory search, which is configured to have a structure suitable for searching, and retrieves index information and search information needed each time a search is performed from an external storage device to an internal storage device.

Conventionally, most information retrieval devices used external memory retrieval, but: 1. With the rapid decline in the price of semiconductor memories, information retrieval devices with large capacity internal storage devices can be manufactured at low cost.

2. OA systems are centered around conversational processing, and the speed of system response time for information retrieval devices has become important.

For these reasons, internal memory retrieval is also often used.

Hereinafter, only information retrieval devices using internal memory retrieval will be described. In the case of an information retrieval device that uses an internal storage retrieval method, the following two processes are problematic.

1 Before starting to use the information retrieval device (for example, at startup), the process of transferring the search information stored in the external storage device to the internal storage device so that search processing can be performed (hereinafter referred to as internal storage device) It is necessary to reconstruct the search information (search index) within the search engine (or simply referred to as reconstruction processing).

2. When updating search information, such as adding a new search information record to the information search device, at the same time as updating the search information in the internal storage device, an external storage device is used to retain the updated content for a long time. It is necessary to perform a process (hereinafter referred to as a search information update process) of storing information corresponding to the update contents in the internal storage device.

The processing efficiency of an information retrieval device using internal memory retrieval is determined by the difference in the processing method for these two points. These two processing methods can be broadly divided into the following three methods, depending on the storage format of search information in internal storage and external storage.

1. When both the search information in the external storage device and the search information in the internal storage device are stored for each search information record, for example, in the primary normal form as shown in FIG. 2. The search information in the external storage device is Each search information record is stored in a primary normal form as shown in Figure 3, for example, and the search information in the internal storage is stored in a search index format suitable for searching, for example in a reverse file format as shown in Figure 4. If

3. When both the search information in the external storage device and the search information in the internal storage device are stored in a search index format suitable for searching, for example in a reverse file format as shown in FIG.

In order to judge the advantages and disadvantages of each method, it is necessary to evaluate the following three processing speeds.

A Reconstruction processing speed of search information in the internal storage device B Search processing speed C Update processing speed of search information Next, the advantages and disadvantages of each of the above methods will be considered regarding these three points.

A. In methods 1 and 3, the process of reconfiguring the search information in the internal storage device, which is performed when the system starts up immediately before starting to use the information retrieval device, simply reconstructs the search information in the search information format in the external storage device. Since the search information is stored in the internal storage device, the process is completed by simply transferring the search information from the external storage device to the internal storage device. Therefore, rapid reconfiguration processing can be expected.

In the second method, the search information in the external storage device is retrieved one record at a time, and the search information is reconstructed in the same way as the normal update process of the search information in the internal storage device. In this reconfiguration process, the search index in the internal storage device is updated for all registered search information, so
It takes more processing time than methods 1 and 3.

B Comparing the search processing speed of the search information in the internal storage after reconfiguring the search information in the internal storage, method 1 also stores the search information in the internal storage in units of records (for example, as shown in Figure 3). (linear normal form). Therefore, when searching for a certain keyword, it is necessary to scan (examine) the entire search information to see if the corresponding keyword matches all the keyword groups with the same attribute as the attribute to which the keyword belongs, and the search processing time is It takes. (There are many duplicate keywords stored in the search information stored in the internal storage device, and the storage efficiency is poor.) In the case of methods 2 and 3, in the case of the internal storage device, the search information has a logical structure suitable for searching, For example, since the search information is stored in the internal storage device in a reverse file format as shown in FIG. 4, the search process can be completed by scanning the minimum number of keywords, so a faster search process can be expected compared to 1.

C Comparing the update processing speed of search information in the internal storage device, methods 1 and 2 update the search information in the internal storage device and then transfer only the search information record for which an update request was made to the external storage device. The process ends by recording to the storage device. Therefore, high-speed update processing can be expected.

In case 3, it is necessary to reflect the updated contents in the internal storage device as they are in the external storage device.
In general, the search index in the internal storage device is changed over a wide range of areas through update processing. Therefore, in the process of reflecting all the updated parts in the external storage device, it becomes necessary to transfer some of the scattered search information to the external storage device.
The update processing speed is lower than that of the above method.

As mentioned above, each of the above three methods has its advantages and disadvantages, but the problem of update processing time, which is a drawback of method 3, is considered to be less important than the other two points, so external storage In the case of a rewritable medium such as a magnetic storage medium, method 3 (actually a variation of method 3) is used.

However, if the external storage device is a rewritable medium such as a magnetic storage medium, method 3 can be applied, but if the external storage device is a write-once medium such as an optical disk device, rewriting is not possible, so method 3 is applied. It was difficult to use.

Problems to be Solved by the Invention As described above, the conventional information retrieval device using internal memory retrieval has the following problems.

1. When method 1 is used, the search processing time and the storage efficiency of search information in the internal storage device are poor.

2 When method 2 is used, the processing speed for reconstructing search information in the internal storage device is slow.

3. When method 3 is used, the update processing speed is slow.

4 If a write-once type medium is used as the external storage device, method 3 cannot be used.

The present invention has been made in view of the above points, and employs the second method in order to overcome the above-mentioned drawbacks, and also has the problem of the reconfiguration processing speed of search information in the internal storage device, which is a drawback of the second method. The purpose is to provide an improved method.

Means for Solving the Problems In order to solve the above-mentioned drawbacks, the present invention performs the following processing.

When updating the search information of the information search device, the search information in the internal storage device (hereinafter referred to as a search index to distinguish it from the search information in the external storage device) and the search information in the external storage device are updated at the same time. Needs to be updated. At this time, when the search index in the internal storage device is updated, the experience of the update process is used to provide information (for example, update method, location and storage method of the search information, location and content of changed information, etc.), and stores this auxiliary information in record units together with the search information in an external storage device. Next, when reconfiguring the search information in the internal storage device, where processing speed is an issue, records containing the search information are retrieved from the external storage device one update at a time, and the search information in the record is By using auxiliary information to update the search index in the internal storage device, it is possible to perform an update process that eliminates the scanning of keywords in the search index, which is essential during normal update processing. Reconfiguration processing becomes possible by repeating. Therefore, faster reconfiguration processing can be achieved than conventional reconfiguration processing using normal update processing.

Effect The present invention was created using the experience of normal update processing when reconfiguring the search information stored in the external storage device as a search index in the internal storage device with the above configuration. By using auxiliary information (information on the storage method and storage location in the internal storage device), the process of reconfiguring the search index in the internal storage device can be performed using a process that is simpler than the normal update process. It is an information retrieval device that can achieve both high-speed reconstruction processing and advanced search functions, which have been difficult in the past.

Embodiment FIG. 1 is a block diagram showing an embodiment of the information retrieval apparatus of the present invention. In FIG. 1, 102 is an external storage device, 103 is an internal storage device, 101 is an information format conversion function that retrieves search information in the external storage device for each record and reconstructs a search index in the internal storage device, and 104 is an information format conversion function. an information retrieval device that performs a search process using a search index stored in the internal storage device 103; 105 an information retrieval mechanism that retrieves search information and auxiliary information from the external storage device;
Reference numeral 6 denotes an auxiliary information extraction mechanism that extracts a portion of auxiliary information from the retrieved information; 107 an information storage position determination mechanism that determines a location to store search information using the retrieved auxiliary information; and 108, an information storage position determination mechanism that determines This is an information storage mechanism that stores information at a specified location and updates the search index in the internal storage device using a specified method. Further, 110 and 111 are information retrieved from the external storage device 102 (records storing search information and auxiliary information), 112 is information divided into search information and auxiliary information, and 113, 1
Reference numeral 14 indicates information indicating the search information, the storage location of the search information in the search index, the update method, etc., and 115 indicates the result of searching the search index in the internal storage device 103.

FIG. 2 shows the information created by updating the search information in the information search device of the present invention.
2 is a diagram illustrating an example of a record format for storing search information and auxiliary information stored in the external storage device 102, in which a is an example in which a search information area and an auxiliary information area are separated, and b is an example in which a search information area and an auxiliary information area are separated; FIG. This is an example in which information is stored in pairs with corresponding auxiliary information.

FIG. 3 shows an example in which information (search information and auxiliary information) is actually stored in an external storage device using the record format a in FIG. 2, and FIG. FIG. 3 is a diagram showing an example of a logical structure when a portion of the search information 1 is converted into a reverse file structure and developed in an internal storage device.

In FIG. 3, the search information is configured in a primary normal form, so if it is stored as a search index in the internal storage device in this format, the search efficiency and storage area usage efficiency will be poor. Therefore, in the information retrieval device of the present invention, the retrieval index is stored in the internal storage device after being reconfigured into a reverse file structure as shown in FIG. 4, for example. This embodiment describes only an example in which search information stored in an external storage device in primary normal form is converted into a reverse file structure, reorganized, and stored in an internal storage device, but it can also be used for index sequential organization, etc. An embodiment in which the information is reconfigured into a logical structure suitable for another search and then stored in the internal storage device is also possible.

FIG. 5 is a block diagram showing a different embodiment of the information retrieval device of the present invention. The upper half of Figure 5 is the first
The search information update mechanism 5, which has the same configuration as the embodiment shown in the figure, and which performs processing when updating search information, is shown in FIG.
This is a configuration in which 05 is added. Search information update mechanism 50
5 includes a search information change mechanism 513 that receives update information 530 sent to the search device and updates the search information in the internal storage device. The information 531 is retrieved and the search index in the internal storage device 503 is updated using the specified update method, and the result is comprised of information such as the update method and update location. receiving information 532 for generating auxiliary information; information 532 for generating this auxiliary information;
and the search information in the update information to the auxiliary information generation mechanism 5.
Send to 12th. The auxiliary information generation mechanism 512 generates auxiliary information from the sent information and retransfers it to the storage information editing mechanism 511 together with the sent search information. The storage information editing mechanism 511 edits the sent search information and auxiliary information into information in the format shown in FIG.
send to The information storage mechanism 510 stores the transferred information in the external storage device 502.

FIG. 6 is a block diagram when the information retrieval apparatus of the present invention is applied to a document file system using an optical disk device as an external storage device. 601 is a control device that controls the entire system;
Information conversion mechanism 501 and search information update mechanism 5 shown in the figure
05 is also implemented in this control device. 602
is an external storage device using a write-once optical disk device, and stores document information and search information input into the system. 603 is an internal storage device using a semiconductor memory, 605 is an image input/output device for inputting and outputting documents, and 604 is an image display device for displaying documents and search information.

FIG. 7 shows one embodiment of the format of auxiliary information, and FIG. 8 shows another embodiment of the format of auxiliary information. The example shown in Fig. 7 is an example in which five types of update processing are expressed in one type of format, and the case in Fig. 8 is an example in which three types of update processing are expressed in separate formats. It is. Although the format itself is different,
Both are auxiliary information for similar processing.

FIG. 9 to FIG. 20 show the update process of the search index in the internal storage device 503 for the update process 530 performed by the search information update mechanism 505 of FIG. Processing performed within the information conversion mechanism 501 to reconstruct a search index in the internal storage device 503 using each search information and auxiliary information created by the search information update mechanism 505 and stored in the external storage device 502; It is a figure for comparing and explaining.

Based on these explanations, a search information record group containing search information and auxiliary information in the format shown in FIG. 2 is actually stored in the external storage device 502 in the format shown in FIG. One method of implementing the process of reorganizing the search index into the reverse file structure as shown can be understood.

Here, before explaining the actual update process, auxiliary information creation process, and reconfiguration process using Figures 9 to 20, we would like to explain some matters that are assumed in the explanation of this figure. explain. The search index in the internal storage device 503 has a logical structure of a reverse file structure as shown in FIG.
``Database'' by Martin, translated by Yoshihisa Kunitomo et al., chapter ``Reverse File System'' from page 613).

In the example described here, the third
Search information as shown in the figure is stored, and this search information is reconstructed into a reverse file format search index for each attribute in the internal storage device. Keyword searching for this information retrieval device uses a search index in the internal storage device to retrieve a group of record numbers in which the search target keyword is stored, and then retrieves the actual search information record from the external storage device using the record number group as a key. It is to take it out.

The logical structure of the reverse file format search index for each attribute consists of a keyword area where keywords are stored in a logically sorted order, as shown in Figure 4, and a keyword area for referencing records that store each keyword. There is a record number area that stores a group of record numbers, and an index area that stores indexes of keywords that are logically sorted and lined up.

In the keyword area in the reverse file format search index for each attribute, the keywords are logically sorted, but physically they are not consecutive in the sorted order. The logical order is constructed using pointers between keyword cells that store individual keywords (blocks that store individual keywords and associated information).

In addition, the record number area in the search index includes a record number cell (a block storing a plurality of record numbers and auxiliary information) that stores a group of record numbers for each keyword. The logical structure of this search index is different from its physical structure, which is its actual arrangement in memory. For this reason, various storage methods can be considered to realize this logical structure, but here, as one example, a case will be described in which the physical structure is as follows.

Each keyword cell contains, in addition to the keyword itself, a pointer to a keyword cell that stores the logical next ranking keyword for that keyword.
A pointer to the record number cell referenced by the relevant keyword and the total number of record numbers referenced are stored. Furthermore, the record number cell stores an area for storing a plurality of record numbers and a pointer for connecting the record number cells in an order that follows the logical order of keywords. To explain how to manage the memory area in the internal storage device in this example, in order to simplify the memory management of the keyword area and record number area, the keyword area and record number area may be separated in large segment units. However, it is basically stored in a contiguous area. In addition, since the free area in the keyword area is managed from the beginning of the area, it can be managed if there is a pointer (empty keyword area pointer) pointing to the next area of the last area currently in use. Management of free areas in the record number area takes into account reuse of the area, so each free record number cell in the record number area is connected with a pointer, and a pointer pointing to the first free record number cell ( Free record number area pointer).

In summary, each keyword cell is linked by pointers in a logical order. The storage position of a new keyword cell is managed by an empty keyword area pointer. The record number cells store a plurality of record numbers grouped by keyword, and are linked by pointers according to the logical order of the keywords. The storage position of a new record number cell is managed by an empty record number area pointer.

Taking the above preconditions into consideration, details of the update process and the reconfiguration process of the search index in the internal storage device will be explained below using the figures from FIG. 9 to FIG. 20. Each figure shows how the state within the internal storage device changes due to the re-new process and the reconfiguration process. Here, to explain how to read each figure, there are three rows of horizontal pointer columns with each figure, and the upper column is each keyword arranged in a logical order using pointers within the keyword area. The middle column shows each record number cell arranged according to a logical order using a pointer within the record number area, and the bottom column shows a column for managing free space in the record number area. , shows an empty record number area list created by connecting empty record number cells, and the empty box between the top and middle columns represents the beginning of the empty keyword area. Next, each of FIGS. 9 to 20 is a set of two pages, with odd-numbered pages representing the search index state before the update process and even-numbered pages showing the search index state after the update process. The numbers in parentheses (for example, {1}) in even-numbered pages indicate the processing order in the case of normal update processing performed by the search information update mechanism in FIG. The processing order in the case of reconstruction processing using auxiliary information performed by the information format conversion mechanism 501 shown in the figure is shown. Furthermore, the reconfiguration process shown here is an example of the reconfiguration process using the present invention, and in the case of the conventional reconfiguration process, the same process as the update process is performed. The difference in processing efficiency of the constituent types when using auxiliary information becomes the effect of the information retrieval device of the present invention.

Hereinafter, registration processing, deletion processing, and change processing will be explained in this order according to the notation shown in the above diagram.

The basic update processing method is that the search information update mechanism 505 extracts the update target keyword for each attribute from the update information 530 and updates the search index for each attribute in the internal storage device 503. Process. Similarly, as a method of reconfiguration processing, the information format conversion mechanism 501 extracts the keywords to be updated for each attribute and the corresponding auxiliary information from the information 520 extracted from the external storage device 502, and The search index for each attribute in 503 is reconstructed.

Showing the order of update processing, it is necessary to select the update processing according to the content of the update processing as described below.

1. Scan the keyword cell list in the keyword area to find out where in the search information the retrieved keyword to be updated is stored. (You can narrow down the scanning target by using the index screen. Check the presence or absence of the keyword.)

2. When a corresponding keyword is found, various update processes (FIGS. 11 to 20) are performed on the keyword (update process on registered keywords).

3. If a keyword is not found, and if the update process is a registration process, a new keyword registration process (Figures 9 and 10) is performed (new keyword registration process).

4 For any other processing, perform abnormal processing.

In the case of reconfiguration processing, each processing method is stored in the auxiliary information, and the type of update processing that needs to be performed is known, so the above processing is unnecessary.

First, the keyword registration process will be explained using FIGS. 9 to 14. In the case of normal registration processing, the keyword area is first searched, and based on the results, it is determined whether the keyword exists or not, and if the keyword exists, the record number managed by the keyword is numbered. There are three processing steps, depending on whether there is a free space in the record number cell in the area or not.

First, an example of the registration process will be explained by comparing the update process performed by the search information update mechanism and the update process performed by the information format conversion mechanism.

Normal registration processing is performed in the following order.

1. Receive the keyword and record number to be registered as an update instruction from the outside.

2. Scan the index area and keyword area shown in Figure 4 with the retrieved keyword to find the location where the corresponding keyword is stored.

3 If the keyword cannot be found, please refer to the 9th page.
The new keyword registration process shown in FIGS. 1 and 10 is performed.

4. If found, trace the pointer to the record number area storing the record number that uses the keyword, and check whether there is space in the record number area for the keyword.

5 If there is space in the record number area, the first
The record number additional registration process 1 shown in FIGS. 1 and 12 is performed.

6. If there is no space, perform record number addition process 2 shown in FIGS. 13 and 14.

Next, when the same processing is performed by reconstruction processing using auxiliary information, the process is as follows.

1' Extract the record number and keyword to be registered, and the auxiliary information corresponding to the keyword from the search information in the external storage device, record by record, in the order of registration.

2′ Check the type of auxiliary information for the retrieved keyword and determine the type of processing (new keyword registration,
Either record number addition process 1 or record number addition process 2) is determined.

3' Pass the keyword, auxiliary information, and record number to each process and perform the process. Even if we only look at the common parts of the above registration processing, we can see the difference in the amount of work between the update processing and the reconfiguration processing. The process of scanning the keyword area in step 2 takes time.

Next, specific methods of each registration process (new keyword registration, record number addition process 1, record number addition process 2) will be described using diagrams.

In the case of new keyword registration processing (reconstruction processing) (FIGS. 9 and 10), the processing is performed in the following order (the numbers in brackets representing the processing order correspond to those in FIG. 10).

{0} Perform the process of 1.2. {1} Store the relevant keyword in the first cell of the empty keyword cells indicated by the empty keyword area pointer.

{2} Store the position of the keyword (keyword immediately before the position where the keyword is logically inserted) cell immediately before the position where the process in 2. failed as position information 1 (this process is used to create auxiliary information special processing).

{3} Follow the pointer to the record number of the immediately preceding keyword and store the position of the last cell of the record number managed by the immediately preceding keyword cell as position information 2 (this process is a special process for creating auxiliary information).

{4} Set the information of the next keyword cell pointer in the previous keyword cell as the content of the next keyword cell pointer in the registered keyword cell.

{5} Change the information of the next keyword cell pointer of the immediately previous keyword cell to the position information of the registered keyword cell.

{6} Update the number of registered keywords (increase by 1).

{7} Store the record number to be registered at the beginning position of the empty record number cell indicated by the empty record number area pointer.

{8} Set the contents of the free record number area pointer to the record number cell pointer in the registered keyword cell.

{9} Replace the contents of the next record number cell pointer of the record number cell in which the record number has been registered with the contents of the empty record number area pointer.

{10} Replace the contents of the next record number cell pointer of the record number cell to which position information 2 has been added with the contents of the next record number cell pointer of the record number cell in which the record number has been registered.

{11} Change the contents of the next record number cell pointer of the record number cell to which position information 2 has been added to pointer information indicating the position of the record number cell in which the record number has been registered.

{12} Set the auxiliary information type item to 0. (This process is a special process for creating auxiliary information) {13} Update the number of registered record numbers (increase by 1).

The registration process (update process) for new keywords is as follows:
In addition to the common part processing in 1.2., the above processing is required, and in particular, the processing in 2. takes time because it scans the keyword area. By the way, among the processes from {1} to {13} explained here, {2}
The processes {3} and {12} are special processes for creating auxiliary information, and are not necessary if the process for creating auxiliary information is not performed. The new keyword registration process (reconstruction process) is performed in the following order (the numbers in brackets representing the process order correspond to those in FIG. 10).

(0) Process common parts 1' and 2'.

(1) Store the relevant keyword at the beginning of the empty keyword cell indicated by the empty keyword area pointer.

(2) Set the information of the next keyword pointer in the keyword cell indicated by auxiliary information 1 to the content of the next keyword pointer information of the keyword cell in which the keyword is registered.

(3) Set the position information of the keyword cell in which the keyword is registered to the contents of the next keyword pointer in the keyword cell indicated by auxiliary information 1.

(4) Store the record number to be registered at the beginning of the empty record number cell indicated by the empty record number area pointer.

(5) Set the information of the next record number cell pointer in the record number cell in which the record number has been registered to the contents of the empty record number area pointer.

(6) Set the information of the next record number cell pointer in the record number cell indicated by position information 2 to the contents of the next record number cell pointer in the record number cell in which the record number is registered.

(7) Pointer information indicating the position of the record number cell in which the record number has been registered is set to the contents of the next record number cell pointer in the record number cell indicated by position information 2.

(8) Pointer information indicating the position of the record number cell in which the record number is registered is set to the contents of the pointer to the record number in the keyword cell in which the keyword is registered.

(9) Update the number of registered keywords (increase by 1).

(10) Update the number of registered record numbers (1
increase).

As mentioned above, there is no need to perform step 2 (keyword scanning process), which is a time-consuming process, and the process can be easily completed by simply changing the pointer information of the cell indicated by the auxiliary information and the newly registered cell. I understand. As mentioned above, in the case of new keyword registration processing, when comparing normal update processing and reconstruction processing using auxiliary information, it can be seen that there is a considerable difference in the amount of processing required as described above. . For this reason, if normal update processing is used instead of reconstruction processing using auxiliary information to reconstruct the search index in the internal storage device, which is necessary at system startup, etc., for each record number (for example, Since the medium applied to the document file system shown in Figure 6 can store approximately 10,000 to 20,000 documents on an optical disk, it is necessary to assume that registration processing will be performed this many times. Since it is necessary to perform update processing, it takes a very long time and is not practical.
Therefore, high-speed reconstruction processing can be achieved by replacing normal update processing with reconstruction processing that uses this auxiliary information.

Next, in the case of record number additional registration process 1 (update process), as shown in FIGS. 11 and 12, it is performed in the following order. (The numbers in brackets representing the processing order correspond to those in FIG. 12).

{0} Performs common processing, 1. and 2.

{1} The location of the cell storing the keyword to be registered, found as a result of the process in step 2, is stored as location information 1 (this process is a special process for creating auxiliary information).

{2} Perform the process in step 4 to find the location in the record number cell where a new record number is to be registered.

{3} The location where the record number to be registered, found as a result of the process in {2}, is stored is stored as location information 2. (This process is a special process for creating auxiliary information) {4} Store the record number to be registered at the position indicated by position information 2.

{5} Update the number of registered record numbers (increase by 1).

{6} Set the auxiliary information type item to 1 (this process is a special process for creating auxiliary information).

In this process, processing time is consumed by scanning the registered keyword position in the keyword area at {0} and finding a free area in the record number cell at {2}.

The same registration process will be described in the case of reconfiguration process using auxiliary information. (The bracketed numbers representing the processing order correspond to those in FIG. 12).

(0) Perform common processing 1' and 2'.

(1) Store the record number to be registered at the location indicated by location information 2.

(2) Update the number of registered record numbers in the keyword area indicated by location information 1 (increase by one).

It can be seen that the process can be completed in two steps, and is extremely fast.

Furthermore, in the case of record number additional registration process 2 (update process), as shown in FIGS. 13 and 14, it is performed in the following order. (The numbers in brackets representing the processing order correspond to those in FIG. 14).

{0} Performs the common process 1.2.

{1} The location where the keyword to be registered, found as a result of the process in step 2, is stored is stored as location information 1 (this process is a special process for creating auxiliary information).

{2} Performing the process in step 4 fails to find a place to register a new record number in the record number cell.

{3} As a result of the process of {2}, the storage position of the next record number cell pointer in the last record number cell of the record number cell group of the same keyword is stored as position information 2. (This process is a special process for creating auxiliary information) {4} Store the record number to be registered at the beginning position of the free record number cell indicated by the free record number area designation pointer.

{5} Set the information of the next record number cell pointer of the record number cell that newly stores the record number to the contents of the free record number area specification pointer (prior to this process, save the contents of the free record number area specification pointer) ).

{6} Set the contents of the next record number cell pointer at the position indicated by position information 2 to the contents of the next record number cell pointer in the record number cell that stores the record number.

{7} Replace the information of the saved empty record number area pointer with the contents of the next record number cell pointer at the position indicated by position information 2.

{8} Update the number of registered record numbers (increase by 1).

{9} Set the auxiliary information type item to 2 (this process is a special process for creating auxiliary information).

In this process, processing time is consumed by scanning the registered keyword position in the keyword area at {0} and finding a free area in the record number cell at {2}.

A case in which the same registration process is reconfigured using auxiliary information will be described in detail. (The bracketed numbers representing the processing order correspond to those in FIG. 14).

(0) Perform common processing 1' and 2'.

(1) Store the record number to be registered at the start position of the free record number cell indicated by the free record number area pointer.

(2) Set the information of the free record number area specification pointer to the contents of the next record number area pointer in the record number cell that stores the record number (prior to this process, save the contents of the free record number area specification pointer ).

(3) Set the information of the next record number cell pointer at the position indicated by position information 2 to the contents of the next record number cell pointer in the record number cell that stores the record number.

(4) Set the information of the saved empty record number area pointer to the contents of the next record number cell pointer in the record number cell indicated by position information 2.

(5) Update the number of registered record numbers in the keyword cell indicated by location information 1 (increase by 1).

It can be seen that the registration process can be completed through the above process, and the processing speed is high.

In the above, a comparison between normal update processing and reconstruction processing using auxiliary information was performed using examples of registration processing (three cases: new keyword registration, record number registration 1, and record number registration 2). In this result registration process,
It can be seen that the reconstruction process using auxiliary information is faster than the normal update process.

Next, the process of deleting search information will be described. Deletion of search information can be performed by an information deletion process that pairs the record number of the search information record and the keyword group in the record in order to delete the search information record to be deleted. The process of deleting a specific record number of a specific keyword will be explained using FIGS. 15 and 16. The deletion process used here does not involve actually releasing an area, but is an example of logical deletion by replacing the record number with a deletion mark. Also, the record numbers deleted here can be deleted all at once through garbage collection processing. This process will be explained using FIGS. 17 and 18.

First, deletion processing during normal update processing will be explained.
(The numbers in brackets representing the processing order correspond to those in FIG. 16).

{1} Extract the record number and keyword to be deleted.

{2} Scan and discover where in the keyword area the retrieved keyword is stored. If it cannot be found, perform error processing.

{3} Store the discovered position as position information 1 (this process is a special process for creating auxiliary information).

{4} Determine the beginning of the record number cell using the record number area pointer in the keyword area, and scan the record number cell for the position where the record number to be deleted is stored.

{5} As a result of the process in {4}, a deletion mark is set at the position of the record number to be deleted in the discovered record number cell.

{6} Update the number of deleted record numbers (decrease by 1).

{7} The position where the deletion mark was set in {5} is stored as position information 2 (this process is a special process for creating auxiliary information).

{8} Set the auxiliary information type item to 3 (this process is a special process for creating auxiliary information).

Even in this process, the search information for the position of the deletion keyword in the keyword area performed in {2} and the process for finding the record number to be deleted in the record number cell performed in {4} take time.

Next, the deletion process during the reconstruction process using auxiliary information will be described (the numbers in brackets representing the processing order correspond to those in FIG. 16).

(1) Store a deletion mark at the location indicated by location information 2.

(2) Update the number of registered record numbers in the keyword cell indicated by location information 1 (decrease by one).

The deletion process can be completed with the above process,
It can be seen that the processing speed is very fast.

Furthermore, an example in which auxiliary information at the time of deletion is used in garbage collection processing will be explained as a special process using auxiliary information for record number deletion processing using FIGS. 17 and 18.

(1) Check whether a record number is registered after the position of position information 1 in the record number cell indicated by position information 1. If not, delete the deletion mark for location information 1 and exit.

(2) If there is, move the record number forward one by one.

(3) Check whether there is a record number in the last record number cell. If not, the contents of the next record number cell pointer in the final record number area are replaced with the contents of the immediately preceding next record number cell pointer. If it's hot, finish.

(4) Replace the contents of the empty record number area pointer with the contents of the next record number area instruction pointer of the last record number cell.

(5) Replace the contents of the empty record number area pointer with the position information of the last record number cell.

With the above processing, garbage collection processing can be performed at high speed without brute force checking the deletion marks in the record number cells. Therefore, the application of the present invention, which creates auxiliary information and uses it for update processing, is not limited to processing for expanding search information at startup, but can also be used for general data structure modification processing. I understand.

Furthermore, the case of record number change processing will be explained using FIGS. 19 and 20. The update processing is performed in the following order (the numbers in brackets representing the processing order correspond to those in FIG. 20).

{1} Extract the record number to be changed, new record number, and keyword.

{2} Scan and discover where in the keyword area the retrieved keyword is stored. If it cannot be found, perform error processing.

{3} Store the new record number itself as position information 1 (this process is a special process for creating auxiliary information).

{4} Use the record number cell pointer in the relevant keyword cell to take out the first record number cell, and scan the position in the record number cell where the record number to be changed is stored.

{5} As a result of the process {4}, the position of the record number to be changed in the discovered record number cell is stored as position information 2 (this process is a special process for creating auxiliary information).

{6} Store a new record number in the position of position information 2.

{7} Set the auxiliary information type item to 4 (this process is a special process for creating auxiliary information).

In this process as well, the process of searching for the deletion keyword position in the keyword area in {2} and the process of finding the record number to be deleted in the record number cell in {4} consume processing time.

Next, the process of changing the record number during the reconstruction process using auxiliary information will be described (the numbers in brackets representing the processing order correspond to those in FIG. 20).

(1) Store location information 1 at the location indicated by location information 2.

The processing can be completed with the above processing, and it can be seen that the processing speed is extremely high.

The above describes the various update processes for the search index in the internal storage device, using examples using the auxiliary information format shown in Figure 7, and comparing the normal update process and the reconstruction process performed using auxiliary information. I've explained it. Similar processing is possible when using other auxiliary information formats such as the one shown in FIG.

In summary, when updating the search index in the internal storage device, auxiliary information is created, this auxiliary information is stored together with the search information in the external storage device, and is needed when starting up the system of an information retrieval device that uses internal memory search. By using this search information and auxiliary information, the reconfiguration process of the search index in the internal storage device (developing the search index in the internal storage device into a format suitable for searching) is performed using the search information and auxiliary information. It can be performed faster than when processing is performed. Particularly when applied to an information retrieval device for a document file system as shown in Figure 6, this is a highly effective internal system that simultaneously satisfies both high-speed and advanced search processing functions and high-speed system startup processing. It is possible to configure an information retrieval device using a memory retrieval method.

Effects of the Invention As described above, according to the present invention, when updated information is added to an information retrieval device, auxiliary information and search information that streamline the process of reconfiguring the search index in the internal storage device are provided. By storing the search information in an external storage device and using this search information and auxiliary information, the internal storage device has a logical structure suitable for searching, and the search information in the external storage device can be restored to the search index in the internal storage device. This makes it possible to perform the constituent processing at a higher speed than when performing normal update processing. As a result, it is possible to realize an information retrieval device using an internal storage retrieval method that simultaneously satisfies both high-speed and advanced retrieval processing functions and high-speed system startup processing.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an embodiment of the information retrieval device of the present invention, and FIG. 2 shows search information created in the information retrieval device of the present invention in accordance with one update process and stored in the external storage device 102. and Fig. 3 is a diagram showing an example of the auxiliary information format.
A diagram showing an example in which the information in a of the figure is actually stored in an external storage device, and FIG. FIG. 5 is a block diagram of an information retrieval device according to a different embodiment of the present invention; FIG. 6 is a conceptual diagram of a document file system to which an embodiment of the present invention is applied; FIG. 7 is a diagram showing the logical structure when expanded. Figure 8 shows an example of the auxiliary information format.
The figure shows another example of the auxiliary information format, FIGS. 9 to 20 are diagrams showing the update process contents in the keyword area and record number area to explain the update process, and FIG. FIG. 22 is a block diagram of a conventional information retrieval device, in which search information is stored in a primary normalized form in an external storage device. 101... Information format conversion mechanism, 102... External storage device, 103... Internal storage device, 104...
Information retrieval mechanism, 105... Information retrieval mechanism, 106
...Auxiliary information extraction mechanism, 107...Information storage position determination mechanism, 108...Information storage mechanism, 110,1
11... Information retrieved from external storage device, 1
12...Information divided into search information and auxiliary information,
113, 114... Information instructing search information, its storage location, method, etc.

Claims (1)

[Scope of Claims] 1. An external storage device that holds search information in which data is stored for each record, which is a logical unit; and an internal storage device that holds index information of the search information;
Means for creating index information from the search information, wherein when reproducing the index information, the search information is retrieved record by record, and added to or subtracted from the updated data stored in the search information for each record. or index information creation means for creating index information by updating a corresponding update position in the index information with the update data using the update method and update position information of the update data, which are either changes; ,
information retrieval means for searching the index information and extracting corresponding data from the search information; and means for updating the index information and the search information in response to a request for addition, deletion, or modification of the search information, When an information update request and update data are input, information regarding an update method and update position in the index information of the update data is obtained, and based on the result, the index information is updated with the update data, and An information retrieval device characterized by comprising a search information update means for storing the update data, the update method, and the information regarding the update position in the search information. 2. When the index information update means inputs a search information update request and update data, the index information update means determines the update method and update position in the index information of the update data, and based on the result, updates the index information with the update data. After updating and storing the update data, update method, and update position information in the search information, if a certain number of update requests are processed, the entire index information and the last recorded search information record in the search information and the position information of the last recorded search information record in an external storage device, and when reproducing the index information, the index information creation means stores search information from the external storage device after the position indicated by the position information of the last recorded search information record. Retrieving each record and updating the corresponding update position in the index information with the update data using the update data recorded in the search information for each record, the update method of the update data, and the update position information. An information retrieval device according to claim 1, wherein index information is created by: 3. The information retrieval device according to claim 1 or 2, wherein the external storage device is a write-once recording medium.