JPH02178876A - Data retrieval system - Google Patents

Abstract

PURPOSE:To decrease the frequency of the input and output of data and the use rate of a CPU, and to shorten a retrieval time by generating a key file regarding only optional retrieval items and deciding other retrieval items as to only hit records. CONSTITUTION:The key file 4 which is rearranged by retrieval items is provided and a hit table 5 wherein hit records among all records are set to '1' and unhit records are set to '0' while one bit correspond to one record is generated. Then unhit records corresponding to bits of '1' in the hit table 5 are corrected to '0' repeatedly as to other retrieval items and when all retrieval items are handled, records satisfying record retrieval conditions corresponding to bits of '1' in the hit table 5 are decided. Consequently, the frequency of the input and output of data and the use rate of the CPU are decreased to shorten the retrieval time.

Description

[Detailed Description of the Invention] [Field of Industrial Application] The present invention relates to a data search method, and more particularly to a data search method for searching records that satisfy logically expressed search conditions from all target records in a database. Regarding.

[Conventional technology]

FIG. 5 shows a first conventional example of searching for a record that satisfies a search condition consisting of a plurality of search items (hereinafter also referred to as "r items") logically expressed from all target records in a database. FIG. 2 is an explanatory diagram showing the status of each table in memory in FIG.

Fig. 5 (a), (b), and (c) respectively show search condition table 1. The key file (search data table) is 2 degrees and the hit table 3. The numbers on the left side of each margin indicate the record number, and the alphabetic characters above indicate the type of search item.

" in the search condition table 1 and search data table 2 columns.
The numbers "o" to "5" tentatively indicate the content of each search item, and even if they have the same value, different items do not necessarily have the same content.

In the hit table 6, 1 bit corresponds to 1 search item,
If it hits, it becomes 1°, and if it doesn't, it becomes 0°.

Each record in the key file 2 is compared for each search item with the corresponding search item in the search condition table 1,
The result is recorded in the corresponding bit of hit table B.

Next, for the completed hit table B, a logical operation is performed for each record based on the logical expression of the search condition, and if the result is 1'', it is determined that the record satisfies the search condition.

For example, the logical formula is AnBnCn...nF, that is, 6
If all items are AND, the 12th record (content = r132504J') satisfies the search condition.

As a second embodiment, for example, Japanese Patent Application Laid-Open No. 62-159222
There is a number.

This invention satisfies the search conditions expressed by a logical formula from the entire set of records in the database in order to reduce the number of bitmap input/outputs (synonymous with a hit table) and the CPU usage rate and speed up the search. When searching for data, a subset of records that are intermediate results of the search is temporarily held as a bitmap, and the search is performed using this held bitmap. Records that correspond to the search criteria and match the intermediate search conditions are set to bit-on.
The bit group obtained by turning off the bits of other records is divided into a plurality of bit sets each consisting of a predetermined number of bits to form a bit map of the second level, and one bit corresponds to one of the bit sets, and at least one bit is turned on. The bit set containing the state is turned on and the other bit sets are turned off, and the obtained bit set is divided into a plurality of bit sets each consisting of a predetermined number of bits to form a second level pit matsub, and at least The first and second level bitmaps are temporarily held as a subset of records that are intermediate results of the search, and this subset is used to perform the final record search.

[Problem to be solved by the invention]

However, in the first conventional example, when a search condition consisting of multiple search items is given, all items are compared for each record to create a hit table, and then it is determined whether the search condition is satisfied or not. Judging.

The second conventional example is a method of extracting records that satisfy the search conditions after making the above determination, and does not speed up the operations from creating a hit table to determining the conditions.

In recent years, as various databases have been developed, the number of records stored in these databases has become enormous.

Regarding this, if the search conditions are simple, the number of output records will be too large for the user to process, so the search conditions will gradually become more complex with more search items.

According to the same method, since the search time is approximately proportional to the product of the number of records and the number of search items, there is a problem that even if computers become faster, the search time does not shorten or even increases.

In particular, there is a problem in that determination using AND logic, which is frequently used, takes more time than determination using OR logic.

This invention has been made in view of the above points, and aims to shorten search time by reducing the number of data input/outputs and the CPU usage rate.

[Means to solve the problem]

In order to achieve the above object, the present invention provides a data search method that searches for records that satisfy search conditions consisting of a plurality of search items logically expressed from all target records in a database. With an organized key file, for any search item, create a hit table where 1 bit corresponds to the code and set the hits of all records as 1" and the non-hits as 0°, and then For the search item, the record corresponding to the 1° bit in the hit table that does not have a hit is successively corrected to O°, and when all search items are completed, the 1° bit in the hit table is corrected. A record corresponding to a certain bit is determined to be a record that satisfies a search condition.

[For production]

According to the data search method according to the present invention, a key file for an arbitrary search item is first created, and then other search items are determined only for the hit records.

In this way, records that do not hit the search items are excluded one after another, so the number of records subject to judgment is drastically reduced, reducing the number of data input/outputs and CPU usage, and shortening search time. .

〔Example〕

Hereinafter, this invention will be specifically explained based on examples.

FIG. 1 is a functional block diagram of an embodiment according to the present invention, which includes a computer 10 and an external storage device 20. As shown in FIG.

The external storage device 20 is a magnetic drum (not shown).

It is composed of a storage medium such as a magnetic disk or magnetic tape and its driver, and the database 21 is stored on the storage medium.

The computer 10 uses search condition table 1. Key file 4. A RAM 1l that stores the hit table 5, a search condition table generation means 12 that generates the search condition table 1 based on the search conditions given by the operator, and a search condition table generation means 12 that reads search data corresponding to the necessary search items from the database 21 and sets the key. A key file generating means 13 that generates a file 4; a hit table generating means 14 that generates a hit table 5 by comparing each record of the key file 4 with the content of the corresponding search item of the search condition table 1; 5, and a record search means 15 for determining a record to be compared for the next search item according to the contents of each bit of 5.

FIG. 2 is a flow diagram of this embodiment.

Fig. 3 is an explanatory diagram showing the contents of each table and file in RAMII, and (a) to (c) in the figure correspond to (a) to (c) of the conventional example (Fig. 5), respectively. ing.

The difference between the embodiment shown in FIG. 3 and the conventional example is the key file 4 shown in FIG. 3 (B) and (C). hit table 5
is the conventional key file 2. While the hit table 3 was a two-dimensional array, it is now a one-dimensional array with one vertical column corresponding to the search items.

FIG. 4 is an explanatory diagram showing how the contents of the hit table 5 are being updated.
The bit that changed to o゛ remains blank from then on (on the right).

This will be explained below with reference to FIGS. 2 to 4.

The given search conditions are 6 as shown in Figure 3 (a).
The content of each of the search items A to F is "1".

r3'', r2'', rJ, ro'', r4J, and the logical formula=AnBnCnDnEnF That is, this corresponds to a case where the frequency of use is generally high and the conventional amount also takes time.

When the process starts, a search condition table 1 is first generated based on the search conditions, and then the key file 4 shown in FIG.

Next, the contents of the key file 4 are compared with the contents "1" of item A of the search condition table 1 in order from record number 0, and depending on whether there is a hit or not, the contents of the key file 4 are changed to the hit table shown in FIG. 5 is generated.

Next, data about other search items, such as item B, is read into the key file 4, and only for records whose content in the hit table is 1°, the content is the content "3" of item B in the search condition table 1. If there is a hit, the corresponding bit in the hit table 4 remains at 1°, and if there is no hit, it is converted to 0°, thereby updating the hit table 5.

In this way, key file 4 is used until all search items are completed.
The generation of and the updating of the hit table 5 are repeated, and the process ends when all search items are completed.

At this time, the record corresponding to the bit set to "1" in the hit table 5 is the record that satisfies the search condition.

As is clear from the example shown in Figure 4, among the 18 records (numbers 0 to 17), 12 records were hit for item A in the first time, but for item B in the second time. There were 7 records that also hit item C (numbers 3, 5, 9, 12.1).
7) is reduced to 5.

Finally, there is one record (number 12) that also hits item F and satisfies the search condition.

Generally speaking, if there are as many key files as the number of search items in the search conditions, 4. The memory capacity required for the hit table 5 is the same as that of the conventional key file 2. The memory capacity required for the hit table 3 is 1/1, or if the memory capacity is the same, the number of records that can be processed in one read is doubled.

Also, the average number of categories per item is 8 (S>1),
Assuming that they exist randomly among all R records, the number of records targeted for each item is reduced to 1/S, and the number of records remaining at the end is R/SK.

Furthermore, the number of times H of comparison and determination as to whether or not there is a hit is as follows.

Compared to this, in the conventional example, since the number of times is RK, the number of times H is reduced to S/K (S-1), and the search time is shortened accordingly.

In this example, K=6. S=6 (rOJ~"5"
), and H=1.19997R:1.2R. In the conventional example, it is 6R, so the number of comparisons is 115.
Therefore, the search time is also reduced to 115.

As explained above, the effect of reducing the required memory capacity and search time becomes more significant as the search items become more complex and the number of items increases.

〔Effect of the invention〕

As described above, according to the present invention, it is possible to reduce the number of data input/output and the CPU usage rate, thereby shortening the search time.

[Brief explanation of the drawing]

FIG. 1 is a functional block diagram of an embodiment according to the present invention, and FIG.
The figure is the same flow diagram, and Figure 3 is the same search condition table, key file,
FIG. 4 is an explanatory diagram showing an example of the contents of the hit table. FIG. 4 is an explanatory diagram showing the update status of the hit table. FIG. 5 is an explanatory diagram corresponding to FIG. 6 in the first conventional example. 1... Search condition table 2, 4... Key file 3.5... Hit table 10... Computer 11... RAM 12.
...Search condition table generation means 13...Key file generation means 14...Hit table generation means 15...Record search means 20...External storage device 21...Database "1 Fig. 1" Fig. 2 Figure 4 (A) (C)

Claims (1)

[Scope of Claims] 1. In a data search method that searches for records that satisfy search conditions consisting of a plurality of search items logically expressed from all target records in a database, keys organized by the search items are provided. A hit table is prepared in which each bit corresponds to one record for an arbitrary search item, and a hit table is set as "1" for a hit among all the records, and a hit table is set as a "0" for a non-hit record. For the search items in the hit table, records corresponding to bits that are "1" in the hit table that are not hit are successively modified to "0", and when all search items are completed, the records in the hit table are A data search method characterized by determining a record corresponding to a bit that is "1" as a record that satisfies a search condition.