JPS61141036A - Data retrieval system - Google Patents

Abstract

PURPOSE:To improve the data retrieval processing efficiency by replacing comparison conditions with data size comparison conditions in terms of the retrieval having forward coincidence pattern matching comparison conditions. CONSTITUTION:The titled data retrieval system is constituted of a data input device 10 such as a keyboard, a central processing unit 20, a ROM30, a RAM40, an external memory device 50 such as a floppy disk and a data output device 60 such as a display. At the time of data retrieval accompanying with the forward coincidence pattern matching, key value conditions limit the scope of retrieval data, and data is retrieved and processed without the pattern matching comparison. For instance, assuming that the pattern matching comparison conditions are 'column ABC %', it can be converted into comparison conditions of data value, 'ABC <= column value < ABD' and further converted into 'ABC <= key value < ABD', and the retrieval and processing are executed under these key value conditions by referring to an index table.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Application of the Invention] The present invention relates to a data retrieval method, and particularly to a data retrieval method suitable for retrieving data by forward matching pattern match comparison in a relational database or the like.

[Background of the invention]

As shown in Figure 3, a relational database has multiple rows (
It is expressed in a tabular format of row information and multiple column information. One way to search such a relational database is to search for data using a pattern match comparison using a prefix match. This is a method in which the search range is limited to an arbitrary character string of a predetermined length from the beginning of a specified column or multiple columns, a pattern match comparison is performed on the character strings, and matching row information is extracted. Figure 3 is column 1
This is an example of specifying 1 condition and selecting rows 2, 4゜, etc. that match the condition ``ABC% total (% is an arbitrary character string of 0 or more characters).゛In this case, conventionally, pattern match comparison are performed sequentially from row 1 to row n.

There was a problem in that the rows that matched the conditions were selected, and the search time was long.

For example, the technical magazine r'b i tJ (Vol. 10.N
As introduced in ``Algorithms - About Pattern Matching'' in ``Algorithms - About Pattern Matching'', various algorithms have been considered to improve the efficiency of pattern matching processing. However, when searching for data using prefix pattern comparison in a database, etc., consideration must be given to solving the problem that the search time increases by performing pattern matching on all the target data. It has not been.

[Purpose of the invention]

An object of the present invention is to improve search efficiency in a data search that involves a prefix pattern match, such as a search in a relational database.

[Summary of the invention]

In a pattern match comparison with a forward match, the compared string is greater than or equal to the comparison pattern string, all but the last character of the comparison pattern matches, and the last character has a code of 5 and is next to the last character of the comparison pattern. It only needs to be smaller than the character string.
Therefore, a pattern match comparison condition can be converted into a data value comparison condition and an index key value condition. It is characterized by limiting the range of search data based on conditions and performing data search processing without performing pattern match comparison.

For example, if the pattern match comparison condition is ``column ABC% pattern'', the condition can be converted to a data value comparison condition of ``ABC<column value<ABD''.

2, "ABC" is a comparison pattern, and "ABD" is a character string one larger than the last character 11 Cl# of the comparison pattern. It is assumed that the values of the codes increase in the order of A, B, C, D, E, . . . . The above data value comparison condition can be further converted into a key value condition of "ABC<key value<ABD". By referring to the index table using this key value condition, 1. Data search processing can be performed without performing pattern match comparison for all data.

[Embodiments of the invention]

FIG. 1 shows an overall block diagram of a data retrieval system targeted by the present invention. In FIG. 1, 10 is a search condition;
A data input device such as a keyboard used for other inputs, 20 a central processing unit that performs search condition conversion processing, data search processing, etc., 30 a memory in which programs necessary for processing by the central processing unit 20 are stored. (ROM),
40 is a memory (RAM) used to store input data, intermediate results, index tables, etc., 5G is an external storage device such as a floppy disk in which the database to be searched is stored, and 60 is a display unit that outputs the search results. This is a data output device such as a printer.

FIG. 2 shows an example of the processing flow of the data and search method of the present invention, which is executed by the central processing unit 20 in FIG. 1, and its program is stored in the memory 30.

FIG. 2(a) shows the processing procedure when ``Column LIKE ABC%'' is specified as the t* pattern match comparison condition. ``Column LIKE ABC%'' is input as the pattern match comparison condition from the data input device 10. Then, the central processing unit 20 analyzes this and generates a comparison condition "ABC column value <ABD" regarding the column value (step 101), that is, Bacalum LIK
A search with the condition ``E ABC%'' will search from ``ABC...'' to 'ABD...
Since it is possible to think of extracting the following part up to - before ``, in step 101, the fixed character string ``ABC'' obtained by removing the par %'' symbol from the 1-condition pattern string and the last character of the fixed character string are extracted. Character string ``''A'' with ``C'' one larger
Next, in the central processing unit 20, the comparison condition regarding the column value is generated as the condition regarding the key value of the index "ABC < key value <ABD". ” Convert to (
Step 102) indicates the search conditions for the index table 41. The central processing unit 20 searches the index table 41 of the memory 40 under this key value condition and obtains the read address of the searched data (step 1).
03).

The index table 41 has a tree structure, for example, and the corresponding read address is obtained by searching the table in several stages. Since the search itself using an index is well known, the details thereof will be omitted here.

Next, the central processing unit W20 accesses the external storage device 150 using the read address obtained by the index search,
The corresponding data is read out (step 104). This read data is data that matches the pattern match comparison condition "LIKE ABC%" and is displayed or printed out by the data output device !60.

Figure 2(b) shows the processing procedure when "LIKE ABCXX" is specified as a pattern match comparison condition, where the symbol "'x" represents one arbitrary character, and the symbol "x" represents one arbitrary character. 'ABCXX'" means that the first 3 characters are "AB"
C'' and the following two characters mean searching for arbitrary data with a data length of 5.

In FIG. 2(b), steps 201 to 2.4 are the second
Corresponding to steps 101 to 104 in FIG.
In addition to ", a condition regarding data length "LENGTHC column value) = 5" is generated. One of these conditions is "ABC≦-
Step 202: Convert the column value "ABD" to the condition "ABC<key value<ABD" regarding the index key value.
).

The read address of the searched data is obtained by index search based on this key value condition (step 203), and the corresponding data is read out (step 204). Next, from this read data, the condition regarding the data length generated in step 202 "LENGTHC column value) = 5"
When data in the corresponding range is retrieved by , the data matches the pattern match comparison condition
(Step 205)
.

As described above, the present invention does not use pattern match comparison, but uses only index search, or only index search and data length check.

Search processing can be performed.

〔Effect of the invention〕

According to the present invention, by replacing the pattern match comparison condition with the data size comparison condition in a search that has a prefix pattern match comparison condition, it is possible to retrieve the corresponding data without performing a pattern match comparison. This has the effect of improving search processing efficiency.

[Brief explanation of drawings]

FIG. 1 is an overall block diagram of a data search system targeted by the present invention, and FIGS. 2(a) and 2(b) are process flow diagrams illustrating the data search system of the present invention. FIG. 3 is a diagram illustrating a relational database. 10...Data input device, 20...Central processing unit. 30...Memory (ROM), 40...Memory (RAM), 41...Index table. 50... External storage device, 60... Data output device. Figure 1 Figure 3

Claims (1)

[Claims] (1) In a method of searching data using a pattern match comparison of prefix matches, the condition of the pattern match comparison is converted into a condition regarding the key value of the index, the index table is searched according to the key value condition, and the searched data is searched based on the result. A data search method characterized by retrieving more relevant data.