JPH02132559A - Natural connection arithmetic processing system - Google Patents

Abstract

PURPOSE:To shorten the time required for production of an index of relations by producing an index table consisting of the codes obtained by putting the attributes of two index values on each other also for a 3rd relation obtained by connecting the attributes of two relations to each other. CONSTITUTION:The relation tuples are retrieved via the 1st and 2nd index tables 1 and 2 of two relations respectively to obtain a 3rd index table 3 consisting of the index records obtained by connecting two relations to each other. Therefore the relation connecting two relations to each other is also retrieved at a high speed via the table 3. In such a way, the existing index table and hash table are used for the relations applying the arithmetic operations. Thus the time required for the natural connection arithmetic operation carried out between the relations can be shortened. Furthermore the relation tuples can be retrieved at a high speed and at the same time an index is obtained to the executing result of the natural connection arithmetic operation carried out between the relations having indexes.

Description

Detailed Description of the Invention (Industrial Application Field) This invention relates to a data arithmetic processing method, and in particular creates superposition codes for attributes of relations given as data, such as age, gender, etc. When searching for , natural combinations (Natural
Joint) related to the calculation method.

(Prior Art) Conventionally, in relational database systems, 81 trees (B”t
The search for relation table has been speeded up by preparing an index such as ``ree'' or by preparing a hash function and a hash table.

The former method, such as an 8" tree, traces the index for the attribute specified by the search request and finally obtains a tuple that satisfies the request. On the other hand, the latter method using a hash function and a hash table, The hash table is searched based on the result of applying a hash function to the attribute value specified by the search request to obtain table candidates that satisfy the request.

In order to create a B+ tree or a hash table for two relations, the attributes of each are combined using a search method such as a B+ tree that each has, or a search method using a hash function and a hash table. An operation, a natural join operation, is performed.

If a search method is also required for the relation that is the result of the operation, either examine the tuples formed by the result of the operation and create a search record of the attributes that serve as a queue for all tuples of that relation, or Or execute the process to calculate the hash value and create a B+ tree or hash value.
Create a table.

(Problem to be Solved by the Invention) However, in the conventional processing method, it was necessary to search for attributes of a relation that is the result of a natural join operation, and to create a new hash table. Nevertheless, the relation to which the operation is applied does not utilize the existing search or hash table.

Normally, relationships used in practical use have tables with hundreds to tens of thousands of items, so creating a new index or hash table for the key attributes of one relationship would take a huge amount of time. . In this way, when searching for a relation that is the result of a natural join operation, the system becomes preoccupied with searching and creating a hash table, which has the disadvantage of significantly reducing the processing speed of the entire system.

This invention reduces the time required for a natural join operation to be executed between relations, enables high-speed searching of relation table, and provides an index for the execution result of a natural join operation between relations that have the index. The purpose of this paper is to provide a natural join operation processing method that can be used to solve problems.

(Means for Solving the Problems) The natural join operation processing method of the present invention creates an index that combines the attributes of two relations by executing a predetermined logical process on two relations having the same attributes. For each of the two relations, a first step of sequentially shifting a group of bits representing each of the plurality of attributes by a predetermined number of bits in the order of arrangement of the attributes; A second step of creating index values of the first and second index tables by taking a logical OR between the bits of each digit of the prefecture attribute shifted in the first step; and a second step of creating an index value of the first and second index tables; a fourth step of shifting five bits of the second index table so that set bits overlap each other in the bit groups of the first and second index tables corresponding to the selected specific attribute; a fifth step of creating a bit group of a third index table by performing a logical OR between the bits of the first index table and the bits of the second index table shifted by the fourth procedure; The bit groups of the index table are made to correspond to the combined index.

(Operation) According to the present invention, it is possible to search for a tuple in the relation using the first and second index tables of each of the two relations, and the third index is made up of an index record that is a combination of the two relations. Since a table is obtained, it is possible to quickly search for a relation obtained by combining the two relations using the third index table.

(Embodiment) FIG. 1 is a diagram illustrating a procedure for creating a relation index using the natural join calculation method of the present invention. FIG. 2 is a diagram illustrating in detail the procedure for creating index values for person relations in FIG. 1. First, a procedure for creating an index value of a person relation will be explained with reference to FIG.

Attributes of a plurality of people and tables 1 to 6 are set as relations in step STAI. In this case, "person name j,""age," and "gender" are set as attributes, as shown in order in the horizontal direction of tuples 1 to 6.

In step STA2, "Person's name j,""Age" and "Gender" are converted into binary values and "0100" consisting of bit groups of 4 bits, 4 bits and 2 bits long respectively, "Age""00lO" and " Get 01''.

In step STA3, [Age] "0010" is shifted to the right by 1 bit for "Person's name""Old00", and further, "Gender. "01" is shifted for "Age""0010".
is shifted to the right by one bit, and then a logical sum (+) is taken for each bit of each digit to obtain the binary value code "old 010". This binary value code 001010'' becomes the index value of index record 1 for table 1.

Similarly, the same logical processing as for table 1 is performed for tuples 2 to 6. As a result, the index value of index record 2 for table 2 is "11101", the index value of index record 3 for tuple 3 is "10110", and the index value of index record 4 for table 4 is "10011". The index value is '011 old'
The index value of index record 6 for table 6 is “0011”
0”.

Furthermore, step ST84 sets an index table using index records 1 to 6, and changes index record 1 to table 1, . 6 and table 6 respectively according to pointers indicated by arrows.

FIG. 1 is a diagram including the procedure explained in FIG. 2, and is a diagram illustrating the procedure for creating an index table of person relations according to the present invention.

In step STB 1, "r department" and "blood type" are set as the attributes of the relation "person 1", and table 1 and table 2 are set as the table.

By step STB2, the relation "Person 1" is
For example, apply a hash to each attribute value of tuple 2, and set "department""10000" to 1 for "person name""01001".
Shift the bit to the left and also "department""10000"
"Blood type J "010" is shifted to the left by 1 bit, and the bits of each corresponding digit are logically summed to obtain the index value "0" of index record 2 of "index table 1".
101001” is obtained.

For index record 1 as well, the index value "1100100" is obtained by the same process as for index record 2.

In step STB3, similarly to step STC 1, "person name" and "
"Age" is set, and tuples 3 to 6 are set as tuples.

In step ST old, as in step STB2, each attribute of the relation "Person 2" is hashed, the binary value is shifted 1 bit to the left, the bits of each corresponding digit are ORed, and "Index Index values of index records 3 to 6 of “Table 2” “old 0011”, “001110”
．． Obtain "101100" and "001011".

In step STB5, based on the condition that the "person names" are the same, a natural join operation is performed on the relations "Person 1" and "Person 2" to obtain tables 7 and 8 of the relation "Person 3".

Therefore, tables 7 and 8 of relation "Person 3" contain the attributes included in tuples 1 and 2 of "Person 1" and tuples 3 to 6 of "Person 2", namely "Person name", ""age","department" and "blood type" are all included.

In step STB6, a natural join operation is executed based on the contents of "index table 1" and "index table 2" to create "index table 3" corresponding to the relation "person 3". For example, the index value of index record 2''
0101001”, and the index value of index record 3 “01”
0011'' is extracted and the index value ``0101001'' of index record 2 is extracted so that the attribute used as the join condition, that is, the binary value ``Old゜'' representing ``Noguchi'' of ``Person's name'' overlaps with each other.
By shifting the index value "010011" of index record 3 to the right by 2 bits and taking a logical OR between the corresponding bits, a binary value "ototo old" is obtained, and this is transferred to the index table 3. The index value of index record 7 is as shown in FIG.

By the same process, the index value 8 of the index record 7 shown in the index table 3 is obtained as "1110110".

FIG. 3 is a diagram showing a procedure for searching the person relations shown in FIG. 1 using an index record according to the present invention. Step STCI sets the attribute "age" in the person relation to "17" and "gender" to "17" as search conditions.
Set to ``Female''.

Step ST (:2) executes the procedure of converting attributes into binary values, and sets "Age" to "October" and "Gender J" to "10". In this case, "Person's name" is specified. Since this is not the case, it is set to "0000".

Step ST(:3) arranges these attributes in the same order as when creating the binary value index, shifts the bit of the next attribute to the right by one bit as shown in the figure, Next, perform a logical OR on the bits of each digit.This results in a query mask of "0".
1101''' is obtained. The Query mask is the result of such logical processing.

In step STC4, this query mask is compared with the bit contents of index records 1 to 6, and the index record that matches for bit 1, in this case, index record 2.
and 5 are selected as indicated by the solid arrows. Unmatched index records are indicated by dotted arrows.

In step STC5, the bit contents of extracted tables 2 and 5 are checked, and it is determined that "age" is "17" and "gender"
It is determined which one satisfies the condition that is "10". In this case, it is determined that Table 2 matches, and the search ends.

FIG. 4 is a diagram focusing only on the procedure for creating index values for tuples 2 and 7 in FIG. 1. Fourth
The difference between FIG. 1 and FIG. 1 is that in step STC8,
By performing the arithmetic processing of shifting the attributes in table 7 and calculating the logical sum, "01010011" is obtained as index value 7' to be compared with index value 7.

This confirms that index value 7' gives the same result as index value 7.

(Effects of the Invention) As described in detail above, according to the present invention, even for the third relation that combines the attributes of two relations, an index consisting of a code obtained by superimposing the attributes of the two index values is obtained. By creating a table, you can reduce the time required to create an index for relations.
Also, such an index can speed up the search time for relations.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing a procedure for creating an index of a relation using the natural join calculation processing method of this invention, FIG. 2 is a diagram showing a procedure for creating an index value for a person relation according to this invention, and FIG. FIG. 4 is a diagram illustrating a procedure for indexing a person relation. FIG. 4 is a diagram illustrating a procedure for creating an index value by focusing only on a specific table in FIG. STAI~STA4, STBI NSTB4, ST
CI P-ST (:8-...step

Claims (1)

[Scope of Claims] A natural join operation processing method that creates a combined index for two relations by executing predetermined logical processing on two relations having the same attribute, comprising: a first step of sequentially shifting a group of bits representing each of a plurality of attributes by a predetermined number of bits according to the arrangement order of the attributes;
a second step of creating an index value of the first and second index tables by taking a logical OR between the bits of each digit of the attribute shifted by the step; and a second step of creating an index value of the first and second index tables; the first corresponding to a specific attribute;
and a fourth step of shifting the bits of the second index table so that set bits overlap each other in the bit group of the second index table, and the bits of the first index table; Perform a logical OR with the shifted bits of the second index table,
a fifth step of creating a bit group of a third index table; and a natural join operation processing method characterized in that the bit group of the third index table is made to correspond to the combined index.