JPH0447350B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION Field of Industrial Application The present invention relates to an information retrieval device generally referred to as a database.

BACKGROUND TECHNOLOGY In conventional information retrieval devices, tables are each independent management unit, tables are not grouped, and search indexes are usually created for each table.

Relational databases have the concept of a view, and although it is possible to create virtual tables that do not actually exist, views are fixed at the time of definition and are imported every time a new table is created. There are no dynamic functions that can be used.

Even in the world of simple languages, if not databases, there are some that manage multiple tables in units called binders, and allow searches not only within individual tables but also across all tables within the binder.

Problems to be Solved by the Invention However, when it becomes necessary to search across multiple tables of the same type, the indexes created for each attribute are independent. When searching for an attribute value included in one of the attributes, the search condition expression becomes quite complex, resulting in a significant drop in search speed.

Further, the binder is still a binder, and the management of tables within the binder and the contents of the tables are at the discretion of the user, and the management of groups is left to the user, and the format thereof is limited to the binder name.

In view of this, the present invention automatically creates a common index for corresponding attributes of multiple tables,
To provide an information retrieval device that automatically searches all indexes of grouped tables even if no common index has been created, and enables high-speed searches beyond the boundaries of individual tables. purpose.

Means for Solving the Problems The present invention provides information management means for grouping and managing tables, which are sets of records consisting of a plurality of attributes, and a common index for the corresponding attributes of each table in the group. This information retrieval device includes an index creation means for creating an index, and a search means for retrieving a target record by referring to an index created in common for corresponding attributes of each table in a group.

Effects According to the present invention, the index creation means automatically creates a common index for corresponding attributes in the grouped tables managed by the information management means, and performs the search. The means are
By automatically searching indexes for corresponding attributes regarding searches within grouped tables managed by the information management means, attribute values across multiple tables can be searched at high speed.

Embodiment FIG. 1 shows a block diagram of an information retrieval device according to an embodiment of the present invention. In FIG. 1, 1 is an information management means, 2 is an index creation means, 3 is a search means, and 4 is an index.

The information management means 1 manages information consisting of a collection of a plurality of attributes in units called records, and manages a table, which is a collection of records, on a main memory or a secondary storage. Now, if we consider the case where records are of fixed length and all records are held in secondary storage,
Its storage form is as shown in FIG. In Figure 2, 5 is secondary memory, 51, 52, ... are tables, 5
11, 512, . . . are records. Every record is given a record serial number, and the storage position of each record in the table can be determined by (record serial number) x (record length). below,
The record number will be expressed as # (table number): (record serial number). A record consists of multiple attributes, and the format of the record, that is, the type of the table, is managed using template information. An example is shown in FIG. In FIG. 3, 6 is type definition information, 61 is a type number, 62 is an attribute name, 63 is an attribute type (character string, numerical value, etc.),
64 is the length (width) of the attribute, and 65 is the index number.
1 in a series of records with the same value of type number 61
Two types are defined. The table is created using one of the types registered in the type definition information 6, and its correspondence is managed using the table management information. An example is shown in FIG. 4, where 7 is table management information, 71 is a table number, 72 is a type number, 73 is a table index commonization flag, and 74 is a table group number.

When the index creation means 2 is instructed to create an index regarding a certain attribute AN _ij in the table T whose type is i, first, the value F _T (1: If F _T = 1, the position of the attribute AN _ij in the record and its length are known from the type definition information 6, and the correspondence between the attribute value and the record number is determined. An index, which is information, is created within the index indicated by index number _Iij (FIG. 5). If F _T =O, index commonization is not applied, so an index TI _ij dedicated to table T is created in the same way. In FIG. 5, 8 is the index I _ij , 81 is the attribute value, and 82
is the record number of the record with that attribute value.

When the search means 3 searches for attributes AN _ij in a group of tables grouped by type,
The index number I _ij regarding the attribute AN _ij to be searched is known from the type definition information 6, and the attribute value 8 in that index is found.
1 and an attribute value given as a search condition (match, mismatch, size, inclusion, etc.), and when an attribute value 81 that meets the condition is found, a record number 82 corresponding to that attribute value is found, and that record number 82 is found. The target record is retrieved from the secondary storage 5. Furthermore, for tables that have not been grouped by index sharing (value of the index sharing flag 73 = O), the same operation is performed on the index TI _ij related to the attribute AN _ij to be searched.

The operation of the information retrieval device of this embodiment configured as described above will be described below.

FIG. 6 shows a first management form in the case where the information management means 1 in this embodiment groups information by sharing an index. In FIG. 6, 10 is table management information A, 1 to M, (M+
1) to (M+N) are tables. Here (M+1)
~(M+N) is a group of first tables having the same type i and a common index, and index commonization flags F _M+1 to _{F M+N} of these tables are all set to 1.

Now, they belong to the group of the first table, but have different table numbers T1 and T2. However, (M+1)≦T1≦(M+
Attribute AN _ij for two records #T ₁ : R ₁ and #T ₂ : R 2 belonging to N), (M+1)≦ _T2≦ (M+N)
When the index creation means 2 is instructed to create an index for record number #T ₁ :R ₁
Thus, the table number _T1 is obtained, and the table management information A10 is used to know that the type is i. Next, the structure of the attributes of type i is known from the type definition information 6, and the index number I _ij for the attribute AN _ij is known, and the attribute value K ₁ of the attribute AN _ij of record #T ₁ : R ₁ and the record The numbers #T ₁ and R ₁ are registered in the index I _ij in correspondence with each other.
Follow the same procedure for record #T ₂ :R ₂ ,
The attribute value K ₂ and the record number #T ₂ :R ₂ are associated and registered in the index I _ij (FIG. 7).

When searching for a table in which the index is shared by type, the search means 3 learns that the attribute AN _ij in the type i has the index I ij by referring to the type definition information 6, and searches the index I _ij by referring to the type definition information ₆ . By searching, it becomes possible to find records #T ₁ :R ₁ and #T ₂ :R ₂ beyond the framework of the table.

FIG. 8 shows a second management form in the case where the information management means 1 in this embodiment groups information by sharing an index. In FIG. 8, 20 is table management information B, 1 to M, (M+
1) to (M+N) are tables. Here (M+1)
~(M+N) are tables belonging to the second group S that have the same type but do not have a common index, and the index commonization flags F _M+1 ~ _{F M+N} of these tables are all O.
shall be.

Now, although they belong to the second group S, they have different table numbers T3 and T4, where (M+1)≦T3≦(M+N).
When an instruction is given to create an index for the attribute AN _K for two records #T ₃ : R ₃ and #T ₄ : R ₄ belonging to (M+1)≦T4≦(M+N), the index creation means 2 is from record number #T ₃ :R ₃ ,
Obtaining table number T ₃ , knowing from table management information B20 that table T ₃ has a dedicated index, and adding attribute value K ₃ and record number #T ₃ to index TI _K3 for attribute AN _K : R ₃
and register it in correspondence. For record #T ₄ :R ₄ , the attribute value K ₄ and record number #T ₄ :R ₄ are registered in the index TI _K4 in correspondence with each other by the same procedure (FIG. 9).

The search means 3 knows that tables T ₃ and T ₄ belong to the same group S by referring to the table management information B20, and retrieves the indexes TI _K3 and TI _K4 and all other corresponding indexes TI _KS for the attribute AN _K. By searching for records #T ₃ :R ₃ and #T 4 :R 4 beyond the framework of the table, it becomes possible to search for records #T 3 :R 3 and #T ₄ :R ₄ .

As described above, according to this embodiment, by defining grouping related to index creation or grouping related to searching for a plurality of tables in the database, searching beyond the boundaries of the tables can be performed at high speed.

For information on how indexes work, see ISAM, B-
There are several methods such as TREE and HASH, which are not limited to this embodiment, and it goes without saying that grouping related to index creation and grouping related to searching can be defined independently of each other.

Effects of the Invention As explained above, according to the present invention, it is possible to perform a high-speed search for attribute values across multiple tables having the same format without complicating the search condition expression, and it is possible to put this into practical use. The effect is significant.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of an information retrieval device according to an embodiment of the present invention, FIG. 2 is a storage format diagram of tables and records in secondary storage, FIG. 3 is a structure diagram of type definition information, and FIG. 4 is a diagram of the storage format of tables and records in secondary storage. A structural diagram of table management information, Fig. 5 is a structural diagram of an index, Fig. 6 is a diagram of the first table management form in the same embodiment, and Fig. 7 is an explanatory diagram of registration in the index in the first table management form. , FIG. 8 is a diagram of a second table management format in the same embodiment, and FIG. 9 is an explanatory diagram of registration in an index in the second table management format. 1... Information management means, 2... Index creation means, 3
... Search means, 4 ... Index, 5 ... Secondary memory, 5
1,52~...Table, 511,512~...Record, 6...Type definition information, 7...Table management information, 8...
...Index, 10, 20...Table management information.

Claims (1)

[Scope of Claims] 1. Information management means for managing a plurality of tables, which are sets of records having a plurality of attributes, by grouping a group of tables having at least one common attribute among the plurality of tables; index creation means for creating an index of attribute values for a specific attribute in the table; and search means for obtaining a target record in the table by searching the index, the index creation means , identifying which group the table belongs to and creating an index for attributes common to each table in the group; An information retrieval device characterized by having a function of referencing and retrieving the target record.