JPH04245563A - Preparation of retrieving table - Google Patents

Abstract

PURPOSE:To prepare a retrieving table which can secure the approximately equal time required for retrieval of the key words included in each index. CONSTITUTION:A key word that shows the contents of the document stored in a storage medium is provided with a pointer for the document number that can be derived from the key word. Then the key words are divided into groups that are shown in the indexes. These indexes contain the pointers that can derive the key words from these indexes, and the number of key words included in indexes are approximately equal to each other through a retrieving table.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for creating a search table for searching documents stored on a storage medium.

0002

2. Description of the Related Art Optical disks have become quite popular in recent years. A feature of optical discs is that they have a huge memory capacity. Therefore, a large number of documents (units of information storage) are stored on one optical disk, and it is necessary to quickly retrieve a desired document from among the documents. For this reason, the search information necessary to search each document is also stored in the storage medium, and this search information is used to access the desired document.

An example of such search information will be explained using FIG. 5. FIG. 5 shows a search table in which keywords representing the contents of a document are divided into predetermined groups, and the groups are represented as indexes, and the keywords are classified for each index. A keyword has a pointer from which a document number, which is the number of a document having the content represented by the keyword, can be retrieved, and an index has a pointer from which a keyword belonging to the index can be retrieved. To search for a desired document using this search table, first, an index that is thought to include a keyword representing the content of the desired document is retrieved, and keywords belonging to that index are searched. If a target keyword is found, one or more document numbers represented by the keyword are obtained from the keyword pointer, and the target document is accessed from among these. If the desired keyword is not included in this index, the keyword in another index is searched.

0004

In creating the above-described search table shown in FIG. 5, the number of keywords belonging to one index (Km in FIG. 5) is often set to a fixed value. That is, if the total number of keywords representing the contents of a document stored in one storage medium is I, the above-mentioned search table is created by setting the number S of indexes to I/Km. However, if the number of keywords exceeds S×Km and is subsequently registered, the number of keywords belonging to the final index (that is, the S-th index) greatly exceeds Km. Therefore, up to (S-1) indexes can be searched in approximately the same amount of time for each index, but the search time for a keyword belonging to the final index is longer than for other indexes.

The present invention has been made in view of the above-mentioned problems, and it is an object of the present invention to provide a method for creating a search table in which the time required to search for keywords in each index is approximately equal.

[0006]

[Means for Solving the Problems] In order to achieve the above object, the search table creation method of the present invention provides a pointer to a document number that can be derived from a keyword that represents the content of a document stored in a storage medium. The keywords are divided into groups, and this group is represented by an index. This index is provided with a pointer to which the keywords contained therein are retrieved. When creating a search table consisting of this index and the keywords belonging to this index, each group is represented by an index. The number of keywords belonging to the index is made approximately equal.

[0007]

[Operation] With the above configuration, the number of keywords belonging to each index is almost uniform, so that the search time can be speeded up as a whole.

[0008]

Embodiments Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 shows a search table created according to an embodiment of the invention. Before explaining this search table, a system for searching using this search table will be explained.

FIG. 2 shows an optical disc search system, in which 1
is an optical disk control device that has a built-in optical disk, 2 is an input/output device that inputs keywords and documents necessary for searching, and displays and records search results and read documents; 3 is an optical disk controller that uses instructions from input/output device 2. This is a system control device that extracts search information from the search device 4 and controls the search performed by the search device 4. The search device 4 expands the search information read from the optical disc and performs a search.

FIG. 3 is a block diagram of the search device 4 shown in FIG. 2. The search device 4 includes a system interface unit 41 that interfaces with the system control device 3, a search control unit 42 that reads search information stored on an optical disk, expands it on a main memory 43 (described later), and performs a search.
It is composed of a main memory 43 that serves as a search work area.

FIG. 4 is a diagram showing a state in which search information is expanded in the main memory 43 shown in FIG. The area of the main memory 43 is composed of an index area 431, a keyword area 432 in which keywords belonging to each index are written, and a document number area 433 representing the document pointed to by each keyword.

Next, the operation will be explained. The search device 4 accesses the index information through the system interface section 41 to the optical disk and expands it into each area shown in FIG. 4 on the main memory 43. Next, calculate the number of keywords A managed by each index by dividing the number of all registered keywords by the number S of indexes that can be created in the index area 431, and sort them in the keyword area 432 in ascending order (or descending order). An index is created for each keyword number A for each index calculated from the beginning of the keyword, and is set in the index area 431. Note that this index has a pointer for accessing the keyword to which it belongs. In this way, the search table shown in FIG. 1 is created.

Next, the operation of searching for a desired document using this search table will be explained. When a certain keyword is input from the input/output device 2 and there is an instruction to search for a document number related to this keyword, the system interface unit 41 receives this search specified keyword, performs a binary search of the index area 431, and then specifies the search specification. A keyword list to which the keyword belongs is detected, and a search specified keyword is detected from this keyword list. Since the number of keywords managed by each index is averaged, the search time for searching the specified search keyword is averaged for each index. Therefore, unlike the search table described in the conventional technology, the number of keywords belonging to each index is determined by the number of indexes and the total number of keywords, so the number of keywords is not concentrated in a certain index, and the search time is faster overall. can be achieved.

[0014] Binary search is called a binary search method, and is an effective search method when data is arranged in a certain order, for example, based on the magnitude relationship between integers or the lexicographical order between character strings. In this method, the target data is searched for by first dividing the whole into two parts and checking which part contains the target data, and then dividing the first part into two parts and checking which part contains the target data one after another.

[0015]

Effects of the Invention As is clear from the above description, the present invention uses a search table created so that the number of keywords managed by each index is approximately equal, so that searches for all keywords are performed to the same extent on average. This can be done in a short amount of time, improving search efficiency.

[Brief explanation of the drawing]

[Figure 1] Diagram showing a search table configured according to this embodiment

[Figure 2] System configuration diagram for executing this embodiment

[Figure 3] Configuration diagram of the search device shown in Figure 2

[Fig. 4] A diagram showing an area in which search information is expanded on the main memory shown in Fig. 3.

[Figure 5] Diagram showing an example of a conventional search table

[Explanation of symbols]

1 Optical disc control device 2 Input/output device 3 System control device 4 Search device 41 System interface section 42 Search control section 43 Main memory

Claims (1)

[Claims] Claim 1: A pointer to a document number that can be retrieved from a keyword is provided for a keyword representing the content of a document stored in a storage medium, the keywords are divided into groups, and this group is represented by an index, and this index includes A pointer is provided to retrieve the keyword contained in the index, and when creating a search table consisting of this index and the keywords belonging to this index, the number of keywords belonging to each index is made to be approximately equal. How to create a search table.