JP2001022775A - Information retrieval device, information compressing method for information retrieval device, and recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To suppress the capacity of an index file by compressing index information more efficiently and to enable high-speed retrieval using the index file. SOLUTION: When entries of index information are successively identical, this information compressing method deletes by an index information compression part 107 the entries except only one entry, puts (n) successive pieces of index information with the same document number in one group as to index information in the same character chain, and generates compressed index information having the document number compressed into one. When byte data of position numbers are not larger than a reference value (0×0f), the low-order four bits of each of the byte data of the position numbers are connected to generate compressed index information having the position number converted into 1-byte data.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an information retrieval apparatus, an information compression method of the information retrieval apparatus, and a recording medium storing a program for executing the information compression method. The present invention relates to an information search device for searching for a document including a character string, an information compression method of the information search device, and a recording medium.

[0002]

2. Description of the Related Art With the spread of word processors and personal computers in recent years, a large amount of electronic document data has been accumulated, and practical use of a document database for retrieving document data as necessary has been progressing. Also, a full-text search method for searching document data from the contents of a document without adding a keyword has attracted attention in recent years and has begun to be used in offices and the like. In this full-text search method, an index file using a character chain is created in advance, and at the time of search, high-speed search is realized by reading out index information for each character chain from the index file.

As a conventional information retrieval apparatus using such a full-text retrieval method, for example, there is one as shown in FIG. FIG. 18 is a configuration diagram showing a conventional information search device. In FIG. 1, an information search device of the conventional example is a terminal 1
101, document input unit 1102, search condition input unit 110
3. Result output unit 1104, overall control unit 1105, index creation unit 1106, bibliographic entity registration unit 1107, search unit 110
8, a document number storage unit 1109, a bibliography acquisition unit 1110, an entity acquisition unit 1111, an index position storage unit 1112, an index information storage unit 1113, a bibliography information storage unit 1114, and an entity information storage unit 1115.

In the conventional information retrieval apparatus 1100, when a document is first registered, a document input from a terminal 1101 is indexed by a document input unit 1102 into an index creation unit 1106.
After converting the format to a format that can be handled by
Send to 105. The format conversion referred to here is, for example, format conversion of a document for creating an index, such as extracting bibliographic information from the document by extracting bibliographic information from the document.

Next, the overall control unit 1105 sends the format-converted document data to the index creation unit 1106. The index creation unit 1106 creates index information for each character chain from the format-converted document data using a two-character chain described in Japanese Patent Application Laid-Open No. 6-52226. The index information is a set of position information that specifies the position of a character chain in the document data. More specifically, a document number for identifying the document, and
A position number for determining where the character chain is located in the document corresponds to the position information. Index creation unit 110
6 is stored in the index information storage unit 11
13 to form an index file. At this time, for each character chain, head position information indicating the head position of the index information corresponding to the character chain in the index file is associated with length information (number of bytes) indicating the data length of the index information. , In the index position storage unit 1112.

A general control unit 1105 registers a bibliographic entity registration unit 1107 for registering bibliographic information and entity information of a document.
Also, the document data that has undergone format conversion is sent. In the bibliographic entity registration unit 1107, only the bibliographic information in the document data is stored in the bibliographic information storage unit 1114.
115 stores only the entity information in the document data.

On the other hand, when searching for a document, the terminal 11
01 and the search condition input from the search condition input unit 1103 via the overall control unit 1105 to the search unit 1108
Send to The search unit 1108 creates character chain information from the search character string of the search condition, and then creates the index position storage unit 111.
2 to obtain the head position information and the length information associated with the character chain information. Next, it accesses the index information storage unit 1113 using the acquired head position information and length information, and acquires the corresponding index information from the index file. Next, the search unit 1108 performs a full-text search by determining the continuous status of the index information, stores the document number and the search formula number that match the search conditions in the document number storage unit 1109, and determines the number of search cases by the overall control unit 1105. Send to Next, the overall control unit 1
When 105 sends the number of searches to the result output unit 1104, the number of searches as a search result is displayed on the terminal 1101.

[0008]

As described above, in the conventional information search apparatus 1100, an index file is created from document data using a character chain at the time of document registration, and each index is created from the index file at the time of search. A high-speed search is realized by reading out the index information. However, assuming a Japanese document, the data size of the index file is about twice as large as the document data, so a large-capacity storage medium is required for the index information storage unit 1113 for holding the index file. There was a problem. That is, while the data size of the document data is “the number of characters of the document data × 2 bytes”, the data size of the index file is such that the index is created using a character chain of two characters, and the number of indexes per document is Is almost the same as the number of characters in one document data, so that if the index information is composed of a 2-byte document number and a position number, the number of characters is 4 characters.

In addition, since the search is performed using an index file having a large data size, especially when the index file is held in a recording medium such as a CD-ROM which has a low data read speed, the search is not performed. There is also a problem that a relatively long time is required, resulting in a low search speed.

In order to solve these problems, attempts have been made to compress the index file. However, since the index information is binary data with little continuity, for example,
When compression was performed using an existing compression method such as "compress" of NIX, only about 10% was compressed, and a significant effect was not obtained. In addition, when a complicated compression method is used, there is a problem that a relatively long time is required for decompressing the compressed file, which results in a problem that a search speed becomes low. Had not been done.

SUMMARY OF THE INVENTION The present invention has been made in view of the above-mentioned conventional problems, and it is an object of the present invention to compress index information more efficiently, reduce the size of an index file, and perform high-speed search using the index file. It is an object of the present invention to provide an information retrieval apparatus, an information compression method of the information retrieval apparatus, and a recording medium that can perform the information retrieval.

[0012]

According to a first aspect of the present invention, there is provided an information retrieval apparatus having an index for specifying a specific character chain in document data by position information. In the apparatus, for each character chain appearing in the document data, index information creating means for creating index information including the position information, and for the index information of the same character chain, the position information is stored in accordance with the value of the position information. Index information compression means for creating compressed compressed index information.

According to a second aspect of the present invention, in the information search apparatus according to the first aspect, the index information compressing means includes a first information included in the position information for the index information of the same character chain. A plurality of pieces of index information having the same position information are set as one group, and compressed index information is created by compressing the first position information of the group into one.

According to a third aspect of the present invention, in the information search apparatus according to the first or second aspect, the index information compressing means is configured to execute position information for index information or compressed index information of the same character chain. When the second position information included in the second position information is equal to or smaller than a predetermined reference value, compressed index information is created by compressing the second position information to a smaller number of bits or bytes.

According to a fourth aspect of the present invention, in the information search apparatus according to the second aspect, the index information compressing means is configured to compress the first position information into one piece of compressed index information. Calculating a difference between the second position information included in the position information, and compressing the second position information into a difference represented by a smaller number of bits or bytes when the difference is equal to or smaller than a predetermined reference value. It is.

According to a fifth aspect of the present invention, in the information search apparatus of the first, second, third or fourth aspect, the index information includes a plurality of hierarchical position information.

According to a sixth aspect of the present invention, in the information search apparatus according to the second, third or fourth aspect, the first position information is used as document identification information and the second position information is used as the document identification information. This is the position information in the document specified by the first position information.

According to a seventh aspect of the present invention, in the information search apparatus according to the first, second, third, fourth, fifth, or sixth aspect, the information retrieval apparatus stores compressed index information compressed by the index information compression means. Compressed index information storage means, index expansion means for reading out compressed index information in the compressed index information storage means based on a given search condition and expanding the original index information, and index information expanded by the index expansion means And a search means for performing a search based on.

According to a still further aspect of the present invention, there is provided an information compression method for an information retrieval apparatus having an index for specifying a specific character chain in document data by position information. An index information creating step of creating index information including the position information for each character chain that appears, and creating compressed position information obtained by compressing the position information for the index information of the same character chain according to the value of the position information Index information compression step.

According to a ninth aspect of the present invention, in the information compression method of the information retrieval apparatus according to the eighth aspect, the index information compression step includes the steps of: A plurality of pieces of index information having the same first position information included in the position information are taken as one group, and compressed index information is created by compressing the first position information of the group into one.

According to a tenth aspect of the present invention, in the information compressing method of the information searching apparatus according to the eighth or ninth aspect, the index information compressing step comprises the steps of: Alternatively, when the second position information included in the position information is equal to or smaller than a predetermined reference value, the compressed index information is used to create compressed position information obtained by compressing the second position information to a smaller number of bits or bytes. It is.

[0022] According to an eleventh aspect of the present invention, in the information compression method of the information retrieval apparatus according to the ninth aspect, the index information compression step includes the step of storing one piece of the first position information. About the compressed index information compressed to
Calculating a difference between the second position information included in the position information, and compressing the second position information into a difference represented by a smaller number of bits or bytes when the difference is equal to or smaller than a predetermined reference value. It is.

According to a twelfth aspect of the present invention, in the information compression method of the information retrieval apparatus according to the eighth, ninth, tenth, or eleventh aspect, the index information includes a plurality of hierarchical hierarchical information. It contains location information.

According to a thirteenth aspect of the present invention, in the information compression method of the information retrieval apparatus according to the ninth, tenth, or eleventh aspect, the first position information is used as document identification information. , The second location information to the first
Is the position information in the document specified by the position information.

Further, a computer-readable recording medium according to claim 14 is a computer-readable recording medium.
It is recorded as a program for causing a computer to execute the information compression method of the information search device described in 11, 12, or 13.

In the information retrieval apparatus, the information compression method of the information retrieval apparatus, and the recording medium according to the present invention, the index information creating means (index information creating step) includes an index including position information for each character chain appearing in the document data. Create information,
By index information compression means (index information compression step),
For index information of the same character chain, compressed index information is created by compressing the position information according to the value of the position information. Here, the position information included in the index information is, as in the information search device according to claim 5, the information compression method of the information search device according to claim 12, and the recording medium according to claim 14,
It is often hierarchical information. For example, as in the information search device according to claim 6, the information compression method of the information search device according to claim 13, and the recording medium according to claim 14,
Generally, the first position information is the identification information of the document, and the second position information is the position information in the document specified by the first position information. In addition, the hierarchical position information includes a section number of a section or a chapter of the document, a page of the document, a paragraph number of the document, and the like.

In particular, in the information retrieval apparatus according to the second aspect, the information compression method of the information retrieval apparatus according to the ninth aspect, and the recording medium according to the fourteenth aspect, the same index information compression means (index information compression step) is used. , The plurality of pieces of index information having the same first position information included in the position information are regarded as one group, and the first position information of the group is defined as 1
To create compressed index information. As described above, since the position information included in the index information is often hierarchical, it is easy to imagine a case where there are n pieces of index information having the same first position information in a higher hierarchy. Therefore, by creating the compressed index information in which the first position information is compressed into one by using the index information as one group, it is possible to reduce the number of n-1 pieces of the first position information, To reduce the size of the index file. If index information is created in the order of appearance of the character chain in the index information creating means (index information creating step), index information having the same first position information appears continuously. (Compression process) can be easily performed.

Further, in particular, an information retrieval apparatus according to claim 3,
In the information compression method of the information search device according to the tenth aspect, and the recording medium according to the fourteenth aspect, the index information compression means (index information compression step) uses the index information of the same character chain or the compressed When the second position information included in the second
Compressed index information is created by compressing the position information of into a smaller number of bits or bytes. For example, when the second position information is position information (appearance order) in the document specified by the first position information, the number of bits or the number of bytes is reduced for an item whose appearance order is equal to or less than a predetermined reference value. The index information can be compressed to reduce the capacity of the index file. If the index information creating means (index information creating step) creates the index information in the order of appearance of the character chain, the possibility that the index information whose second position information is equal to or less than the predetermined reference value continuously appears increases. Therefore, compression processing by the index information compression means (index information compression step) can be easily performed.

[0029] Further, in particular, an information retrieval apparatus according to claim 4,
In the information compression method of the information search device according to the eleventh aspect and the recording medium according to the fourteenth aspect, the index information compressing means (index information compression step) is configured to compress the first position information into one piece of the compressed index information. Calculating a difference between the second position information included in the position information, and when the difference is equal to or smaller than a predetermined reference value, compressing the second position information into a difference represented by a smaller number of bits or bytes. ing. For example, when the second position information is position information (appearance order) in the document specified by the first position information, the number of bits or bytes of those that appear close to each other and whose difference is equal to or less than a predetermined reference value is determined. The number can be reduced, and the index information can be compressed to reduce the capacity of the index file. If the index information creating means (index information creating step) creates the index information in the order of appearance of the character chain, there is a possibility that the index information whose difference between the second position information is equal to or less than a predetermined reference value appears continuously. , The compression processing by the index information compression means (index information compression step) can be easily performed.

By compressing the index information as described above, the index information (compressed index information) has a smaller data size than the uncompressed index information. The area, that is, the capacity of the index file can be reduced. Further, regarding the amount of data to be loaded from the recording medium to the memory when obtaining the same information, the compressed index information is smaller than the uncompressed index information, so that the load time to the memory can be reduced. Therefore, even when the compressed index information is held in a recording medium such as a CD-ROM having a low data reading speed, the reading time can be reduced and a high-speed search can be realized.

Further, in the information retrieval apparatus according to the present invention, the compressed index information storage means holds the compressed index information compressed by the index information compressing means, and the compressed index information based on the search condition given by the index decompression means. The compressed index information in the information storage unit is read and expanded to the original index information, and the search unit performs a search based on the index information expanded by the index expansion unit. Claims 1, 2, 3,
An information retrieval apparatus according to any one of claims 4, 5, and 6, wherein
Since the information compression method of the information retrieval device according to the present invention is simple, and the compression method performed on the recording medium according to the present invention is simple, the original index information is decompressed by the index decompression means. As a result, the index information can be compressed more efficiently, the capacity of the index file can be reduced, and the search time can be reduced by searching using the index file.

[0032]

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram showing an embodiment of an information retrieval apparatus, an information compression method of an information retrieval apparatus, and a recording medium according to the present invention. This will be described in detail with reference to FIG. In the description of each embodiment, the information search device according to the present invention and the information compression method of the information search device will be described in detail. However, the recording medium according to the present invention is provided with a program for executing the information compression method. The description is included in the following description of the information compression method of the information retrieval device because the recording medium is a recorded recording medium.

[First Embodiment] FIG. 1 is a block diagram of an information retrieval apparatus according to a first embodiment of the present invention. The information search apparatus 100 according to the present embodiment creates an index file using a chain of characters in advance, and reads out index information for each character chain from the index file at the time of search, thereby realizing a high-speed search. FIG.
8 is a configuration in which an index information compression unit 107 and an index decompression unit 110 are added to the configuration of the conventional example of FIG.

In FIG. 1, an information retrieval apparatus 100 according to this embodiment includes a terminal 101, a document input unit 102, a search condition input unit 103, a result output unit 104, an overall control unit 105, and an index creation unit (index information creation means). 106, an index information compression unit (index information compression unit) 107, a bibliographic entity registration unit 108,
Search unit (search means) 109, index expansion unit (index expansion unit) 110, document number storage unit 111, bibliography acquisition unit 11
2. It is provided with an entity acquisition unit 113, an index position storage unit 114, an index information storage unit (compressed index information storage unit) 115, a bibliographic information storage unit 116, and an entity information storage unit 117.

The index creating unit 106 creates an index for performing a high-speed full-text search from a large amount of document data. The index information compression unit 107 creates compressed index information by compressing the index created by the index creation unit 106. Index information storage unit 11
Reference numeral 5 stores the compressed index information created by the index information compression unit 107. The index position storage unit 114 stores the head position and length of the compressed index information in the index information storage unit 115 for each character chain. Further, the bibliographic entity registration unit 108 registers bibliographic information and entity information of the document. The bibliographic information storage unit 116 stores bibliographic information, and the entity information storage unit 117 stores entity information.

The index decompression unit 110 reads out the compressed index information from the index information storage unit 115 according to the input search condition and decompresses the original index. The search unit 109 performs a full-text search from the index expanded by the index expansion unit 110. The document number storage unit 111 stores a document that matches the search condition in the search unit 109. The bibliography acquisition unit 112 acquires a bibliography list corresponding to the search condition of the document number storage unit 111 from the bibliography information storage unit 116. The entity acquisition unit 113
The entity information corresponding to the document number in the document number storage unit 111 is acquired from the entity information storage unit 117. Further, the overall control unit 105 issues an instruction to each unit based on a request from the terminal 101 and returns a result to the terminal 101.

Next, the operation of the information retrieval apparatus 100 according to the present embodiment at the time of document registration and retrieval will be described. First, at the time of document registration, the format of a document input from the terminal 101 is converted by the document input unit 102 into a format that can be handled by the index creation unit 106, and then the entire control unit 10
Send to 5. Here, the format conversion is a format conversion of a document for creating an index, for example, separating bibliographic information from entity information of the document by extracting bibliographic information from the document.

Next, the overall control unit 105 sends the text part of the format-converted document data to the index creation unit 106. The index creation unit 106 creates index information for each character chain from the format-converted document data using a two-character character chain. For example, for document data of "substance and whole", four indexes of "substance", "body and", "to all" and "whole" are created.

The index information is a set of position information for specifying the position of a character chain in the document data. More specifically,
A document number for identifying a document and a position number for determining where a character chain is located in the document correspond to the position information, and each is 2-byte data.
For example, when the document data “substance and entirety” in the above example exists at the document number 1, the entry of the index “entire” is the 4-byte data “(0x00) (0x01) (0x01)
(0x02) "(where" 0x "indicates a hexadecimal number). That is, the document number is indicated by the data of the first byte and the second byte, and 3
The position number is indicated by the data in the byte and the fourth byte. The data in the first byte and the second byte of the position number indicate the order of appearance of the characters “all” and “body” in the document with the document number 1 respectively.

Next, the index information created by the index creating unit 106 is sent to the index information compressing unit 107. The index information compression unit 107 compresses the index information by a compression method described later. The compressed compressed index information is stored in the index information storage unit 115 to form an index file. At this time, for each character chain, head position information indicating the head position of the compressed index information corresponding to the character chain in the index file is associated with length information (number of bytes) indicating the data length of the compressed index information. And stored in the index position storage unit 114.

The overall control unit 105 also sends the format-converted document data to a bibliographic entity registration unit 108 for registering bibliographic information and entity information of the document. The bibliographic entity registration unit 108 stores only bibliographic information in the document data in the bibliographic information storage unit 116, and the entity information storage unit 117 stores only the entity information in the document data.

On the other hand, when performing a full-text search, the terminal 101
The search condition input from the search condition input unit 103 is transmitted to the search unit 109 via the overall control unit 105 using the search command. The search unit 109 creates character chain information of two characters from the search character string of the search condition, and sends the character chain information to the index expansion unit 110. In the index decompression unit 110, the index position storage unit 11
4 and obtains the head position information and length information associated with the character chain information, accesses the index information storage unit 115 using the obtained head position information and length information, And obtains the corresponding compressed index information from the index file. Then, the obtained compressed index information is expanded to the original index information, and the expanded index information is sent to the search unit 109.

Next, the search section 109 performs a full-text search by determining the continuity of the character chain from the index information, and stores the document number and the search formula number that match the search conditions in the document number storage section 11.
1 and sends the number of searches to the overall control unit 105. Then, the overall control unit 105 sends the number of searches to the result output unit 104, so that the number of searches is displayed on the terminal 101.

Further, when a list of bibliographic information or entity information is acquired, the search formula number input from the search condition input unit 103 using the terminal 101 is used as the general control unit 105.
Is transmitted to the bibliography acquisition unit 112 or the entity acquisition unit 113 via the
Acquires the document number corresponding to the search formula number stored in the document number storage unit 111. Next, when acquiring a bibliographic information list from the obtained document number, the bibliographic acquisition unit 112
Thus, the bibliographic information is read from the bibliographic information management unit 116 and sent to the overall control unit 105. When acquiring entity information from the obtained document number, the entity acquiring unit 113
, The entity information is read from the entity information storage unit 117 and sent to the overall control unit 105. And the overall control unit 1
05 sends the bibliographic information or the entity information to the result output unit 104, so that the terminal 101 displays the bibliographic information or the entity information.

Next, examples of the information compression method performed by the index information compression unit 107 of the information search device 100 of the present embodiment will be described (first example), (second example), (third example). The order will be described in detail with reference to the drawings.

Here, the data format of the compressed index information will be described with reference to FIG. FIG. 2 is an explanatory diagram showing one format of the compressed index information in the information search device 100 of the present embodiment. As shown in FIG.
00 is, for example, a 3-bit compression method block 20 that indicates the compression method used to compress the compression index information.
One, a 5-bit entry number block 202 indicating the number of entries included in the compression index information, and one entry of the compression index information form a group of the compression index information. That is, the compression index information includes a compression method block 201, an entry number block 202, and n
One (n ≧ 1) entry is defined as one group, and thereafter, the compressed index information of another group is continuously stored.

First Embodiment First, a method of compressing index information according to a first embodiment will be described with reference to FIGS. FIG. 3 is a flowchart illustrating a main routine of the index information compression method according to the first embodiment.
FIGS. 4, 5, and 6 are flowcharts for explaining a subroutine of a process for putting an entry in a queue, and FIG. 7 is a flowchart for explaining a subroutine of a queue output process.

First, in FIG. 3, in step S301, one entry (4-byte data) is read from the index created by the index creating unit 106. Next, step S30
In step 2, the next one entry (4-byte data) is read from the index. Next, in step S303, the process proceeds to a subroutine of a process of putting the entries shown in FIGS. 4, 5, and 6 into a queue. The process of putting this entry in the queue (step S303) is performed in step S304.
Read the entry (4-byte data) and execute step S
This is repeated until there are no more entries to read in 305.

In the subroutine of the process of placing an entry in a queue, two entries read consecutively from the previous time and this time are compared, and if all 4 bytes of the entry can be deleted, the entry is placed in a 4-byte queue. , Document number 2
If a byte can be deleted, it is placed in a 2-byte queue, and if it cannot be compressed, it is placed in a 0-byte queue. Less than,
This will be described in detail with reference to FIGS.

First, in step S401 in FIG. 4, the 4-byte data of the previous entry (read in step S301) and the current entry (read in step S302) are compared, and the 4-byte data of the two entries is compared. It is determined whether or not all the data is the same. In step S401, if all the 4-byte data of the two entries are the same, the process proceeds to step S421 in FIG. 6, and if not, the process proceeds to step S402.

Next, in step S402, the document numbers (2-byte data) of the previous entry and the current entry are compared to determine whether or not the document numbers are the same. In step S402, if the document numbers are the same, FIG.
To step S411, otherwise to step S403.

That is, by comparing two entries read consecutively from the previous time and the current time, if all the 4-byte data of the entries are the same, one of the entries can be deleted. Branching to a subroutine (FIG. 6) for performing various processes associated with the distribution of the document. If the two-byte data of the document number is the same, the document number of one entry can be deleted. The subroutine branches to a subroutine (FIG. 5) for performing various processes accompanying the above. If none of the above two conditions is satisfied, compression is impossible. Various processes will be performed.

First, various processes associated with the distribution to the 0-byte queue from step S403 will be described. Step S4
At 03, it is determined whether or not there is data in the 2-byte queue. In step S403, if there is no data, the process proceeds to step S405. If there is data, the process proceeds to step S40.
In step 4, the previous entry is placed in the 2-byte queue, and the 2-byte queue is output by a queue output processing subroutine (FIG. 7) described later.

Next, in step S405, it is determined whether or not there is data in the 4-byte queue. Step S40
In step 5, if there is no data, the flow advances to step S407 to put the previous entry in the queue for 0 bytes. If there is data, the flow advances to step S406 to put the previous entry in the queue for 4 bytes. Output the queue. Since the contents of the 4-byte queue are entries from which all 4-byte data can be deleted,
The output only clears the contents of the queue, but does not output the contents of the queue to the index file. Further, in step S408, steps S403 and S403
Regardless of the result of the determination at 405, if the current entry is the last, the current entry is put into the 0-byte queue, and the subroutine ends.

The algorithm of FIGS. 4, 5 and 6 compares the two entries read continuously from the previous and the present, and basically distributes the previous entries sequentially. I am taking it. Therefore, step S
If it is determined in steps 403 and S405 that there is data in the 2-byte queue or the 4-byte queue, the last entry when the distribution of entries to the 2-byte queue or the 4-byte queue occurs continuously is performed. Since it indicates that the previous entry corresponds, it is necessary to distribute the previous entry to a 2-byte queue or a 4-byte queue. Step S408
In this case, if the current entry is the last one, the entry is put in the 0-byte queue because basically the method of sequentially distributing the previous entry is adopted.

Next, in step S402 of FIG.
Performed when the 2-byte data of the document number is the same,
Various processes associated with distribution to the 2-byte queue will be described with reference to FIG. In step S411, it is determined whether there is data in the 0-byte queue. Step S411
In step S413, if there is no data, the process proceeds to step S413.
If there is data, the flow advances to step S412 to put the previous entry in the 0-byte queue and output the 0-byte queue. Since the contents of the 0-byte queue are entries that cannot be deleted, the contents of the queue are cleared by output, and the contents of the queue are indexed with compressed contents corresponding to the compression conditions of the table shown in FIG. Output to a file.

Next, in step S413, it is determined whether there is data in the 4-byte queue. Step S41
In 3, if there is no data, the flow proceeds to step S 415 to put the previous entry in the 2-byte queue. If there is data, the flow goes to step S 414 to put the previous entry in the 4-byte queue, Output the queue. Further, in step S416, regardless of the determination results in steps S411 and S413, if the current entry is the last, the current entry is put in the 2-byte queue, and the subroutine is terminated.

In the distribution to the two-byte queue in FIG. 5, if the two-byte data of the document number is the same, the previous entry must be distributed to the two-byte queue without fail. The processing of S415 is the same as that of FIG.
The processing contents are different from those of steps S403 to S407. In step S416, if the current entry is the last entry, the entry is placed in the 2-byte queue because basically, the method of sequentially distributing the previous entry is adopted.

Next, in step S401 of FIG.
FIG. 6 shows various processes performed when the 4-byte data of an entry is the same and is distributed to a 4-byte queue.
This will be described with reference to FIG. In step S421, it is determined whether there is data in the 0-byte queue. Step S4
In 21, if there is no data, the process proceeds to step S 423, and if there is data, the process proceeds to step S 422 to put the previous entry in the 0-byte queue and output the 0-byte queue.

In step S423, it is determined whether or not there is data in the 2-byte queue. In step S423, if there is no data, the process proceeds to step S425, and if there is data, the process proceeds to step S424 to put the previous entry in the 2-byte queue and to execute the 2-byte queue by a queue output processing subroutine (FIG. Is output.

Next, in step S425, the previous entry is put in the 4-byte queue, and furthermore, in step S426
Then, regardless of the determination results in steps S421 and S423, if the current entry is the last, the current entry is changed to 4
Queue the byte and end the subroutine.

The reason why the previous entry is distributed to the 0-byte queue, the 2-byte queue, or the 4-byte queue by performing the case classification based on the judgments in steps S421 and S423 is that in step S426
If the current entry is the last entry, the entry in the 4-byte queue is basically due to the method of sequentially distributing the previous entry.

As described above, the process of distributing entries to the 0-byte queue, the 2-byte queue or the 4-byte queue (step S303) is performed in step S305.
Is repeated until there are no more entries to be read, the process proceeds to step S306 to determine whether data is present in any of the 0-byte queue, the 2-byte queue, and the 4-byte queue. Step S3
In step 06, if there is no data in each queue, the compression process ends.
Outputs the byte queue, 2-byte queue or 4-byte queue. The output of the 2-byte queue is performed by a queue output processing subroutine (FIG. 7) described later.

Next, referring to the flowchart of FIG.
The queue output process will be described. First, step S7
In 01, one entry is read from the queue A with the queue for two bytes as the queue A. Next, step S70
In step 2, it is determined whether or not each byte data of the two bytes of the position number is equal to or less than “0x0f”. In step S702, each byte data is set to “0x0
f ”or less, the process proceeds to step S703, where“ 0x0
If it is larger than "f", the process proceeds to step S708.

In step S708, the entry read from the queue A is put into another queue B as a temporary. Next, in step S709, one more entry is read from the queue A. Next, in step S710, it is determined whether or not the queue A is empty. If the queue A is empty, the process proceeds to step S711; if not, the process proceeds to step S702.
Move on to In step S711, the compression method and the number of entries of the entries registered in the queue B are output to the index file. Next, in step S712, the entry is read from the queue B, and the position number is output to the index file without compressing the position number according to the compression conditions of the table shown in FIG. I do.

As described above, if each byte data of the position number is equal to or less than "0x0f", step S7 is executed.
03, and in Step S703, Step S701
Puts the entry read from the queue A into the queue B. Next, in step S704, one more entry is read from the queue A. Next, in step S705, it is determined whether or not the queue A is empty. If the queue A is empty, the process proceeds to step S706, and if not, the process proceeds to step S702.

In step S706, the compression method and the number of entries of the entries registered in the queue B are output to the index file. Next, in step S707, the queue B
, And compresses the position number with compressed contents according to the compression conditions of the table shown in FIG. 8 and outputs the compressed position number to an index file, thereby forming a group of compressed index information. After ending the subroutine of the queue output process after ending step S707 or S712 in this manner, FIG.
Alternatively, return to the subroutine of FIG.
Alternatively, the compression processing is terminated by returning to the main routine of FIG.

The table shown in FIG. 8 is based on the comparison of the entries and the value of the position number of the entry.
It shows the compression methods classified into eight types. Hereinafter, compression conditions and compression contents of each compression method will be described. In the figure, reference numeral 801 denotes a compression condition;
Indicates compression contents that match the compression condition 801.

First, in the compression method 0, when the previous entry is compared, the two-byte data of the document number is different from each other, and one of the two-byte data of the position number is "0x1".
The compression condition is “0” or more. This case is 0
The entries distributed to the byte queue correspond to the output, and all the entries are output without compression.

Next, in the compression method 1, the compression condition is that, when compared with the previous entry, the two-byte data of the document number is different from each other, and each byte data of the position number is both "0x0f" or less. In this case, the entries distributed to the 0-byte queue correspond to the above. However, since the position number can be compressed, 2-byte data of the document number is output, and the lower 4 bytes of each byte data of the 2 bytes of the position number are output. The data is converted into 1-byte data by combining bits and output.

Next, in the compression method 2, when the document number is the same as that of the previously output compressed index information group, and when compared with the previous entry, the 2-byte data of the document number is the same, Either of the byte data is "0x1
The compression condition is “0” or more. This case is 2
The entry distributed to the byte queue corresponds to the above. The 2-byte data of the document number is omitted because the document number of the group of the compressed index information previously output can be used, and only the 2-byte data of the position number is used. Output.

Next, in the compression method 3, when the document number is the same as that of the previously output compressed index information group, and when compared with the previous entry, the 2-byte data of the document number is the same, The compression condition is that each byte data is equal to or less than “0x0f”. In this case, the entry distributed to the 2-byte queue is applicable, and the 2-byte data of the document number is omitted because the document number of the group of the compressed index information previously output can be used, and the position number 2 is omitted. The data is converted into 1-byte data in which the lower 4 bits of each byte data are connected and output.

Next, in the compression method 4, when the document number is different from the group of the compression index information previously output, and when compared with the previous entry, the 2-byte data of the document number is the same and each of the position numbers is different. If either byte data is "0x1
The compression condition is “0” or more. This case is 2
The entries distributed to the byte queue correspond to the first entry, and the 4-byte data is output as it is for the first entry. For the second and subsequent entries, the 2-byte data of the document number uses the document number of the first entry. This is omitted because it is possible to output only 2-byte data of the position number.

Next, in the compression system 5, the document number differs from the group of the compression index information previously output, and when compared with the previous entry, the 2-byte data of the document number is the same and each of the position numbers is different. The compression condition is that both the byte data are “0x0f” or less. In this case, the entries distributed to the 2-byte queue correspond to the first entry, and output the 2-byte data of the document number only for the first entry. For each entry, the lower 4 bits of each byte data of the 2-byte position number Is converted to 1-byte data in which the data is connected and output.

Next, in the compression method 6, the compression condition is that the 2-byte data of the document number and the 2-byte data of the position number are the same when compared with the previous entry. In this case, only the first entry is distributed to the 0-byte queue, and the entry is output without performing compression.

Further, in the compression method 7, when compared with the previous entry, the two-byte data of the document number and the two-byte data of the position number are the same, and both the byte data of the position number are both "0x0f" or less. Is the compression condition. In this case, only the first entry is distributed to the 0-byte queue, but since the position number can be compressed, 2-byte data of the document number is output and the lower byte of each byte data of the 2 bytes of the position number is output. It is converted into 1-byte data in which four bits are connected and output.

Next, with reference to a specific example of index information shown in FIG. 9, a method of compressing index information according to the present embodiment will be specifically described. FIG. 9 is an explanatory diagram illustrating a specific compression process of the index information compression method according to the first embodiment. In the figure, reference numeral 901 denotes an entry (index information) to be processed, and reference numeral 902 denotes a compressible entry of the index information shown in 901. Note that in 902, the document numbers of the entries of the indexes k to k + 4 are all “(0x
27) (0x7b) ", and each byte data of the position number of the entry is all less than or equal to" 0x0f ". Also,
903 indicates an entry compressed for the position number,
904 indicates the resulting compressed index information.

In the following, a processing procedure when the index information 901 shown in FIG. 9 is applied to the index information compression method (the flowcharts shown in FIGS. 3 to 7) in this embodiment will be described with reference to FIG. Will be explained. First, the entry of the index k-1 is read in step S301 of FIG. 3, and the entry of the index k is read in step S302. Since these entries have different 4-byte data, in the subroutine of step S303,
Steps S401, S402, S403, S40 in FIG.
After following steps S5, S407, and S408, the process proceeds to step S304 of the main routine in FIG. Therefore, the index k
The entry of −1 is registered in the 0-byte queue as shown in FIG.

Next, the entry of the index k + 1 is read again in step S304. When the entries of the index k and the index k + 1 are compared, each 4-byte data is different but the document number is the same. Since the entry of index k-1 is registered in the queue, in the subroutine of step S303, steps S401 and S402 in FIG. 4 and step S41 in FIG.
1, S412, S415 and S416 are traced. Therefore, the entry of the index k-1 held in the 0-byte queue is output by the compression method 0, and the entry of the index k is registered in the 2-byte queue.

Next, the entry of the index k + 2 is read out again in step S304, and the process proceeds from step S305 to the subroutine of step S303 again. Index k +
Comparing the entries of 1 and the index k + 2, each 4-byte data is different but the document number is the same. Therefore, in this subroutine, steps S401 and S40 in FIG.
2, and steps S411, S413, S41 in FIG.
5 and S416. For this reason, the entry of the index k + 1 is also registered in the 2-byte queue similarly to the index k.

Next, the entry of the index k + 3 is set in step S
Although read at 304, by following the same steps as above, the entry with index k + 2 is also registered in the 2-byte queue. Then, the entry of the index k + 4 is read in step S304. By following the same steps as above, the entry of the index k + 3 is also registered in the 2-byte queue.

Next, the entry of the index k + 5 is set in step S
When the entries at index k + 4 and index k + 5 are compared, the 4-byte data is different. At this time, the index k to index k are stored in the 2-byte queue.
Since the entry of +3 is registered, step S30
In the subroutine No. 3, step S40 in FIG.
1, S402, S403 and S408 are traced. Therefore, after the entry of the index k + 4 is registered in the 2-byte queue, the entries of the indexes k to k + 4 registered in the 2-byte queue are output by the compression method 5.

Here, the output processing of the entries of the index k to the index k + 4 registered in the 2-byte queue will be described with reference to FIG. First, in step S701, the entry of the index k is read from the 2-byte queue (queue A). In the entry of the index k, since each byte data of the position number is equal to or less than “0x0f”, the process proceeds to step S504 and is output to the queue B. Thereafter, the index k + 1 to the index k +
In the entry No. 4 as well, each byte data of the position number is “0x”.
0f ", and is similarly output to the queue B. Next, in step S705, since the 2-byte queue is empty (queue A), the process proceeds to step S706.

Next, in step S706, since the compression condition of the compression method 5 in FIG. 8 is satisfied, “5” is output to the compression method block 904a of the index file, and the number of entries held in the queue B is “5”. "5" is output to the entry number block 904b. And step S70
7, the entry is read from the queue B, and the compression method 5
Compresses and outputs each entry.

As described above, the document number and position number of each entry are compressed, and the 3-bit compression method block 90 is used.
When the entry number block 904b of 4 bits and 5 bits is added, as shown in 904, a group of a total of 8 bytes of compressed index information is formed. That is, in this specific example,
20 when not compressing the entries from index k to index k + 4
The index information that required bytes could be reduced to 8-byte compressed index information by the index information compression method of the first embodiment.

As described above, in the index information compression method of the present embodiment, the index information compression step (FIG. 3, FIG. 4, FIG. 5,
According to FIGS. 6 and 7), when the entries of the index information are continuously the same, the other entries are deleted leaving only one entry, so that the index information is compressed to reduce the capacity of the index file. be able to. In addition, for index information of the same character chain, compressed index information is generated by compressing the document number into one by creating a group of n consecutive index information with the same document number. Data for the document number can be reduced, and the index information can be compressed to reduce the capacity of the index file. Further, when each byte data of the position number is equal to or smaller than the reference value (0x0f), compressed index information having a position number converted into 1-byte data by connecting lower 4 bits of each byte data of the position number is created. Therefore, the number of bytes can be reduced for characters whose appearance rank is equal to or less than the reference value, and the index information can be compressed to reduce the capacity of the index file.

(Second Embodiment) Next, FIGS.
A method of compressing index information according to the second embodiment will be described with reference to FIG. 11 and FIG.
10 is a flowchart illustrating a subroutine of a compression method of index information according to the second embodiment. Here, the main routine of the compression processing of the index information shown in FIG. 3 and the subroutine (FIGS. 4, 5 and 6) shown in step S303 are the same as those in the first embodiment, and therefore the description is omitted.

Hereinafter, the main routine (step S30) will be described with reference to the flowcharts of FIGS.
In the queue output of 7) and the subroutine (FIGS. 4 and 6), the output processing of the 2-byte queue which is executed by branching will be described. First, step S1 in FIG.
At 101, one entry is read from queue A, which is a 2-byte queue. Next, in step S1102, the process proceeds to step S1201 in FIG. 12, and it is determined whether each byte data of the position number is equal to or less than “0x0f”. In step S1201, if each byte data is equal to or less than “0x0f”, the process proceeds to step S1202, and if larger than “0x0f”, the process proceeds to step S1212.

In step S1202, it is determined whether or not the temporary queue C is empty. If the queue C is not empty, the process proceeds to step S1203. If the queue C is empty, the process directly proceeds to step S1204. .
In step S1203, the compression method (see FIG. 8) of the entries registered in the queue C and the number of entries are output to the index file, the entries are read from the queue C, and the position numbers are output to the index file without compressing the position numbers. I do.

Next, in step S1204, regardless of the determination result in step S1202, that is, regardless of whether the queue C is empty or not, step S1101 is executed.
Puts the entry read from the queue A into the queue B. Next, in step S1205, one more entry is read from the queue A. Then, step S12
At 06, it is determined whether or not the queue A is empty.
Is empty, the process proceeds to step S1103 shown in FIG.
If the queue A is not empty, the process returns to step S1201.

Also, as described above, step S120
1, each byte data of the position number is "0x0
If "f" is larger than "f", the process proceeds to step S1212. In step S1212, it is determined whether or not the queue B is empty. If the queue B is not empty, the process proceeds to step S1213. It proceeds to step S1214 as it is. In step S1213, the queue B
And outputs the compression method and the number of entries of the entries registered in the index file to the index file, reads the entries from the queue B, and outputs the position numbers to the index file without compressing the position numbers.

Next, in step S1214, the entry read from queue A in step S1101 is output to queue C regardless of the result of determination in step S1212, that is, regardless of whether queue C is empty. Next, in step S1215, one more entry is read from queue A. Then, step S12
At 16, it is determined whether or not the queue A is empty.
Is empty, the process proceeds to step S1103 shown in FIG.
If the queue A is not empty, the process returns to step S1211.

Next, in step S1103 of FIG.
It is determined whether or not the queue B is empty. If the queue B is empty, the process proceeds to step S1104. If the queue B is not empty, the process proceeds to step S1105. In step S1105, the compression method and the number of entries of the entries registered in the queue B are output to the index file, the entries are read from the queue B, the position numbers are compressed and output to the index file. 6 to continue the processing, or return to the main routine of FIG. 3 to end the compression processing.

In step S1104, the queue C
Is determined to be empty. If the queue C is not empty, step S1106 is executed. If the queue C is empty, the process returns to the subroutine of FIG. 4 or FIG. Alternatively, the compression processing is terminated by returning to the main routine of FIG. In step S1106, the compression method and the number of entries of the entries registered in the queue C are output to the index file, the entries are read from the queue C, the position numbers are compressed and output to the index file.

Next, referring to a specific example of index information shown in FIG. 13, a method of compressing index information according to the present embodiment will be specifically described. FIG. 13 is an explanatory diagram illustrating a specific compression process of the index information compression method according to the second embodiment. 9, 1301 denotes an entry (index information) to be processed as in FIG.
It shows a compressible entry in the index information shown in FIG. Reference numeral 1303 denotes an entry compressed with respect to the position number, and reference numeral 1304 denotes obtained compression index information.

The index information 1301 shown in FIG. 13 is used to compress the index information in this embodiment (FIGS. 3 to 6 and FIG. 11).
And the flowchart shown in FIG. 12),
In the index information, the document numbers of the entries of the index k to the index k + 4 are all the same as “(0x27) (0x7b)”. be able to. In addition, since each byte data of the position number of the entry of the index k to the index k + 2 is equal to or less than “0x0f”, by converting the lower 4 bits of each byte data of the position number into 1 byte data, one entry is converted. One byte can be reduced.

On the other hand, since each byte data of the position number of the entry of the index k + 3 and the index k + 4 is larger than “0x0f”, the position number cannot be compressed.
Since the document numbers of the entries of the index k + 4 are all the same, a group of two consecutive pieces of compressed index information is formed. That is, as shown in FIG.
Each entry of the index k + 2 is compressed by the compression method 5 with 3 entries and is a group of 6-byte compression index information, and each of the indexes k + 3 and k + 4 is compressed by the compression method 2 with 5 entries. Compression index information 1304 including a group of byte compression index information
Can be obtained. This allows 2
0-byte index information could be reduced to 11-byte compressed index information. In addition, it is possible to reduce the number of bytes by two in comparison with the compressed index information (13 bytes) in which the document number is compressed and the position number is not compressed for the five entries of the index k to the index k + 4.

As described above, in the index information compression method of the present embodiment, the index information compression step (FIG. 3, FIG. 4, FIG. 5,
According to FIGS. 6, 11 and 12), as in the first embodiment, when the entries of the index information are continuously the same, other entries are deleted except one entry, and With regard to the index information of the same character chain, compressed index information is created by compressing the document number into one by grouping n consecutive index information with the same document number as one group.
Data for n-1 document numbers can be reduced,
By compressing the index information, the capacity of the index file can be reduced. Further, when one byte data of the position number of a part of the index information is less than or equal to a reference value (0x0f) in a group of n consecutive index information having the same document number, a part of the index information is , The compressed index information having the position number converted into 1-byte data by connecting the lower 4 bits of each byte data of the position number is created. Can be reduced, and the index information can be compressed to reduce the capacity of the index file.

(Third Embodiment) Next, a method of compressing index information according to a third embodiment will be described with reference to FIGS. 3 to 6 and FIG. FIG. 14 is a flowchart illustrating a subroutine of a method for compressing index information according to the third embodiment. Here, the main routine of the compression processing of the index information shown in FIG. 3 and the subroutine (FIGS. 4, 5 and 6) shown in step S303 are the same as those in the first embodiment, and therefore the description is omitted.

In the following, with reference to the flowchart of FIG. 14, in the queue output of the main routine (step S307) and the subroutine (FIGS. 4 and 6), the output processing of the 2-byte queue which is executed in a branched manner will be described. I do. It should be noted that the flowchart of FIG. 14 is compared with the flowchart of FIG. 7 (first embodiment), and as is clear, steps S701 and S702 in FIG.
Between steps S1402 and S1403 before the repetition loop, and steps S140 and S1403 during the repetition loop.
4 and S1405 are respectively inserted.

In this embodiment, a group of n consecutive pieces of index information having the same document number is registered in the 2-byte queue, and branches off as queue A in FIG. , The difference between the position numbers is calculated, and each byte data of the difference is set to the reference value (0
x0f) or less, the 2-byte position number is converted to 1-byte data obtained by joining the lower 4 bits of each byte data of the difference and compressed.

First, in step S1402, it is determined whether or not the entry read from the queue A in step S1401 is the first entry. If the entry is the first entry, the process proceeds to step S1403, where one more entry is read from the queue A. Is read, and the process proceeds to step S1404. If it is not the first entry, the process proceeds to step S1.
Proceed to 404.

Next, in step S1404, the difference between the pieces of position information of the preceding and succeeding entries is obtained.
Proceeding to 05, the difference is replaced with the position information of the subsequent entry. Then, after step S1406, as in FIG. 7 (first embodiment), each byte data of the difference is set to “0x0”.
f ”or less, the 2-byte data of the position number is
The low-order 4 bits of each byte data of the difference are converted into 1-byte data by joining and compressed.

Next, referring to a specific example of index information shown in FIG. 15, a method of compressing index information according to the present embodiment will be specifically described. FIG. 15 is an explanatory diagram illustrating a specific compression process of the index information compression method according to the third embodiment. 9, 1501 indicates an entry (index information) to be processed, as in FIG. 9, and 1502 indicates 15 entries.
It shows a compressible entry in the index information shown in FIG. Reference numeral 1503 denotes an entry compressed by the difference in the position number, and reference numeral 1504 denotes compression index information obtained as a result.

When the compression method of the index information (the flowcharts shown in FIGS. 3 to 6 and FIG. 14) in this embodiment is applied to the index information 1501 of FIG. 15, the index k to the index k + 4 of the index information is obtained. Since the document numbers of all entries are the same as "(0x27) (0x7b)", compression can be performed by first omitting the document numbers of the entries at the index k + 1 to the index k + 4.

The byte data of the position number of the entry of the index k + 2 to the index k + 4 includes data larger than “0x0f”, but the 2-byte data of the position number of each entry of the index k + 1 to the index k + 4 is one. When the difference from the position number of the previous entry is obtained, all the byte data of the difference are equal to or less than “0x0f”. Therefore, the lower 4 bits of each byte data of the difference of the position number are connected, and
By converting the data into a difference of one byte data, each entry can be reduced by one byte. As a result, the 20-byte index information without compression can be reduced to the 8-byte compressed index information. In addition, compared with the compressed index information (13 bytes) in which the document number is compressed and the position number is not compressed for the five entries of the index k to the index k + 4, the number can be reduced by 5 bytes. (1)
(2 bytes) can be reduced by 4 bytes.

As described above, according to the index information compression method of the present embodiment, the index information compression step (FIG. 3, FIG. 4, FIG. 5,
According to FIGS. 6 and 14), similarly to the first embodiment, when the entries of the index information are continuously the same, the other entries are deleted except for one entry.
Regarding the index information of the same character chain, n consecutive index information having the same document number is regarded as one group, and the document number is set to 1
Since the compressed index information is created, the data for n-1 document numbers can be reduced, and the index information can be compressed to reduce the capacity of the index file. Further, for a group of n pieces of consecutive index information having the same document number, a difference between the position numbers of the position numbers of the index information is obtained, and when each byte data of the difference is equal to or smaller than the reference value (0x0f). Since the compression index information is created by converting the 2-byte position number into 1-byte data by connecting the lower 4 bits of each byte data of the difference, character chains appear close to each other and the difference between the position numbers is used as a reference. The number of bits or bytes can be reduced for those below the value,
By compressing the index information, the capacity of the index file can be reduced.

As described above, according to the information retrieval apparatus according to the first embodiment and the information compression methods according to the first, second and third examples of the information retrieval apparatus, the index information is compressed. Since the index information (compressed index information) has a smaller data size than the uncompressed index information, an area for holding the index information on the recording medium,
That is, the capacity of the index file can be reduced. Also,
Regarding the amount of data to be loaded into the memory from the recording medium when obtaining the same information, the compressed index information is smaller than the uncompressed index information, so that the load time to the memory can be reduced. Therefore, CD-RO
Even when the compressed index information is held in a recording medium having a low reading speed of data such as M, the reading time can be reduced and a high-speed search can be realized.

[Second Embodiment] FIG. 16 shows a second embodiment of the present invention.
It is a lineblock diagram of information search device 1600 concerning an embodiment. The information search device 1600 of the present embodiment differs from the information search device 100 of the first embodiment (see FIG. 1) in that a bibliographic information compression unit 1601 that compresses bibliographic information and an entity information compression unit that compresses entity information The configuration is such that a bibliographic expansion unit 1603 for expanding compressed bibliographic information and a substance expanding unit 1604 for expanding compressed substance information are added.

That is, the information search device 160 of the present embodiment
0 not only reduces the storage capacity of the index information by compressing the index file (index information storage unit 115), but also compresses the bibliographic information of the bibliographic information storage unit 116 and the entity information of the entity information storage unit 117. This also reduces the storage capacity of bibliographic information and entity information. Therefore, the bibliographic information storage unit 116 associates the compressed bibliographic information with the document number, the length information, the compression method information, and the head position information. Similarly, the entity information storage unit 117 stores the compressed bibliographic information. The document information, the length information, the compression method information, and the head position information are associated with the entity information.

The basic operation of the information search device 1600 of this embodiment is the same as that of the first embodiment, but differs from the information search device 100 of the first embodiment in the following points. That is, when the bibliographic entity registration unit 108 registers bibliographic information and entity information, the bibliographic information is compressed by the bibliographic compression unit 1601 and then stored in the bibliographic information storage unit 1.
16 and the entity information is stored in the entity compression unit 1602
Is stored in the entity information storage unit 117 after compression. When the bibliographic acquisition unit 112 acquires bibliographic information, the bibliographic decompression unit 1603 sets the bibliographic information storage unit 1
16 to read the compressed bibliographic information and expand the data in accordance with the compression method to obtain the original bibliographic information. Also, when the entity obtaining unit 113 obtains the entity information, the entity expanding unit 1604 Another difference is that the compressed entity information in the entity information storage unit 117 is read, and the original entity information is obtained by performing decompression according to the compression method.

Further, the information search device 1600 of the present embodiment
Also, in the same manner as in the first embodiment,
In addition, the search information compression method of the third embodiment can be applied. Further, as a method of expanding the compressed index information in the index expanding section 110, a procedure as shown in a flowchart of FIG. 17 is performed. First, step S
At 1701, 1-byte data including the data of the compression method and the number of entries is read from the compressed index information. Next, in step S1702, the compression method and the number of entries are extracted from the one-byte data. Next, step S1
In step 703, entries corresponding to the number of entries are read from the compressed index information, and the entries of the compressed index information are expanded according to the compression method. Further, in step S1704, it is determined whether or not the compressed index information is the end. If it is the end, the decompression process is ended. If not, the process returns to step S1701 to repeat the process.

In the information retrieval apparatus according to the first and second embodiments described above and the information compression methods according to the first, second and third embodiments of the information retrieval apparatus, the index information includes the document number and the position. Although constituted by numbers, other information may be used as long as the information has properties similar to these. For example,
A variant is also conceivable in which the document number is hierarchically and finely decomposed to include section numbers of sections, chapters, etc. of the document, page numbers of the document, paragraph numbers of the document, and the like. Also, index information 1
Although the data size of the entry is 4 bytes, the present invention is not limited to this, and may be configured with an arbitrary number of bits or bytes. In the data format of the compression index information shown in FIG.
And a 5-bit entry count block, constitutes a header that divides the group of compressed information. The number of bits of each block can be set arbitrarily, and a block for holding other information is included in the header. An additional configuration may be used. Further, the compression conditions and the contents of compression shown in FIG. 8 are examples of the compression method, and are not limited to the contents of FIG. Further, the information compression methods according to the first, second and third embodiments, and specific examples of each,
It is merely an example and is not limited to specific contents.

[0114]

As described above, according to the information retrieval apparatus, the information compression method of the information retrieval apparatus, and the recording medium of the present invention, the characters appearing in the document data are generated by the index information creating means (index information creating step). Index information including position information is created for each chain, and compressed index information obtained by compressing the position information according to the value of the position information for the index information of the same character chain by an index information compression unit (index information compression step) Is created, the index information can be more efficiently compressed, the capacity of the index file can be reduced, and a high-speed search can be performed using the index file.

According to the information retrieval apparatus, the information compression method of the information retrieval apparatus, and the recording medium of the present invention, the index information compression means (index information compression step) converts the index information of the same character chain into the position information. Since a plurality of pieces of index information having the same first position information are included in one group, compressed index information in which the first position information of the group is compressed into one is created. When there are n pieces of index information having the same position information of one, by creating these pieces of index information as a group and compressing the first position information into one, n-1 pieces of compressed index information are created. The first position information can be reduced, and the index information can be compressed to reduce the capacity of the index file.

Further, according to the information retrieval device, the information compression method of the information retrieval device, and the recording medium of the present invention, the index information compression means (index information compression step) uses the index information or compressed index information of the same character chain. When the second position information included in the position information is equal to or smaller than a predetermined reference value, compressed index information is created by compressing the second position information to a smaller number of bits or bytes. In the case where the second position information is position information (appearance order) in the document specified by the first position information, the number of bits or the number of bytes is reduced for those whose appearance order is equal to or less than a predetermined reference value. The index information can be compressed to reduce the capacity of the index file.

According to the information search device, the information compression method of the information search device, and the recording medium of the present invention, the first position information is compressed into one by the index information compression means (index information compression step). For the index information, a difference between the second position information included in the position information is obtained, and when the difference is equal to or less than a predetermined reference value, the second position information is represented by a difference represented by a smaller number of bits or bytes. Therefore, for example, when the second position information is position information (appearance order) in the document specified by the first position information, the second position information appears close and the difference is equal to or less than a predetermined reference value. The number of bits or the number of bytes can be reduced, and the index information can be compressed to reduce the capacity of the index file.

Further, according to the information search device of the present invention,
The compressed index information storage means holds the compressed index information compressed by the index information compressing means, and the index decompression means reads out the compressed index information in the compressed index information storage means based on the given search condition and retrieves the original index information. The information retrieval apparatus according to the present invention, the information compression method of the information retrieval apparatus, and the compression method performed on the recording medium are all simple. Therefore, the process of decompressing the original index information by the index decompression means can be easily performed. As a result, the index information can be more efficiently compressed to reduce the size of the index file, and Searching using a file can reduce the search time.

[Brief description of the drawings]

FIG. 1 is a configuration diagram illustrating an information search device according to an embodiment of the present invention.

FIG. 2 is an explanatory diagram showing one format of compressed index information in the information search device of the embodiment.

FIG. 3 is a flowchart illustrating a main routine of a method for compressing index information according to the first embodiment.

FIG. 4 is a flowchart (a process of queuing entries: 1) illustrating a subroutine of a compression method of index information according to the first embodiment.

FIG. 5 is a flowchart (a process of queuing an entry: part 2) for explaining a subroutine of a method of compressing index information according to the first embodiment;

FIG. 6 is a flowchart (a process of queuing an entry: 3) illustrating a subroutine of a compression method of index information according to the first embodiment.

FIG. 7 is a flowchart (queue output process) for explaining a subroutine of a compression method of index information according to the first embodiment.

FIG. 8 is an explanatory diagram showing compression methods classified into eight cases according to the comparison of entries and the value of the position number of the entry.

FIG. 9 is an explanatory diagram illustrating a specific compression process of a compression method of index information according to the first embodiment.

FIG. 10 is an explanatory diagram showing entries input to a 0-byte queue, a 2-byte queue, and a 4-byte queue in a specific example.

FIG. 11 is a flowchart illustrating a subroutine (part 1) of a method for compressing index information according to the second embodiment.

FIG. 12 is a flowchart illustrating a subroutine (part 2) of a method for compressing index information according to the second embodiment.

FIG. 13 is an explanatory diagram illustrating a specific compression process in a method of compressing index information according to the second embodiment.

FIG. 14 is a flowchart illustrating a subroutine of a method for compressing index information according to the third embodiment.

FIG. 15 is an explanatory diagram exemplifying a specific compression process in a compression method of index information according to the third embodiment.

FIG. 16 is a configuration diagram illustrating an information search device according to a second embodiment.

FIG. 17 is a flowchart illustrating a method of expanding compressed index information.

FIG. 18 is a configuration diagram showing a conventional information search device.

[Explanation of symbols]

 Reference Signs List 101 terminal 102 document input unit 103 search condition input unit 104 result output unit 105 overall control unit 106 index creation unit 107 index information compression unit 108 bibliographic entity registration unit 109 search unit 110 index decompression unit 111 document number storage unit 112 bibliography acquisition unit 113 Entity acquisition unit 114 Index position storage unit 115 Index information storage unit 116 Bibliographic information storage unit 117 Entity information storage unit 1601 Bibliographic information compression unit 1602 Entity information compression unit 1603 Bibliography decompression unit 1604 Entity decompression unit

Claims (14)

[Claims] 1. An information retrieval apparatus having an index for specifying a specific character chain in document data by position information, wherein an index for creating index information including the position information for each character chain appearing in the document data. An information retrieval apparatus comprising: an information creating unit; and index information compressing unit that creates compressed index information obtained by compressing position information of index information of the same character chain according to the value of the position information. 2. The index information compressing means, for index information of the same character chain, sets a plurality of pieces of index information having the same first position information included in the position information as one group.
2. The information retrieval apparatus according to claim 1, wherein compressed index information is created by compressing the position information into one. 3. The index information compressing means, when the second position information included in the position information is equal to or less than a predetermined reference value, for the index information or the compressed index information of the same character chain. 3. The information retrieval apparatus according to claim 1, wherein compressed index information is created by compressing the information to a smaller number of bits or bytes. 4. The index information compressing means obtains a difference between second position information included in the position information for the compressed index information obtained by compressing the first position information into one, and determines the difference as a predetermined reference value. The information search device according to claim 2, wherein when the value is equal to or less than the value, the second position information is compressed into a difference represented by a smaller number of bits or bytes. 5. The information retrieval apparatus according to claim 1, wherein the index information includes hierarchical position information. 6. The method according to claim 2, wherein the first position information is identification information of a document, and the second position information is position information in a document specified by the first position information. 3. The information search device according to 3 or 4. 7. A compressed index information storage means for holding compressed index information compressed by said index information compression means, and reads out the compressed index information in said compressed index information storage means based on a given search condition. 2. The apparatus according to claim 1, further comprising: index expansion means for expanding the index information; and search means for performing a search based on the index information expanded by the index expansion means.
The information search device according to 2, 3, 4, 5, or 6. 8. An information compression method of an information search device having an index for specifying a specific character chain in document data by position information, wherein index information including the position information is provided for each character chain appearing in the document data. And an index information compression step of creating compressed position information obtained by compressing the position information for the same character chain index information in accordance with the value of the position information. Information compression method. 9. The index information compressing step includes, for index information of the same character chain, a plurality of index information having the same first position information included in the position information as one group, and 9. The information compression method according to claim 8, wherein compressed index information is created by compressing the information into one. 10. The index information compressing step includes the step of, when the second position information included in the position information is equal to or less than a predetermined reference value, for the index information or the compressed index information of the same character chain. 10. The information compression method according to claim 8, wherein compressed position information is generated by compressing the information to a smaller number of bits or bytes. 11. The method according to claim 11, wherein the step of compressing the index information comprises:
Calculating a difference between the second position information included in the position information, and when the difference is equal to or smaller than a predetermined reference value, compressing the second position information into a difference represented by a smaller number of bits or bytes. 10. The information compression method according to claim 9, wherein: 12. The information compression method according to claim 8, wherein the index information includes hierarchical position information. 13. The method according to claim 9, wherein the first position information is identification information of a document, and the second position information is position information in a document specified by the first position information. 12. The information compression method according to 10 or 11. 14. A computer-readable recording medium recorded as a program for causing a computer to execute the information compression method according to claim 8, 9, 10, 11, 12, or 13.