JPH08161357A - Document control system - Google Patents

Abstract

PURPOSE: To construct an efficient document control system by changing the constitution of a character component table depending on the purpose. CONSTITUTION: A document register means 2 register-operates a registering document 1 to a document data base 9. The document data base 9 includes document main sentence data 8, the character component table 7 and a character component table entry designation 6. The character component table 7 is a table recording the presence/absence of information extracted from each character and a character string included in the registering document 1 for each document. The document register means 2 stored the registering document 1 and registers/records a value calculated from each character and consecutive two characters included in the registering document 1 in the character component table 7. A document retrieving means 4 quickly finds out a document 5 corresponding to a retrieving condition 3 through the use of the character component table 7. Then, the character component table entry designation 6 is capable of changing the constitution of the character component table 7.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a document management apparatus,
More specifically, the present invention relates to a document management device that searches for character strings in all documents using a character component table.
It can be applied to a document management system, an image management system, a database management system, and the like.

[0002]

2. Description of the Related Art A known document describing a conventional document management apparatus is, for example, Japanese Patent Laid-Open No. 5-324722. According to this publication, in order to reduce the size of the character component table to be used in the character string search, increase the degree of search, and enable high-speed document registration, the search character string input to the input unit is processed. The document search processing unit searches for a document that is considered to include a character string by using the character component table of the data unit. The document data corresponding to the retrieved document is output to the output unit by the document output process. In the document registration process, the document to be registered is registered in the document data, the character component is extracted from the document data, and registered in the character component table. When searching for a search character string from a document, the character component table indicates whether or not a character component exists in the document, and whether or not a certain bit string extracted from an adjacent character exists in the document. A document is searched by using the adjacent character component table indicating. That is, it is a system that searches documents at high speed using a character component table that records characters that have appeared for each document.

Further, the above-mentioned Japanese Patent Laid-Open No. 5-324722 and the previously proposed Japanese Patent Application No. 5-298433 propose a single character component table which indicates whether or not a character component is present in a document. , And a document is searched by using the adjacent character component table that indicates whether or not a certain character component extracted from the adjacent character exists in the document. Furthermore, the character component table fixes variable-length character entries. It divides into long blocks, stores them in secondary storage, and performs processing to combine small blocks into large blocks in order to speed up search processing.

[0004]

However, the methods described in the above publications have the following problems. ． The composition of the character composition table was fixed. ． Search time increases as the search string lengthens. ． Only a single character string can be specified as a search condition. Therefore, it is not possible to search for a document that satisfies a condition in which a plurality of character strings are combined by logical operators (AND, OR). ． Since the size (block size) of the block storing the bitmap data of the character component table is fixed, the search speed decreases when the block size is reduced, and the registration speed decreases when the block size is increased. ． There is no function to register multiple documents at once, and it takes processing time to register a large number of documents. ． Since the document data is stored in the system, it is troublesome for the user to refer to it, and the processes such as document registration, deletion, and update are troublesome.

The present invention has been made in view of the above circumstances, and it is possible to change the structure of the character component table according to the purpose, unify and optimize duplicate entries, and Include three or more characters in the entry of the character component table, support logical operators, optimize according to logical operators, prepare two types of block sizes, conventional character components In the table, the data blocks are small and distributed over the secondary storage, which means that a wide range of secondary storage is accessed, and the search speed is slow. There was a lot of access to the site, which caused a slowdown in search speed. To improve it,
Improving the registration processing speed of a large number of documents, facilitating the reference of registered documents by users and other applications, facilitating the document designation when registering a plurality of documents, and document registration, It is an object of the present invention to provide a document management device that automates the document management of the device when it is updated or deleted.

[0006]

In order to achieve the above-mentioned object, the present invention (1) stores a registered document, and a value calculated from each character or two consecutive characters contained in the registered document ( In a document management device having a document registration unit for registering and recording (character component) in a character component table and a document searching unit for rapidly searching for a document satisfying a search condition using the character component table, a character component table entry designation is made. To change the composition of the character component table by
(2) In (1), the registered document can be divided into a plurality of folders and managed, and a character component table entry can be designated for each folder; or (3) In (1), the document search means. When there are two or more identical character components extracted from a single character string, the character component in the character component table is accessed only once, or (4) In (1) above. As the calculated value of the character component, a value calculated from a continuous character string of three characters or more is also used, or (5) above (4)
In the case of using the value calculated from a continuous character string of three or more characters in the calculation of the character component at the time of document registration, the value calculated from one character or two continuous characters from the character string is used as the character component. Do not register in the table, or
(6) In the above (1), the document search means can process a search condition in which a single character string or a plurality of character strings are combined by an AND logical operator or an OR logical operator, or ( 7) In (6), the same character component extracted from the two character strings combined by the AND logical operator in the document search means is 2
If there are more than one, the character component in the character component table is accessed only once, or (8) above (6)
In the above, if there are two or more identical character components extracted from the two character strings combined by the OR logical operator in the document search means, the character component in the character component table is accessed only once. Or (9) In (1), the character component table is a data file composed of two types of blocks, large and small, for storing the character component table, and an index file for recording the block position for each character component. Or a block merging means for storing a plurality of small blocks into a large block, or (10) in (9), the block merging means groups a large block in front of the data file and a small block in the rear. Grouping blocks, or (11) in the block merging means in (10) above. , Subjecting only the area of the data file having a small block to the block fusion, or (12) in the above (9), the block fusion means rapidly stores the character component table entry from the secondary storage at the time of retrieval. When reading each character component table entry divided into multiple fixed-length blocks into a large fixed-length block for reading, the large fixed-length block is used as the unit of allocation, and the combined large fixed-length block and the rest that cannot be combined By writing out large fixed-length blocks packed with small fixed-length blocks in sequence, the data of the character component table can be processed in one scan, and the secondary storage area required at high speed can be minimized. Or, (13) Retrieval input from the input device, which holds a large amount of document data A document management device for searching a document including a character string and outputting the searched document by an output device, wherein a bit string extracted from each character code component (single character component) and two or more adjacent characters from the document at the time of document registration A component (adjacent character component) is extracted, a single character component table indicating whether or not each document includes each component, and one or more adjacent character component tables are generated, and each variable-length character of the character component table is generated. The entropy of the component is divided into a plurality of fixed-length blocks and registered in the secondary memory. At the time of search, a single character component and two or more adjacent character components are extracted from the search character string, and the adjacent character components are extracted. In a document management device that integrates fixed-length blocks in secondary storage corresponding to the above, generates each character entry, and retrieves a document, the bit in which a character component appears in a character entry is set to 1 and the bit that does not appear is If 0 is set to 0, a target character entry is appropriately selected when AND processing of the bitmaps of a plurality of character entries is performed at the time of search, and the bit string is scanned to change other By performing a full-text search with a minimum access to the bitmap of character entries, by checking the corresponding bits of the entries, and by using the document in which the value of the corresponding bits of all the entries is 1 as the search result, or (14) In the above (13), the number of 1 bits appearing in each character entry (the number of appearances of bits) is recorded in the secondary storage in advance for each character entry, and the bitmap of a plurality of character entries is searched at the time of retrieval. When the AND process is performed, the character entry with the smallest number of occurrences of the bit is selected from the target character entries, the bit string is scanned, and the value is 1. By checking the corresponding bits of other character entries in the ascending order of the number of bit appearances, and finding the documents in which the value of the corresponding bit of all character entries is 1, the access to the bitmap of character entries can be performed. Do a full-text search with a minimum, or
(15) In (13), the bitmap data of a variable-length character entry is converted into a plurality of fixed-length blocks so that it is not necessary to read all blocks of the character entry when accessing a part of the character entry. It is divided into two parts and stored in the secondary storage, and a block table that manages each fixed-length block is provided in the secondary storage, and an arbitrary fixed-length block can be accessed from the block table. A full-text search is performed by reducing access to a block on the secondary storage that is not present, or (16) a value calculated from each character or two consecutive characters included in the registered document while saving the registered document ( A document registration unit for registering and recording (character component) in the character component table, and a document search unit for quickly searching for a document satisfying the search condition using the character component table. And the configuration of the character component table can be changed by designating the character component table entry, and the configuration of the character component table records each character of the document and a value calculated from two consecutive characters. The character search means extracts only a value calculated from two consecutive characters from the search character string, or (17) the above (16)
In the above, the document search means extracts a value calculated from two consecutive characters from the search character string and a value calculated from one character at the end of the search character string, or (1
8) In (16), the document search means extracts a value calculated from two consecutive characters from the search character string and a value calculated from the first character of the search character string, or (19) In the case of (18), in the case where a value (character string entry) calculated from a character string of three or more consecutive characters is also used in the configuration of the character component table, the document search means uses the search character string. When the character entry is extracted, a value calculated from one character included in the character string corresponding to the character string entry or two characters included in the character string including the characters before and after the character string is not extracted. Or (20) In (16), the document retrieval means combines a single character string or a plurality of character strings with a "logical product", a "logical sum", or a "logical difference" logical operator. conditions Can be processed, or (2
1) In the above (20), in the processing of the two character strings combined by the "logical difference" by the document search means, the character string on the rear side is not processed, or (22) The above (1)
In the case of registering a large number of documents at once, the document registration means does not directly write the character component table data generated each time one document is registered to the character component table on the secondary storage, but once. The document is registered in the memory, and then the character component table data in the memory is collectively written in the character component table of the secondary storage to perform the batch registration of documents at high speed, or (23) (1) above. When the directory path name of the registered document on the file system registered by the document registration means is managed on the secondary storage and the document contents are referred to, the file path directly on the file system is registered based on the registered directory path name. Not only does the document management system not need to have the original data internally by referencing the original data of To facilitate the reference of the registration document,
Alternatively, (24) in (1), the directory path name of the original data of the registered document on the file system registered by the document registration means is managed, and when referring to the document contents, the registered directory path name Directly refer to the original data on the file system based on, specify the directory when registering the document,
By registering all the documents in the directory and its subordinate directories, the document designation at the time of registering a plurality of documents can be facilitated, or (25) above (1)
In managing the directory path name of the original data of the registered document on the file system registered by the document registration means, and referring to the document contents, the original data on the file system is directly copied based on the registered directory path name. The directory to be registered is specified in advance so that the registration, update, and deletion of documents in that directory are always monitored, and when a document is registered, updated, or deleted, it is automatically In addition, by registering, updating, and deleting the document with respect to the character component table, the user's time and effort for document management is reduced.

[0007]

The document management apparatus according to the present invention includes (1) document registration in which a registered document is stored and a value calculated from each character or two consecutive characters included in the registered document is registered and recorded in the character component table. And a document search unit that searches for a document that meets the search condition at high speed using the character component table, and makes it possible to change the configuration of the character component table by specifying the character component table entry. Can be managed by dividing it into multiple folders, and the character component table entry can be specified for each folder, and the structure of the character component table can be changed according to the purpose, so an efficient document management system can be constructed. .

(2) When there are two or more identical character components extracted from a single character string by the document retrieval means, the character component in the character component table is accessed only once. In addition, a value calculated from a continuous character string of three or more characters is used in the calculation of the character component, and further, a value calculated from the continuous character string of three or more characters is calculated in the calculation of the character component at the time of document registration. When a value that is used is used, the value calculated from one character or two consecutive characters from the character string is not registered in the character component table, so that even if the search character string is long, the document can be searched at high speed. .

(3) The document searching means can process a search condition in which a single character string or a plurality of character strings are combined by AND or OR logical operators, and
When there are two or more identical character components extracted from the two character strings combined by the AND logical operator in the document retrieval means, the character component in the character component table is accessed only once. Further, when there are two or more identical character components extracted from the two character strings combined by the OR logical operator in the document retrieval means, the character component in the character component table is accessed only once. Since the search condition is not included, a plurality of character strings can be expressed as a combination by logical operators (AND, OR), so that a complicated search request can be expressed. In addition, since optimization is performed according to the logical operator, documents can be searched at high speed.

(4) The character component table is stored by a data file composed of two types of blocks, large and small, for storing the character component table and an index file for recording the block position for each character component. A block merging unit for merging blocks into a large block, wherein the block merging unit assembles large blocks in front of the data file and assembles small blocks rearward of the data file; Since only the area having a small block is targeted for block fusion, and two sizes, large and small, are prepared for the size of the block that stores the bitmap data of the character component table, both the registration / search speed can be increased.

(5) In the block merging means, the character component table entries divided into a plurality of fixed length blocks are grouped into a large fixed length block in order to read the character component table entries from the secondary storage at high speed during retrieval. Sometimes a large fixed-length block is used as a unit of allocation, and a large fixed-length block packed together and a large fixed-length block packed with the remaining small fixed-length blocks that could not be combined are written out in sequence, thereby writing the data of the character component table. Process in a single scan, speedily and minimize the secondary storage area required during processing,
Search speed is improved by combining small blocks of character entries into large blocks.

(6) A bit of a specific bitmap is sequentially examined in the horizontal direction, and when the bit is 1, the corresponding bit of another bitmap entry is examined. That is, the bits are examined in the vertical direction, and if all the bits are 1, the document corresponding to the bits is the document as the search result. Further, the number of appearances of bit 1 appearing in each character entry is counted in advance, the bits are rearranged in the ascending order of the number of appearances of the bits at the time of the above-described processing for checking the bits, and the same processing is performed, so that the data to be referred to further. It is possible to reduce the amount. Further, when only a part of the character entry is required, the corresponding block can be directly obtained from the block table without accessing the entire block, and the high speed search can be performed. As described above, in the related art, the character entry is frequently accessed and the search speed is lowered. However, the search speed is improved by changing the algorithm and the data structure of the processing at the time of search.

(7) When the composition of the character component table is such that each character of the document and a value calculated from two consecutive characters are recorded, the character searching means calculates from two consecutive characters from the search character string. And a value calculated from two consecutive characters from the search character string and a value calculated from one character at the end of the search character string, and A value calculated from two consecutive characters from the search character string and a value calculated from one character at the beginning of the search character string are extracted, and further the character string of the character component table is composed of three or more consecutive character strings. When using the value calculated from, when the document search unit extracts the character entry from the search character string, one character included in the character string corresponding to the character string entry or the character before and after the character is included in the character string. Including characters Since so as not to extract the value calculated from the 2 characters in string, reducing the number of entries to be extracted from the search string in the search process can speed up the search process.

(8) The document search means can process a search condition in which a single character string or a plurality of character strings are combined by "logical product", "logical sum" or "logical difference" logical operators, and In the processing of two character strings combined by "logical difference" in the document search means, the latter character string is not processed, so that the search condition is a logical operator (AND, OR) of a plurality of character strings. , NOT) can be expressed as a combination, so that a complicated search request can be expressed. In addition, since optimization is performed according to the logical operator, documents can be searched at high speed.

(9) When a large number of documents are registered at once, the character component table data generated every time one document is registered is not directly written in the character component table on the secondary storage, but a large number of documents are registered. In the conventional method, when the document is stored in the secondary storage, the secondary storage is accessed every time one document is registered, and the speed of the document registration is low. Therefore, for documents to be registered in batch, temporarily generate and register the character component table in the memory,
After that, at the end of the processing, the character component table data on the memory is appended to the document component table data on the secondary storage. By doing so, the access to the secondary storage is reduced, and the batch registration of a plurality of documents can be performed at high speed.

(10) There is no document data in the system of the document management device, and instead, as one of the document information, a directory path name indicating the position of the original document on the file system is managed. . For reference, since the original document on the file system is directly referenced based on the directory path name, the system of the device does not need to have the original data internally, and the secondary storage is not wasted. In addition, the registered document can be referred to by the user or another application without going through the system.

(11) When a document is registered, a directory is designated, and all the documents in the directory and its subordinate directories are registered, so that the directory path name of the original data of the registered document on the file system is registered in the document management apparatus. to manage. When referring to the document contents, the original data on the file system may be directly referred to based on the registered directory path name. In this way, when the original document is placed on the file system that the user usually uses, it becomes possible to manage the document existing in one directory hierarchy as it is by the document management device. In addition, by specifying a directory, you can automatically register all the documents included in that directory or the subordinate directories, so that all the documents in the directory can be registered individually. It is possible to reduce the burden on the conventional user, which had to be designated by one user.

(12) The document management device manages the directory path name of the original data of the registered document on the file system. When referring to the document contents, the original data on the file system is directly referenced based on the registered directory path name, and the user specifies in advance the directory in which the document to be registered in the relevant document management device is placed. Then, the system of the device is
The registration, update, and deletion of documents in that directory are constantly monitored, and when registration, update, and deletion of documents are performed,
The same operation is automatically reflected in the character component table, and registration, update and deletion are performed. By doing so, it is possible to reduce the burden of the document operation on the user.

[0019]

Embodiments will be described below with reference to the drawings. FIG. 1 is a block diagram for explaining an embodiment (claim 1) of a document management apparatus according to the present invention. In the figure, 1 is a registered document, 2 is a document registration means, 3 is a search condition, and 4 is a document. Search means, 5 is the relevant document, 6 is a character component table entry designation,
Reference numeral 7 is a character component table, 8 is document text data, and 9 is a document database.

The document registration means 2 performs the registration operation of the registered document 1 in the document database 9. The document database 9 includes document body data 8, a character component table 7, and a character component table entry designation 6. The character component table 7 is a table in which the presence or absence of information extracted from each character or character string included in the registered document 1 is recorded for each document. The document registration means 2 stores the registered document 1 and also registers and records a value calculated from each character or two consecutive characters included in the registered document 1 in the character component table 7. The document search means 4 uses the character component table 7 to quickly find the document 5 that satisfies the search condition 3. The configuration of the character component table 7 can be changed by the character component table entry designation 6.

FIG. 2 is a diagram showing an example of the character component table. In the character component table shown here, only the appearance of each character is recorded. In this method, a function is applied to the code of each character, and the calculated value is used as an entry. Is). An entry calculated from such one character is called a single character entry.

FIG. 3 is a diagram showing another example of the character component table. In the character component table shown here, two consecutive characters
An entry is a value obtained by bit-connecting the lower 4 bits of each character code from a character. For example,
The JIS codes for "gu", "da", and "ba" are 0x2430 and 0x, respectively.
2440, 0x2450, and if 8 bits obtained by concatenating the lower 4 bits are used as an entry in the character component table, "gugu", "guda", "guba" ... Are all grouped into the same 0x00 entry. That is, it is possible to apply a function to a code of two consecutive characters and use the calculated value as an entry (for the character x on the front side and the character y on the rear side, the function g
The value of (x, y) is an entry). An entry (character component) calculated from two consecutive characters is called an adjacent character entry. The character component table of FIG. 3 is a combination of single character entries and adjacent character entries.

As described above, the character component table can have various configurations, and in the present invention, the configuration of the character component table 7 can be designated by the character component table entry designation 6 in FIG. In the following description, for the sake of simplicity, a character component table in which only the appearance of each character is recorded as shown in FIG. 2 is used.

The document registration procedure is as follows. ． The registered document 1 is registered in the document body data 8. ． The contents of the registration document 1 are registered in the character component table 7. The entry specified by the character component table entry designation 6 is extracted from the document body. When the registered document number is i and the extracted entry number is j, the value of point (i, j) in the character component table is set to "1" for all j.

The document search procedure is as follows. ． A document number that may include a search character string is obtained using the character component table 7. (a) The entry defined by the character component table entry designation 6 is extracted from the search character string. (b) The extracted entry bitmap (one horizontal row in FIG. 2) is extracted from the character component table, and bit AND is performed. ． The document body of the document number obtained above is read from the document body data 8 and it is checked whether or not the search character string is included, and the included document set is used as the search result.

The retrieval accuracy of the character component table (the ratio of the retrieval character string included in the document obtained by using the character component table) depends on the structure of the character component table. In the present invention, the configuration of the character component table can be arbitrarily changed by designating the character component table entry. Therefore, an efficient document management system can be constructed according to the registered document.

Next, the invention described in claim 2 will be described. FIG. 4 is a block diagram for explaining another embodiment (claim 2) of the document management apparatus according to the present invention.
Reference numeral 9-n is a document database, and other parts having the same functions as those in FIG. The difference from the configuration of FIG. 1 is that there are many document databases 9-1 to 9-n. Since documents have various uses, it is desirable to store different document sets in different document databases. At that time, different document sets have different document lengths, character appearance frequencies, and the like. Therefore, in the document management apparatus of the present invention, since different character component table entry designations 6 can be used for each document database 9, efficient document management can be performed.

Next, the invention described in claim 3 will be described. With the conventional method, as the search character string becomes longer, the number of entries to be accessed in the character component table increases, so the search speed decreases. In fact, the same character component table entry calculated from the search word may include the same item. In that case, it is not necessary to access that entry multiple times, so by suppressing the access to the entry that appears multiple times in the search term at one time, the number of accesses to the character component table required for the search can be reduced and the search can be performed. Can speed up.

For example, since the search word "man-machine system" is composed of 9 characters, "ma" is displayed in the character component table.
You need to access 9 times, "n", "ma", "shi", "n", "shi", "su", "te" and "mu". However, in reality, "ma" and "n"
Since “shi” appears twice each, access to these entries can be combined into one. That is,
The actual access to the character composition table is "ma""n""shi"
You only need 6 times of "su", "te" and "mu".

Next, the invention described in claim 4 will be described. With the conventional method, the entry of the character component table is composed of only a maximum of two characters. On the other hand, if a long character string of 3 characters or more (a value calculated from) is used for the entry, the number of accesses to the character component table is reduced,
Search can be significantly speeded up. FIG. 5 is a diagram showing a character component table having a long character string as an entry. "System", "pattern", etc. are character string entries. As the character string entry, a character string having a high appearance frequency in the document may be selected.

At the time of registration, from "... that man-machine system ...", characters "a", "no", "ma", "n", "shi" are displayed as characters.
“Su”, “Te”, “Mu”, “Ha” and “System” are extracted as a character string and recorded in the character component table. At the time of search, the search word "man-machine system" begins with "ma" as a character.
"N", "ma", "shi", "n", "system" is extracted as a character string, and "shi" and single character duplicates included in "system" are deleted. After all, it is only necessary to access the three entries of "ma", "n", and "system", and the search time can be greatly reduced.

Next, the invention described in claim 5 will be described. In the method described in claim 4, the character entry included in the character string entry is also extracted at the time of document registration and is recorded in the character component table. However, since that part is often included as a character string even in a normal search character string, it is not always necessary to record it in the character component table. By not registering such a character entry, the character component table can be downsized.

For example, when registering the example sentence "... that man-machine system ..." in the preceding paragraph, only "a", "no", "ma", "n" and "ha" need to be registered as characters (as a character string " The system is extracted and recorded in the character component table.) However, if the search string contains a substring of the string entry, this method may cause a missed search.
For example, if the search string is "system"("system"
Substring), this method cannot be searched.

Next, the invention according to claim 6 will be described. In this embodiment, a combination of a plurality of character strings by logical operators (AND, OR) is accepted as a search condition (a single character string is also included in this search condition). Here, “AND” means to search for a document that includes both the character strings before and after, and “OR” means to search for a document that includes at least one character string before and after. Furthermore, if necessary, in order to clarify the order of operation of operators, use “(”, “)”
Can be used. By allowing a logical operator to be used as a search condition, a complicated search request can be expressed. For example, "man-machine system", "document search AND document registration", "document search OR information search", "(newspaper OR magazine) AND color" are the above-mentioned search conditions.

Next, the invention according to claim 7 will be described. The invention according to claim 3 proposes unification of access within a single search character string, but here, two or more search character strings connected by a logical operator AND are combined. Introduce unification of straddled access. For example, from the search condition “document search AND document registration”, in the conventional method, “sentence”, “writing”, “inspection”, “search”, “sentence”, “writing”, “registration”, “record”
Access the 8 entries. On the other hand, by unifying (optimizing) this item, duplication of “sentence” and “writing” is deleted, and the access to the character component table can be reduced to 6 times.

Next, the invention described in claim 8 will be described. The invention according to claim 3 proposes unification of access within a single search character string. However, here, two or more search character strings connected by a logical operator OR are combined. Introduce unification of straddled access. For example, from the search condition “document search OR information search”, in the conventional method, eight entries of “sentence”, “writing”, “check”, “search”, “information”, “report”, “check”, and “search” are accessed. On the other hand, by unifying (optimizing) this item, the duplication of “check” and “search” is deleted, and the access to the character component table can be reduced to 6 times.

Next, the invention described in claim 9 will be described. The character component table is saved as a file. The composition of the character component table file is desired to be able to easily access the bitmap data corresponding to the entries of the character component table. To realize this, for example, a bitmap composed of an index file and fixed length blocks is used. It can consist of two files, a data file. In this case, the index file can consist of blocks containing the following two fields:・ Start block offset field ・ End block offset field

The number of blocks included in the file is determined by the character component table entry designation. The bitmap data file is composed of blocks including the following two fields.・ Next block offset field ・ Data field

The block size may be set in the range of several tens to several kilobytes according to the performance requirement. Figure 6
(A) is a figure which shows an example of the file structure for a character component table. Note that loading the index file on the semiconductor memory is effective for speeding up.

The block size of the bitmap data file has a great influence on the registration / search performance. If the block size is large, the search is fast but the registration is slow,
If small, registration is fast but search is slow. In addition, there are two types of areas of the data file that are not used for recording bitmap data, namely, a large block and a small block. Below, a small block is called a "bucket" and a large block is called a "container", and the size ratio between the container and the bucket is "M".
Will be written. The size of the container is about several times to several tens of times that of the bucket (M = several times to several tens).

FIG. 6B is a diagram showing an example of the file structure of the character component table when blocks of two different sizes are introduced. Here, the most significant bit of the block offset is "1" or "0" to indicate whether the block at the offset position is a container or a bucket.

At the start of using the document search system, the block size is set to be small and the registration speed is prioritized. (While the number of registered documents is small, the search time is small even if the search speed is a little slow, so there is almost no problem.) . When a large number of documents are registered and the number of blocks included in the bitmap data file increases, block fusion processing is performed to collect a plurality of buckets in a container. This is because, in a normal operating system, it is faster to group data into larger blocks than to divide the data into smaller blocks. As a result, the block merging process improves the search speed, and the search time can be shortened even when the number of registered documents is large.

Block fusion processing procedure . Create a temporary file for writing. ． The following processing is performed on the bitmap data of all entries that make up the character component table. (a) Write the container as it is to a temporary file. (b) The (M) buckets that are put together in a container are set as one container and written to a temporary file. (c) Buckets that cannot be collected in the remaining containers (less than M) are written to a temporary file. ． Delete the existing bitmap data file. ． Use the temporary file as a new bitmap data file.

FIGS. 7A and 7B are diagrams showing an outline of the block merging process. The hatching indicates which entry (here, each character stands for each character) each bucket / container corresponds to. Here, the size of the container is eight times the size of the bucket. For example,
Since “a” has 17 buckets before the fusion processing, there are 2 containers and 1 bucket after the fusion processing. "I" means
Similarly, 11 buckets are 1 container and 3 buckets.
Become individual. Since there are only 7 buckets, no “u” is generated in the container, and there are still 7 buckets. Will be faster and search speed will be improved).

Next, the invention according to claim 10 will be described. In the above method, buckets and containers are mixed in the data file. Data on the secondary storage device is accessed in page units due to optimization of the operating system, etc. Therefore, if buckets and containers are mixed, even if a large block like a container is introduced, The layout may not match the page boundary of the secondary storage device, and the performance may not be improved as expected. Therefore, according to the method of the present invention, the containers are arranged at the front of the data file and the blocks are arranged at the rear of the data file, so that the containers are always arranged at the page boundaries to improve the performance.

Block fusion processing procedure . Create two temporary files for writing. One is called a "temporary file for container" and the other is called "temporary file for bucket". ． The following processing is performed on the bitmap data of all entries that make up the character component table. (a) Write the container as it is to the temporary file for container. (b) The (M) buckets that are put together in a container are set as one container and are written to the temporary file for container. (c) Buckets that cannot be collected in the remaining containers (less than M) are written to a temporary bucket file. ． Delete the existing bitmap data file. ． The temporary file for bucket is connected to the temporary file for container to make a new bitmap data file.

FIGS. 7A and 7C are diagrams showing the outline of the block merging process by the above algorithm. In this algorithm, after the block merging process (state of FIG. 7C), containers are gathered at the beginning of the data file, and buckets are gathered after point A.

Next, the invention according to claim 11 will be described. The document is additionally registered even after the block merging process. Since the block type after additional registration is a bucket, if the number of additionally registered documents increases, the search speed will decrease again. In that case, the block merging process may be performed again according to the block merging process procedure. However, in the block fusion procedure, the total size of the two temporary files is
It becomes almost equal to the data file. This is highly undesirable because the size of the data file becomes huge if a large number of documents are registered. The block merging process procedure shown below improves this point so that the total size of the temporary files is about the size of the bucket portion of the data file.

Block fusion processing procedure . Create two temporary files for writing. One is called a "temporary file for container" and the other is called "temporary file for bucket". ． The following processing is performed on the bitmap data of all entries that make up the character component table. (a) Ignore the container. (b) The (M) buckets that are put together in a container are set as one container and written to a temporary file for container. (c) Buckets that cannot be collected in the remaining containers (less than M) are written to a temporary bucket file. ． Delete the bucket part of the bitmap data file. ． Connect the container temporary file and the bucket temporary file to the bitmap data file.

Differences between the methods described in claims 10 and 11 are shown in FIGS. 8 (a) to 8 (c). As shown in FIG. 8A, when the document is registered again after the bucket merging process, the buckets are sequentially inserted from the end of the data file (point B in FIG. 8A). In the block merging processing procedure, as shown in FIG. 8B, the blocks of the data file are arranged, and the search speed is improved. But 2
The total size of one temporary file is equal to the size of the data file. On the other hand, in the block fusion processing procedure proposed in this section, only the portion after the point A of the data file is processed. FIG. 8C shows the result of the block merging process. The newly created container is placed after point A where the bucket existed before the fusion. Although the containers for the same entry are not necessarily arranged at consecutive positions (for example, the “yes” container), the speed reduction is extremely small.

FIG. 9 is a block diagram for explaining still another embodiment (claim 12) of the document management apparatus according to the present invention.
In the figure, 11 is an input unit, 12 is a processing unit, 13 is a character string input processing unit, 14 is a document search processing unit, 15 is a document output processing unit, 16 is a document registration processing unit, 17 is a data unit, and 18 is a character. The component table, 19 is an output unit, and 20 is document data.

The search character string input to the input unit 11 is processed by the character string input processing 13 of the processing unit 12. The document search processing unit 14 uses the character component table 18 of the data unit 17 to search for a document that is considered to include a character string. And
The document data 20 corresponding to the retrieved document is output to the output unit 19 by the document output device 15. Document registration processing unit 16
Then, the document to be registered is registered in the document data 20, the character components are extracted from the document data 20 and registered in the character component table 18. In the following description, the target document is a 1-byte character code (for example, ASCII) and a 2-byte character code (E
The text data is composed of UC: Extended UNIX CODE. However, the target character code is EU
Other than C, it can be easily applied.

When a document is registered in the data section, a single character component and an adjacent character component are extracted and a character component table is created. The single character component is a 2-byte code of each internal character code, and the adjacent character component is a code converted from the adjacent internal character code. In this embodiment, a bit string obtained by appropriately extracting the bit component of the internal character code is used as the adjacent character component. As shown in FIG. 10, a single character component table and an adjacent character component table shown in FIG. 11 are generated for the character component and the adjacent character component obtained by the above method, respectively. In FIG. 11, the lower 1 bytes of adjacent characters are combined into 2 bytes. Each character component table indicates with 0 and 1 whether each single character component or an adjacent character component is present in each document. Figure 11
In the example of the adjacent character component table of, the bit string of a0a0 (hexadecimal) does not exist in documents 1, 2, 3 and n, but exists in documents 4 and 5. At the time of document registration, character components are extracted from the document by the above method and added to each character component table.

If only the lower 1 byte of each character component is used as the adjacent character component table, there may be a case where a document including an adjacent character which is different from the search character string but has a matching lower byte is searched. . Hiragana and katakana frequently appear, which reduces the accuracy of search. Further, since the kanji character has a low frequency of appearance in the document, the search accuracy will be low due to the influence of other character types with low search accuracy, even though the character type is originally high in search accuracy. Therefore, by creating a different adjacent character component table for each character type and using a different adjacent character component table for each character type of the search character string at the time of search, it is not affected by the character types that frequently appear in documents such as hiragana, Search accuracy can be improved. The single character component table and the adjacent character component table will be described below.

Single character component table: table indicating in which document a character appears. Adjacent character component table-same type adjacent character component table: in which document a pair of adjacent same type characters appears. * Symbol * Alphanumeric time * Hiragana * Katakana * Greek letters, graphic characters, etc. * 1-byte character code * 1st level Kanji * 2nd level Kanji-Different adjacent character composition table: Table showing which document appears

At the time of search, as in the case of registration, a single character component and an adjacent character component are extracted from the search character string, and a document containing each component is searched from the character component table. FIG. 12 is a diagram for explaining a conventional search method. In FIG. 12, a bit 1 indicates that a character component appears, and a bit 0 indicates that a character component does not appear. In the conventional search method, the character type is discriminated from the search character string, the single character component and the adjacent character component are extracted in the same manner as at the time of registration, and the bit of the character entry corresponding to each character component is extracted from the single character component table and the adjacent character component table. The map is extracted and the AND operation is performed. Therefore, the data of all target character entries will be referred to.

Next, the invention according to claim 13 will be described. In the present embodiment, as shown in FIG. 13, the bits of a specific bitmap are sequentially examined in the horizontal direction, and when the bit is 1, the corresponding bits of other bitmap entries are examined. That is, the bits are checked in the vertical direction in FIG. 13, and when all the bits are 1, the document corresponding to the bits becomes the document of the search result. When the bit is 0, the process returns to the uppermost character entry in FIG. 13 and is repeated in the same manner as the above. By doing so, only the bit indicated by the arrow is referred to, and the amount of data to be referred to is significantly reduced as compared with the conventional search method.

Next, the invention according to claim 14 will be described. Further, as shown in FIG. 14, the number of appearances of bit 1 appearing in each character entry is counted in advance, and at the time of processing for examining the above-mentioned bits, as shown in FIG. By rearranging and performing similar processing, it is possible to further reduce the amount of data to be referred to.

In the conventional search method, the variable length bitmap data, which is the bitmap data of each character entry, is
It is divided into a plurality of fixed-length blocks and stored in secondary storage. Therefore, at the time of the above-mentioned search processing, a plurality of fixed-length blocks are combined and restored to the variable-length bitmap again.
Further, even when only a part of the data of each character entry is accessed, it is necessary to read and combine all the fixed length data blocks of the character entry and restore the variable length bitmap data.

Next, the invention according to claim 15 will be described. In the present embodiment, the aforementioned character entry is composed of an index, a block table and a block on the secondary storage as shown in FIG. The index is composed of a block table pointer and a bit appearance number (the number of bit 1 appearing in the entry) pair for each internal character code. The block table has a pointer to the next block table at the head, and is composed of the block pointer and the block final registration document ID (the ID of the document registered last in the block). Therefore, all the blocks corresponding to the internal character code "A" are as shown in FIG.
It becomes the block indicated from the block table.

An example will be described below in which it is checked whether or not the character having the character code “A” appears in the document with the document ID 4000. -Get the block table from the block table pointer corresponding to the index "A". Obtain a block (block pointer 5120) including the document ID 4000 from the block final registration document ID. If the data of the block (5120) is compressed, decompress it to obtain the corresponding bits of the document ID 4000. In this way, when only a part of the character entry is required, the corresponding block can be directly obtained from the block table without accessing the entire block, and the search can be performed at high speed.

The READ time of data from the disk occupies the largest part of the search time. The more pages to be read (units of physical disk read / write), the slower the search process. When the document is registered,
As in (a), the block of character entries is distributed over a plurality of pages. Therefore, in the search processing, the processing becomes slow by reading all the pages including the dispersed blocks. In the present embodiment, as shown in FIG. 16B, the dispersed blocks are grouped into page units to reduce the number of pages to be READ during the search process and to speed up the process. In the example shown in FIG. 16, 6 or more pages had to be READ before grouping, but after grouping, there are 3 pages, and the READ time becomes half or less. The process of grouping blocks in this way is called a grouping process.

In the grouping process, as shown in FIG. 16B, the blocks are grouped into pages, but the pages that could not be grouped into pages and the blocks that were not grouped (the remaining pages Called the block page). Therefore, as shown in FIG. 17, the remaining blocks of various character entries are stored in the remaining block page. Also, the collected pages will be mixed in the file.

The processing procedure for grouping is shown below. The buffer in the description means an area on the memory. Pre-processing (a) Open the character component table. (B) Create and open a grouping character component table. (C) Allocate the remaining block page.

Grouping processing in character entry units (a) Grouping processing into pages i. Read one run length and fill the page buffer. ii. When the page buffer is full of data, write the page buffer and clear the page buffer. iii. I. Until the run length is read completely. Return to (B) Writing processing of blocks (remaining blocks) that cannot be put together on a page i. One run length is read again from the run lengths that could not be put together, and the remaining block buffer is filled. ii. When the remaining block buffer is filled with data, it is written to the remaining block page. iii. When the remaining block page area is completely used up, a new remaining block page is allocated. iv. I. Until the run length is read completely. Return to

Post-processing (a) Write out the remaining block pages that have not been written out. (B) Close the character component table and the combined character component table. By doing so, the character component table can be grouped by scanning once, and not only the processing is fast, but also the secondary storage area required at the time of processing can be minimized.

Next, the invention according to claim 16 will be described. In the search process of the present embodiment, if the number of entries extracted from the search character string is small, the character component table is less accessed, and the search becomes faster. When the single character entry and the adjacent character entry are defined in the character component table entry specification, when the search character string is n characters, n single character entries and n-1 adjacent character entries are extracted. Then, 2n-1 entries are extracted, and the search is slow.

For example, the search character string is "pattern match".
, The next entry is extracted. • Single character entry: Operate the function f (x) on the following characters. Pa, ta,-, n, ma, tsu, chi ・ Adjacent character entry: Function g (x, y) for the following two characters
To act. Pata, Tar, Nun, Ma

However, f (x) = x, g (x, y) = x
Consider a case such as + αy (where α is the maximum value that the character code can take). At this time, g (X,
When Y) is extracted, f (X) and f (Y) are always extracted (for example, when g (pa, ta) is extracted, f (pa) and f (ta) are always extracted. Be done). Therefore, the single character entry does not make sense in identifying the document containing the search string. Therefore, the single character entry is not extracted from the search character string, and only the adjacent character entry is extracted. As a result, only n-1 adjacent character entries are extracted from the n-character search character string, so that the search can be speeded up.

For example, if the search character string is "pattern match"
, The next entry is extracted. -Single character entry: Do not extract.・ Adjacent character entry: Function g (x, y) for the following two characters
To act. Pattern, tar, neon, nmma, mach, ch
It is not limited to y) = x + αy.

Next, the invention according to claim 17 will be described. This embodiment also aims at the same effect as the invention described in claim 16, and the character component table entry definition is f
A case such as (x) = x, g (x, y) = x + α (y mod β) (where α is the maximum value that the character code can take and β is an appropriate constant) is handled. At this time, g from the search character string
When (X, Y) is extracted, f (X) is always extracted (for example, when g (pa, ta) is extracted, f (X) is always
(Pa) is also extracted). Therefore, a single character entry does not make sense except for the last one character in identifying the document containing the search string. Therefore, a single character entry calculated from the last one character and an adjacent character entry are extracted from the search character string. As a result, a total of n entries of one single character entry and n-1 adjacent character entries are extracted from the n character search character string, so that the search can be speeded up.

For example, the search character string is "pattern match".
, The next entry is extracted. • Single character entry: Operate the function f (x) on the following characters. J ・ Adjacent character entry: Function g (x, y) for the following two characters
To act. Pattern, tar, neon, nmma, mach, ch
It is not limited to y) = x + α (y mod β).

Next, the invention according to claim 18 will be described. This embodiment also aims at the same effect as the invention described in claim 17, and the character component table entry definition is f
The case where (x) = x, g (x, y) = y + α (x mod β) is handled. At this time, when g (X, Y) is extracted from the search character string, f (Y) is always extracted (for example, when g (pa, ta) is extracted, f (ta) is always extracted. Extracted). Therefore, in specifying the document containing the search character string, the single character entry does not make sense except for the first character. Therefore, the single character entry calculated from the first character and the adjacent character entry are extracted from the search character string. As a result, a total of n entries of one single character entry and n-1 adjacent character entries are extracted from the n character search character string, so that the search can be speeded up.

For example, the search character string is "pattern match".
, The next entry is extracted. • Single character entry: Operate the function f (x) on the following characters. Par-Adjacent character entry: Function g (x, y) for the following two characters
To act. Pattern, tar, neon, nmma, mach, ch
It is not limited to y) = y + α (x mod β).

Next, the invention according to claim 19 will be described. In the invention according to claim 4 described above, by using a long character string (a value calculated from) of three or more characters for the entry of the character component table, the number of accesses to the character component table is reduced and the search is performed. It was shown that the speed can be significantly increased. FIG. 5 shows a character component table having a long character string as an entry. "System", "pattern", etc. are character string entries.

Consider a search process when a single character entry and an adjacent character entry are defined in the character component table entry designation and a character string entry is introduced. According to the fourth aspect of the invention, the single / adjacent character entry extracted from the character string corresponding to the character string entry included in the search character string is not used in the search process.

For example, if the search character string is "pattern match"
When the "pattern" is defined as a character string entry in, the next entry is extracted. • Single character entry: Operate the function f (x) on the following characters. Ma, tsu, chi ・ Adjacent character entry: Function g (x, y) for the following two characters
To act. Comma, match, switch ・ Character string entry: pattern

Of course, it is possible to combine the invention described in claim 4 with the invention described in claims 16 to 18 of the present invention. On the other hand, in the present invention, for the adjacent character entry, the number of extracted entries is reduced from the search character string by not extracting the entry extracted from the character string obtained by adding the preceding and succeeding characters to the corresponding character string entry, and the search speed is increased. (For the single character entry, the entry extracted from the corresponding character string entry is not extracted as before).

For example, the entries extracted from “pattern matching” are as follows, and the number of entries can be reduced by one. • Single character entry: Operate the function f (x) on the following characters. Ma, tsu, chi ・ Adjacent character entry: Function g (x, y) for the following two characters
To act. Match, character string entry: pattern

Next, the invention according to claim 20 will be described. In the invention described in claim 6, the search condition is a combination of a plurality of character strings by logical operators (AND, OR) (a single character string is also included in the search condition). Here, "AND" means searching for a document that includes both the preceding and following character strings, and "OR"
Means to search for documents that include at least one of the preceding and following character strings. Furthermore, "(", ")" can be used to clearly indicate the order of operation of operators, if necessary. By enabling logical operators to be used as search conditions, complex search requests can be expressed.

However, it is not possible to express a search request including a negation such as "including" document search "but not including" image search "" only by AND and OR. Therefore, in this embodiment, a combination of a plurality of character strings by logical operators (AND, OR, NOT) is accepted as a search condition. Here, “NOT” means to search for a document that includes the preceding character string but does not include the following character string.

Next, the invention according to claim 21 will be described. In the AND and OR processing, it suffices to calculate the bit string for the search character string before and after the operator (step 1 of the search procedure) and take the bit AND or the bit OR. However, in NOT, a similar process (a bit string for the preceding and following search character strings is calculated, bit inversion of the bit string on the rear side is performed, and bit AND is performed with the bit string on the front side) may result in omission of search. This is because the bit string for each search character string obtained using the character component table is
This is because it is not an expression of the document number that includes the character string exactly, but actually includes an erroneous search that does not include the character string. As a result, the bit-inverted bit string does not include a part of the document (number) that does not include the character string, and search omission occurs.

Therefore, in this embodiment, for NOT, the bit string for the search character string on the front side is set as the bit string of the NOT processing result. As a result, the NOT processing result is
A document including the search character string on the rear side is included, and an erroneous search occurs (the erroneous search can be eliminated in step 2 of the search process.
There is no problem in practice). However, the document that does not include the search character string on the rear side is not included, and the omission of the search can be completely prevented.
Because it cannot be saved in, it becomes a practical problem). Further, in this method, since it is not necessary to process the search character string on the rear side, it is effective in speeding up the character component table search.

Next, the invention according to claim 22 will be described. The process of extracting the character component from the document and generating the character component table is the same as that of the embodiment of the invention of claim 12 described according to FIGS. The structure of the character component table obtained by this is composed of indexes and bitmap data, as shown in FIG. The index part is a correspondence table of pointers to character components and bitmap data. The bitmap data is data of 0 and 1 indicating whether or not a character component appears in the document of the character component table. When registering a large number of documents, since the bitmap data becomes huge, it is placed in the secondary storage instead of being placed in the memory.

According to the embodiment of the present invention characterized by the method of registration in the secondary memory, a character component table is generated each time one document is registered, and the character component table is directly generated in the secondary memory each time. Instead of writing, the data is once registered in the memory and then the data in the memory is collectively written in the character component table of the secondary storage. FIG. 18 shows the data structure at the time of batch registration. In this example, the data structure on the secondary storage and the memory structure are the same for the sake of simplicity. At the time of batch document registration, the data temporarily registered in the table on the memory (shaded in the figure) is copied to the bitmap table of the secondary storage at the end of the processing, and the registration processing is completed.

Next, the invention according to claim 23 will be described. The document management apparatus does not have document data in the system, but instead manages a directory path name indicating the position of the original document on the file system as one of the document information on the secondary storage and refers to it. Is designed to directly refer to the original document on the external file system based on the directory path name. Therefore, it is not necessary to have the original document data inside the system of the device, not only uselessly utilizing the internal secondary storage, but also the trouble of using the system, and the registration of documents registered by the user or other applications. Can be referenced.

Next, the invention according to claim 24 will be described. When registering a document, specify the directory of the original data of the registered document on the file system, register all the documents in that directory or its subordinate directories, manage the directory path name in the document management device, and When referring to the contents, the original data on the external file system is directly referred to based on the registered directory path name. Therefore, when the original document is placed on the file system that the user normally uses, it is possible to provide a system in which the document existing in one directory hierarchy is directly managed by the document management apparatus. Also, by specifying a directory, it is possible to automatically register the documents contained in that directory and all the documents contained in subordinate directories, so that all the documents in the directory can be individually registered. The burden on the conventional user, which had to be specified by one user, is reduced.

Next, the invention according to claim 25 will be described. Similarly to the above, in the document management device that directly references the original data on the external file system based on the directory path name, the user specifies the directory in which the document to be registered in the device is placed in advance. Then, the device constantly monitors the directory, and when a document is registered, updated, or deleted, the same operation is automatically reflected in the character component table to perform registration, update, or deletion. By doing so, it is possible to reduce the burden of the document operation on the user. As a method of monitoring a directory, a method of checking a change in a specified directory at regular time intervals or a system operation process of a file operation of a basic system such as an OS is changed so that a document management system can be used when a file operation occurs. A method of notifying can be adopted.

[0089]

As is apparent from the above description, the present invention has the following effects. (1) Effects corresponding to claims 1 and 2: Document registration means for storing a registered document and registering and recording a value calculated from each character or two consecutive characters contained in the registered document in a character component table. And a document search means for quickly searching for a document satisfying the search condition using the character component table, the configuration of the character component table can be changed by designating the character component table entry, and a plurality of registered documents can be stored. Since it is possible to manage by dividing into folders and the character component table entry can be specified for each folder and the structure of the character component table can be changed according to the purpose, an efficient document management system can be constructed. (2) Effects corresponding to claims 3, 4, and 5: When there are two or more identical character components extracted from a single character string by the document retrieval means, the character components in the character component table are Access is performed only once, and a value calculated from a continuous character string of three or more characters is also used in the calculation of the character component. When a value calculated from a character string of three or more characters is used, the value calculated from one character or two consecutive characters from the character string is not registered in the character component table. Even if it is long, you can search documents quickly. (3) Effects corresponding to claims 6, 7, and 8: In the document retrieval means, a single character string or a plurality of character strings is AN.
It is possible to process search conditions combined with a D or OR logical operator, and there are two or more identical character components extracted from two character strings combined by an AND logical operator in the document searching means. In this case, the character components of the character component table are accessed only once, and further, they are connected by the OR logical operator in the document search means 2
When there are two or more identical character components extracted from one character string, the character components in the character component table are accessed only once. Therefore, the search condition is set to the logical operator of multiple character strings. Since it can be expressed by a combination of (AND, OR), it is possible to express a complicated search request. Also, since optimization is performed according to the logical operator,
Documents can be searched at high speed. (4) Effects corresponding to claims 9, 10 and 11: The character component table is recorded with a data file composed of two types of blocks, large and small, for storing the character component table, and a block position for each character component. It has a block merging means for storing a plurality of small blocks into a large block, which is stored by an index file, and in the block merging means, a large block is grouped in front of the data file and a small block is grouped in the rear, In the block merging means, only the area where a small block of the data file exists is targeted for block merging, and two sizes, large and small, are prepared for the size of the block for storing the bitmap data of the character component table.
Both registration / search speed can be increased. (5) Effect corresponding to claim 12: In the block merging means, each character component table entry divided into a plurality of fixed length blocks is large in order to read the character component table entry from the secondary storage at high speed at the time of retrieval. When a large fixed-length block is used as a unit of allocation when it is grouped into fixed-length blocks, a large fixed-length block that has been put together and a large fixed-length block that is packed with the remaining small fixed-length blocks that have not been put together can be written out in order. Processes the data of the composition table in one scan, speeds up and minimizes the secondary storage area required at the time of processing, and improves the search speed by combining blocks with small character entries into large blocks. To do. (6) Effects corresponding to claims 13 to 15: A bit of a specific bitmap is sequentially examined in the horizontal direction, and when the bit is 1, the corresponding bit of another bitmap entry is examined. That is, the bits are examined in the vertical direction, and if all the bits are 1, the document corresponding to the bits is the document as the search result. Further, the number of appearances of bit 1 appearing in each character entry is counted in advance, the bits are rearranged in the ascending order of the number of appearances of the bits at the time of the above-described processing for checking the bits, and the same processing is performed, so that the data to be referred to further. It is possible to reduce the amount. Further, when only a part of the character entry is required, the corresponding block can be directly obtained from the block table without accessing the entire block, and the high speed search can be performed. As described above, in the related art, the character entry is frequently accessed and the search speed is lowered. However, the search speed is improved by changing the algorithm and the data structure of the processing at the time of search. (7) Effects corresponding to claims 16 to 19: When the composition of the character component table is to record a value calculated from each character and two consecutive characters of the document, the character retrieving means retrieves the retrieval character string. Only a value calculated from two consecutive characters is extracted, and the value calculated by the document search means from two consecutive characters from the search character string and one value at the end of the search character string are calculated. A value is extracted, and a value calculated from two consecutive characters from the search character string and a value calculated from one character at the beginning of the search character string are extracted. When a value calculated from a character string of three or more characters is also used, when the document search unit extracts the character entry from the search character string, one character included in the character string corresponding to the character string entry or The character string Since so as not to extract the value calculated from the two characters in the string, including the front and rear of the characters, to reduce the number of entries to be extracted from the search string in the search process can speed up the search process. (8) Effects corresponding to claims 20 and 21: A search in which the document search means combines a single character string or a plurality of character strings with a "logical product", "logical sum" or "logical difference" logical operator Since the condition can be processed and the character string on the rear side is not processed in the processing of the two character strings that are combined by the "logical difference" in the document search means, the search condition is set to a plurality of character strings. Logical operators (AND, OR, NO
Since it can be expressed as a combination in T), it is possible to express a complicated search request. In addition, since optimization is performed according to the logical operator, documents can be searched at high speed. (9) Effect corresponding to claim 22: In the conventional method, when the character component table data is in the secondary storage, the secondary storage is accessed every time one document is registered, and the speed is increased. Although it will be late, according to the present invention, for the document to be collectively registered, the character component table is temporarily generated and registered in the memory, and then the character component table data in the memory is secondarily stored at the end of the process. Append to the above document composition table data. By doing so, the access to the secondary storage is reduced, and the batch registration of a plurality of documents can be performed at high speed. (10) Effect corresponding to claim 23: The directory path name on the file system of the external original document is managed by the document management device, and the external file is directly referenced based on the directory path name. Since it is referred to, the system of the device does not need to have the original data inside, the secondary storage is not wastefully used, and the registered document can be referred to by the user or other application without going through the system. It will be possible. (11) Effect corresponding to claim 24: Similar to the above, when managing the directory path name, all the documents in the directory and its subordinate directories are registered, so that the user usually writes the original document. When it is placed on the file system to be used, it becomes possible to make a system in which the documents existing in one directory hierarchy are managed by the document management device as they are. In addition, by specifying a directory, you can automatically register all the documents included in that directory or the subordinate directories, so that all the documents in the directory can be registered individually. It is possible to reduce the burden on the conventional user, which had to be designated by one user. (12) Effect corresponding to claim 25: When a user previously designates a directory in which a document to be registered is placed, the document management apparatus constantly monitors the registration, update, and deletion of the document in the directory, Document registration, update,
When deletion is performed, the same operation is automatically reflected in the character component table, and registration, update and deletion are performed. By doing so, it is possible to reduce the burden of the document operation on the user.

[Brief description of drawings]

FIG. 1 is a configuration diagram for explaining an embodiment of a document management device according to the present invention.

FIG. 2 is a diagram showing an example of a character component table according to the present invention.

FIG. 3 is a diagram showing another example of a character component table according to the present invention.

FIG. 4 is a configuration diagram for explaining another embodiment of the document management apparatus according to the present invention.

FIG. 5 is a diagram showing an example of a character component table having a long character string as an entry according to the present invention.

FIG. 6 is a diagram showing an example of a file structure for a character component table in the present invention.

FIG. 7 is a diagram showing an outline (No. 1) of block fusion processing of a character component table file according to the present invention.

FIG. 8 is a diagram showing an outline (No. 2) of the block fusion processing of the character component table file according to the present invention.

FIG. 9 is a configuration diagram for explaining still another embodiment of the document management apparatus according to the present invention.

FIG. 10 is a diagram showing a single character component table in the present invention.

FIG. 11 is a diagram showing an adjacent character component table in the present invention.

FIG. 12 is a diagram for explaining a conventional search method.

FIG. 13 is a diagram illustrating a search method (No. 1) according to the present invention.

FIG. 14 is a diagram for explaining a search method (2) in the present invention.

FIG. 15 is a diagram showing a data structure in the present invention.

FIG. 16: Grouping processing (1) in the present invention
It is a figure for explaining.

FIG. 17: Grouping processing in the present invention (part 2)
It is a figure for explaining.

FIG. 18 is a diagram for explaining an example of collective registration processing in the present invention.

[Explanation of symbols]

1 ... Registered document, 2 ... Document registration means, 3 ... Search condition, 4 ...
Document search means, 5 ... Corresponding document, 6 ... Character component table entry designation, 7 ... Character component table, 8 ... Document body data, 9 ... Document database, 9-1 to 9-n ... Document database, 11 ...
Input unit, 12 ... Processing unit, 13 ... Character string input processing unit, 14
... document search processing section, 15 ... document output processing section, 16 ... document registration processing section, 17 ... data section, 18 ... character component table, 19
... Output unit, 20 ... Document data.

─────────────────────────────────────────────────── ───

[Procedure amendment]

[Submission date] June 30, 1995

[Procedure Amendment 1]

[Document name to be amended] Statement

[Name of item to be corrected] Claim 13

[Correction method] Change

[Correction content]

[Procedure Amendment 2]

[Document name to be amended] Statement

[Name of item to be corrected] Claim 16

[Correction method] Change

[Correction content]

[Procedure 3]

[Document name to be amended] Statement

[Name of item to be corrected] 0002

[Correction method] Change

[Correction content]

[0002]

2. Description of the Related Art A known document describing a conventional document management apparatus is, for example, Japanese Patent Laid-Open No. 5-324722. According to this publication, in order to reduce the size of the character component table to be used in the character string search, increase the degree of search, and enable high-speed document registration, the search character string input to the input unit is processed. The document search processing unit searches for a document that is considered to include a character string by using the character component table of the data unit. The document data corresponding to the retrieved document is output to the output unit by the document output process. In the document registration process, the document to be registered is registered in the document data, the character component is extracted from the document data, and registered in the character component table. When searching a search string from the document, present as character component table, in the value value calculated from the character is calculated from the 1 character component table and the adjacent character indicating whether present in the document a document A document is searched using the adjacent character component table that indicates whether or not to do it. That is, it is a system that searches documents at high speed using a character component table that records characters that have appeared for each document.

[Procedure amendment 4]

[Document name to be amended] Statement

[Name of item to be corrected] 0003

[Correction method] Change

[Correction content]

Further, the above-mentioned Japanese Patent Laid-Open No. 5-324722 and the previously proposed Japanese Patent Application No. 5-298433 propose a single character component table showing whether or not a character component exists in a document. , and a character component extracted from adjacent characters is to find a document by using the adjacent character component table indicating whether present in the document.

[Procedure Amendment 5]

[Document name to be amended] Statement

[Correction target item name] 0006

[Correction method] Change

[Correction content]

[0006]

In order to achieve the above-mentioned object, the present invention (1) stores a registered document, and a value calculated from each character or two consecutive characters contained in the registered document ( In a document management device having a document registration unit for registering and recording (character component) in a character component table and a document searching unit for rapidly searching for a document satisfying a search condition using the character component table, a character component table entry designation is made. To change the composition of the character component table by
(2) In (1), the registered document can be divided into a plurality of folders and managed, and a character component table entry can be designated for each folder; or (3) In (1), the document search means. When there are two or more identical character components extracted from a single character string, the character component in the character component table is accessed only once, or (4) In (1) above. As the calculated value of the character component, a value calculated from a continuous character string of three characters or more is also used, or (5) above (4)
In the case of using the value calculated from a continuous character string of three or more characters in the calculation of the character component at the time of document registration, the value calculated from one character or two continuous characters from the character string is used as the character component. Do not register in the table, or
(6) In the above (1), the document search means can process a search condition in which a single character string or a plurality of character strings are combined by an AND logical operator or an OR logical operator, or ( 7) In (6), the same character component extracted from the two character strings combined by the AND logical operator in the document search means is 2
If there are more than one, the character component in the character component table is accessed only once, or (8) above (6)
In the above, if there are two or more identical character components extracted from the two character strings combined by the OR logical operator in the document search means, the character component in the character component table is accessed only once. Or (9) In (1), the character component table is a data file composed of two types of blocks, large and small, for storing the character component table, and an index file for recording the block position for each character component. Or a block merging means for storing a plurality of small blocks into a large block, or (10) in (9), the block merging means groups a large block in front of the data file and a small block in the rear. Grouping blocks, or (11) in the block merging means in (10) above. , Subjecting only the area of the data file having a small block to the block fusion, or (12) in the above (9), the block fusion means rapidly stores the character component table entry from the secondary storage at the time of retrieval. When reading each character component table entry divided into multiple fixed-length blocks into a large fixed-length block for reading, the large fixed-length block is used as the unit of allocation, and the combined large fixed-length block and the rest that cannot be combined By writing out large fixed-length blocks packed with small fixed-length blocks in sequence, the data of the character component table can be processed in one scan, and the secondary storage area required at high speed can be minimized. Suppressing, or (13) Retrieval input from the input device while holding a large amount of document data Finds documents containing strings, a document management apparatus for outputting a document retrieved by the output device, the character code component from the document at the document registration (single character component) and two or more characters adjacent characters or al
Bit string out calculate the component (adjacent character components), each document to produce a single character component table and one or more adjacent character component table indicating whether containing the respective components, the variable of the character component table dividing the entry of each character component length into a plurality of fixed length blocks registered in the secondary storage, like the time of registration from the search string to extract a single character component and two or more characters adjacent characters component when searching In a document management apparatus that integrates fixed-length blocks on secondary storage corresponding to the adjacent character components to generate each character entry and retrieves a document, the bit at which the character component appears in the character entry is set to 1 When the bit that does not appear is set to 0, only one target character entry is appropriately selected when ANDing the bitmaps of a plurality of character entries at the time of search, and the bit string is scanned, and only when the value is 1 Other d Examining the corresponding bit in birds by the corresponding bit values of all entries and search results document 1, to perform a full-text search fewer accesses to the bit map of the character entry to a minimum, or (14) In the above (13), the number of 1 bits appearing in each character entry (the number of appearances) is recorded in the secondary storage in advance for each character entry, and a bit map of a plurality of character entries is searched at the time of retrieval. Of the target character entries when performing the AND process of, the character entry with the smallest number of occurrences of the bit is selected, and the bit string is scanned, and when the value is 1, the other character entries are sorted in ascending order of the number of occurrences of the bit. By checking the corresponding bit and finding the document in which the value of the corresponding bit of all character entries is 1, the character entry Minimum to suppress possible to perform a full-text search and access to the bitmap, or,
(15) In (13), the bitmap data of a variable-length character entry is converted into a plurality of fixed-length blocks so that it is not necessary to read all blocks of the character entry when accessing a part of the character entry. It is divided into two parts and stored in the secondary storage, and a block table that manages each fixed-length block is provided in the secondary storage, and an arbitrary fixed-length block can be accessed from the block table. A full-text search is performed by reducing access to a block on the secondary storage that is not present, or (16) a value calculated from each character or two consecutive characters included in the registered document while saving the registered document ( A document registration unit for registering and recording (character component) in the character component table, and a document search unit for quickly searching for a document satisfying the search condition using the character component table. And the configuration of the character component table can be changed by designating the character component table entry, and the configuration of the character component table records each character of the document and a value calculated from two consecutive characters. , extracting only values the document search <br/> search means is calculated from the 2 consecutive characters from the search string, or (17) (16)
In the above, the document search means extracts a value calculated from two consecutive characters from the search character string and a value calculated from one character at the end of the search character string, or (1
8) In (16), the document search means extracts a value calculated from two consecutive characters from the search character string and a value calculated from the first character of the search character string, or (19) In the case of (18), in the case where a value (character string entry) calculated from a character string of three or more consecutive characters is also used in the configuration of the character component table, the document search means uses the search character string. When the character entry is extracted, a value calculated from one character included in the character string corresponding to the character string entry or two characters included in the character string including the characters before and after the character string is not extracted. Or (20) In the above (16), the document searching means combines a single character string or a plurality of character strings with a "logical product", a "logical sum", or a "logical difference" logical operator. Article It can handle a, or (2
1) In the above (20), in the processing of the two character strings combined by the "logical difference" by the document search means, the character string on the rear side is not processed, or (22) The above (1)
In the case of registering a large number of documents at once, the document registration means does not directly write the character component table data generated each time one document is registered to the character component table on the secondary storage, but once. The document is registered in the memory, and then the character component table data in the memory is collectively written in the character component table of the secondary storage to perform the batch registration of documents at high speed, or (23) (1) above. When the directory path name of the registered document on the file system registered by the document registration means is managed on the secondary storage and the document contents are referred to, the file path directly on the file system is registered based on the registered directory path name. Not only does the document management system not need to have the original data internally by referencing the original data of To facilitate the reference of the registration document,
Alternatively, (24) in (1), the directory path name of the original data of the registered document on the file system registered by the document registration means is managed, and when referring to the document contents, the registered directory path name Directly refer to the original data on the file system based on, specify the directory when registering the document,
By registering all the documents in the directory and its subordinate directories, the document designation at the time of registering a plurality of documents can be facilitated, or (25) above (1)
In managing the directory path name of the original data of the registered document on the file system registered by the document registration means, and referring to the document contents, the original data on the file system is directly copied based on the registered directory path name. The directory to be registered is specified in advance so that the registration, update, and deletion of documents in that directory are always monitored, and when a document is registered, updated, or deleted, it is automatically In addition, by registering, updating, and deleting the document with respect to the character component table, the user's time and effort for document management is reduced.

[Procedure correction 6]

[Document name to be amended] Statement

[Correction target item name] 0009

[Correction method] Change

[Correction content]

(3) The document searching means can process a search condition in which a single character string or a plurality of character strings are combined by AND or OR logical operators, and
When there are two or more identical character components extracted from the two character strings combined by the AND logical operator in the document retrieval means, the character component in the character component table is accessed only once. Further, when there are two or more identical character components extracted from the two character strings combined by the OR logical operator in the document retrieval means, the character component in the character component table is accessed only once. I tried not to . Use multiple operators as search conditions for logical operators (AN
Since it can be expressed as a combination of (D, OR), it is possible to express a complicated search request. In addition, since optimization is performed according to the logical operator, documents can be searched at high speed.

[Procedure Amendment 7]

[Document name to be amended] Statement

[Correction target item name] 0012

[Correction method] Change

[Correction content]

(6) A bit of a specific bitmap is sequentially examined in the horizontal direction, and when the bit is 1, the corresponding bit of another bitmap entry is examined. That is, the bits are examined in the vertical direction, and if all the bits are 1, the document corresponding to the bits is the document as the search result. Further, the number of appearances of bit 1 appearing in each character entry is counted in advance, the bits are rearranged in the ascending order of the number of appearances of the bits at the time of the above-described processing for checking the bits, and the same processing is performed, so that the data to be referred to further. It is possible to reduce the amount. Further, when only a part of the character entry is referred to , the corresponding block can be directly obtained from the block table without accessing all the blocks, and the high speed search can be performed. As described above, in the related art, the character entry is frequently accessed and the search speed is lowered. However, the search speed is improved by changing the algorithm and the data structure of the processing at the time of search.

[Procedure Amendment 8]

[Document name to be amended] Statement

[Correction target item name] 0013

[Correction method] Change

[Correction content]

[0013] (7) When the records the value calculated from the character component table documents the configuration of each character and two consecutive characters, the two characters the document retrieval means is continuous from the search string Only the calculated value is extracted, and the document search means extracts a value calculated from two consecutive characters from the search character string and a value calculated from one character at the end of the search character string, and , A value calculated from two consecutive characters from the search character string and a value calculated from one character at the beginning of the search character string are extracted, and the composition of the character component table further includes three or more consecutive characters. When a value calculated from a string is also used, when the document search unit extracts the character entry from the search character string, one character included in the character string corresponding to the character string entry or the character before and after the character string is included in the character string. Including the character Since so as not to extract the value calculated from the 2 characters in string, the number of entries to be extracted from the search string in the search process is reduced, thereby speeding up the search process.

[Procedure Amendment 9]

[Document name to be amended] Statement

[Name of item to be corrected] 0025

[Correction method] Change

[Correction content]

The document search procedure is as follows. ． A document number that may include a search character string is obtained using the character component table 7. (A) An entry defined by the character component table entry designation 6 is extracted from the search character string. (B) Bitmaps of all the extracted entries (see FIG. 2).
(One horizontal row of) is extracted from the character component table, and bit AND is performed. ． The document body of the document number obtained above is read from the document body data 8 and it is checked whether or not the search character string is included, and the included document set is used as the search result.

[Procedure Amendment 10]

[Document name to be amended] Statement

[Name of item to be corrected] 0029

[Correction method] Change

[Correction content]

For example, when the character component table of FIG. 2 is used,
Since the search term "Man Machine System" consists of 9 characters, the character component table shows
You need to access "n", "shi", "su", "te" and "mu" 9 times. However, in reality, “ma”, “n”, and “shi” are 2
Since they appear once, the access to these entries can be combined into one. That is, the actual access to the character component table is "ma""n""shi""su"
You only need 6 times of "te" and "mu".

[Procedure Amendment 11]

[Document name to be amended] Statement

[Correction target item name] 0031

[Correction method] Change

[Correction content]

At the time of registration, from "... that man-machine system ...", characters "a", "no", "ma", "n" are displayed as characters.
“Si”, “su”, “te”, “mu”, “ha” and “system” are extracted as a character string and recorded in the character component table. During the search, the search word from the "man-machine system", "ma,""down,""ma""death,""down" as a character, but "the system" is extracted as a character string, is included in the "system", " Shi "
And remove single character duplicates. After all, "ma""n"
The search time can be greatly reduced by simply accessing the three entries of "system".

[Procedure Amendment 12]

[Document name to be amended] Statement

[Correction target item name] 0038

[Correction method] Change

[Correction content]

The number of blocks included in the index file is determined by the character component table entry designation. The bitmap data file is composed of blocks including the following two fields.・ Next block offset field ・ Data field

[Procedure Amendment 13]

[Document name to be amended] Statement

[Correction target item name] 0039

[Correction method] Change

[Correction content]

The block size may be set in the range of several tens of bytes to several kilobytes according to the performance requirement. FIG. 6A is a diagram showing an example of a file structure for the character component table. Note that loading the index file on the semiconductor memory is effective for speeding up.

[Procedure Amendment 14]

[Document name to be amended] Statement

[Correction target item name] 0042

[Correction method] Change

[Correction content]

At the start of using the document search system, the block size is set to be small and the registration speed is prioritized. (While the number of registered documents is small, the search time is small even if the search speed is a little slow, so there is almost no problem.) . When a large number of documents are registered and the number of blocks included in the bitmap data file increases, block fusion processing is performed to collect a plurality of buckets in a container. In a typical operating system, who are summarized in large blocks than previously divided data into small blocks access is because high speed. As a result, the block merging process improves the search speed, and the search time can be shortened even when the number of registered documents is large.

[Procedure Amendment 15]

[Document name to be amended] Statement

[Correction target item name] 0061

[Correction method] Change

[Correction content]

[0061] document ID as an example the case to determine whether or not the internal character code character of "A" appears in the document of 4000, will be described below. -Get the block table from the block table pointer corresponding to the index "A". Obtain a block (block pointer 5120) including the document ID 4000 from the block final registration document ID. · If the data block (5120) is compressed and stretched to obtain the bit corresponding to the document ID4000. As described above, when only a part of the character entry is referred to, the corresponding block can be directly obtained from the block table without accessing all the blocks, and the search can be performed at high speed.

[Procedure Amendment 16]

[Document name to be amended] Statement

[Correction target item name] 0089

[Correction method] Change

[Correction content]

[0089]

As is apparent from the above description, the present invention has the following effects. (1) Effects corresponding to claims 1 and 2: Document registration means for storing a registered document and registering and recording a value calculated from each character or two consecutive characters included in the registered document in a character component table. And a document search means for quickly searching for a document satisfying the search condition using the character component table, the configuration of the character component table can be changed by designating the character component table entry, and a plurality of registered documents can be stored. Since it is possible to manage by dividing into folders and the character component table entry can be specified for each folder and the structure of the character component table can be changed according to the purpose, an efficient document management system can be constructed. (2) Effects corresponding to claims 3, 4 and 5: When there are two or more identical character components extracted from a single character string by the document search means, for the character components in the character component table Access is performed only once, and a value calculated from a continuous character string of three or more characters is also used in the calculation of the character component. When using a value calculated from a character string of three or more characters,
Since a value calculated from a character or two consecutive characters is not registered in the character component table, even if the search character string is long, the document can be searched at high speed. (3) Effects corresponding to claims 6, 7 and 8: In the document searching means, a single character string or a plurality of character strings is A
It is possible to process search conditions combined with an ND or OR logical operator, and there are two or more identical character components extracted from two character strings combined by an AND logical operator in the document searching means. In this case, the character components in the character component table are accessed only once, and the document retrieval means further combine them by OR logical operators.
When there are two or more identical character components extracted from one character string, the character component in the character component table is accessed only once . Since the search condition can be expressed by combining a plurality of character strings with logical operators (AND, OR), it is possible to express a complicated search request. In addition, since optimization is performed according to the logical operator, documents can be searched at high speed. (4) Effects corresponding to claims 9, 10, and 11: The character component table is recorded with a data file composed of two types of blocks, large and small, for storing the character component table, and a block position for each character component. It has a block merging means for storing a plurality of small blocks into a large block, which is stored by an index file, and in the block merging means, a large block is grouped in front of the data file and a small block is grouped in the rear, In the block merging means, only the area where a small block of the data file exists is targeted for block merging, and two sizes, large and small, are prepared for the size of the block for storing the bitmap data of the character component table. Both search speed can be increased. (5) Effect corresponding to claim 12: In the block merging means, each character component table entry divided into a plurality of fixed length blocks is large in order to read the character component table entry from the secondary storage at high speed at the time of retrieval. When a large fixed-length block is used as a unit of allocation when it is grouped into fixed-length blocks, a large fixed-length block that has been put together and a large fixed-length block that is packed with the remaining small fixed-length blocks that have not been put together can be written out in order. Processes the data of the composition table in one scan, speeds up and minimizes the secondary storage area required at the time of processing, and improves the search speed by combining blocks with small character entries into large blocks. To do. (6) Effects corresponding to claims 13 to 15: A bit of a specific bitmap is sequentially examined in the horizontal direction, and when the bit is 1, the corresponding bit of another bitmap entry is examined. That is, the bits are examined in the vertical direction, and if all the bits are 1, the document corresponding to the bits is the document as the search result. Further, the number of appearances of bit 1 appearing in each character entry is counted in advance, the bits are rearranged in the ascending order of the number of appearances of the bits at the time of the above-described processing for checking the bits, and the same processing is performed, so that the data to be referred to further. It is possible to reduce the amount. Further, when only a part of the character entry is referred to, the corresponding block can be directly obtained from the block table without accessing all the blocks, and the search can be performed at high speed. As described above, in the related art, the character entry is frequently accessed and the search speed is lowered. However, the search speed is improved by changing the algorithm and the data structure of the processing at the time of search. (7) Effects corresponding to claims 16 to 19: When the composition of the character component table is to record a value calculated from each character and two consecutive characters of the document, the character retrieving means retrieves the retrieval character string. Only a value calculated from two consecutive characters is extracted, and the value calculated by the document search means from two consecutive characters from the search character string and one value at the end of the search character string are calculated. A value is extracted, and a value calculated from two consecutive characters from the search character string and a value calculated from one character at the beginning of the search character string are extracted. When a value calculated from a character string of three or more characters is also used, when the document search unit extracts the character entry from the search character string, one character included in the character string corresponding to the character string entry or The character string Since so as not to extract the value calculated from the two characters in the string, including the front and rear of the characters, to reduce the number of entries to be extracted from the search string in the search process can speed up the search process. (8) Effects corresponding to claims 20 and 21: A search in which the document search means combines a single character string or a plurality of character strings with "logical product", "logical sum" or "logical difference" logical operators Since the condition can be processed and the character string on the rear side is not processed in the processing of the two character strings that are combined by the "logical difference" in the document search means, the search condition is set to a plurality of character strings. Logical operators (AND, OR, NO
Since it can be expressed as a combination in T), it is possible to express a complicated search request. In addition, since optimization is performed according to the logical operator, documents can be searched at high speed. (9) Effect corresponding to claim 22: In the conventional method, when the character component table data is in the secondary storage, the secondary storage is accessed every time one document is registered, and the speed is increased. Although it will be late, according to the present invention, for the document to be collectively registered, the character component table is temporarily generated and registered in the memory, and then the character component table data in the memory is secondarily stored at the end of the process. Append to the above document composition table data. By doing so, the access to the secondary storage is reduced, and the batch registration of a plurality of documents can be performed at high speed. (10) Effect corresponding to claim 23: The directory path name on the file system of the external original document is managed by the document management device, and the external file is directly referenced based on the directory path name. Since it is referred to, the system of the device does not need to have the original data inside, the secondary storage is not wastefully used, and the registered document can be referred to by the user or other application without going through the system. It will be possible. (11) Effect corresponding to claim 24: Similar to the above, when managing the directory path name, all the documents in the directory and its subordinate directories are registered, so that the user usually writes the original document. When it is placed on the file system to be used, it becomes possible to make a system in which the documents existing in one directory hierarchy are managed by the document management device as they are. In addition, by specifying a directory, you can automatically register all the documents included in that directory or the subordinate directories, so that all the documents in the directory can be registered individually. It is possible to reduce the burden on the conventional user, which had to be designated by one user. (12) Effect corresponding to claim 25: When a user previously designates a directory in which a document to be registered is placed, the document management apparatus constantly monitors the registration, update, and deletion of the document in the directory, Document registration, update,
When deletion is performed, the same operation is automatically reflected in the character component table, and registration, update and deletion are performed. By doing so, it is possible to reduce the burden of the document operation on the user.

Claims (25)

[Claims] 1. A document registration means for storing a registered document and for registering and recording a value calculated from each character or two consecutive characters contained in the registered document in a character component table.
A document management device comprising: a document search unit that searches for a document satisfying a search condition at high speed by using the character component table, and is capable of changing the configuration of the character component table by designating a character component table entry. . 2. The document management apparatus according to claim 1, wherein the registered document can be managed by being divided into a plurality of folders, and a character component table entry can be designated for each folder. 3. When the document retrieval means extracts two or more identical character components extracted from a single character string, the character component of the character component table is accessed only once. The document management device according to claim 1. 4. The document management apparatus according to claim 1, wherein a value calculated from a continuous character string of three or more characters is also used as the calculated value of the character component. 5. When calculating a character component at the time of document registration,
5. When using a value calculated from a continuous character string of three or more characters, the value calculated from one character or two continuous characters from the character string is not registered in the character component table. Document management device described. 6. The document search means uses an AND logical operator or O for a single character string or a plurality of character strings.
2. The document management device according to claim 1, wherein the search conditions combined by the R logical operator can be processed. 7. When the document retrieval means has two or more identical character components extracted from two character strings combined by an AND logical operator, the character component of the character component table is accessed once. The document management apparatus according to claim 6, wherein the document management apparatus performs only that. 8. When the document retrieval means has two or more identical character components extracted from two character strings combined by an OR logical operator, the character component of the character component table is accessed once. The document management apparatus according to claim 6, wherein the document management apparatus performs only that. 9. The character component table is stored by a data file composed of two types of blocks, large and small, for storing the character component table and an index file for recording block positions for each character component, and a plurality of small blocks are stored. 2. The document management apparatus according to claim 1, further comprising a block merging unit that collects blocks into large blocks. 10. The document management apparatus according to claim 9, wherein the block merging unit collects large blocks in front of the data file and small blocks in rear of the data file. 11. The document management apparatus according to claim 10, wherein in the block merging means, only a region of a data file in which a small block exists is an object of block merging. 12. The block merging means combines each of the character component table entries divided into a plurality of fixed length blocks into a large fixed length block in order to read the character component table entry from the secondary storage at a high speed at the time of retrieval. The large fixed-length block is used as the unit of allocation, and the large fixed-length blocks that are grouped together and the remaining small fixed-length blocks that cannot be combined are packed into the large fixed-length block. 10. The document management apparatus according to claim 9, wherein processing is performed by scanning once, and the secondary storage area required at the time of processing is minimized at high speed. 13. A document management device that holds a large amount of document data, searches for a document including a search character string input from an input device, and outputs the document searched by the output device, the document management device storing the document when the document is registered. A bit string component extracted from each character code component and two or more adjacent characters is generated, and a single character component table and one or more adjacent character component tables indicating whether or not each document includes each component are generated. The entropy of each variable-length character component of the character component table is divided into a plurality of fixed-length blocks and registered in the secondary memory, and a single character component and two or more adjacent characters are searched for from the search character string when searching. In a document management device that extracts character components, integrates fixed-length blocks in secondary storage corresponding to the adjacent character components, generates each character entry, and retrieves a document, the character component is temporarily output by the character entry. When the presenting bit is set to 1 and the non-appearing bit is set to 0, when the AND process of the bit maps of a plurality of character entries is performed at the time of searching, one target character entry is appropriately selected and the bit string is scanned. Only when the value is 1, the corresponding bits of other entries are checked, and the documents whose corresponding bits of all entries have a value of 1 are set as the search result, so that the access to the bitmap of the character entry is minimized. A document management device characterized by performing full-text search. 14. The number of 1-bits appearing in each character entry is recorded in a secondary storage in advance for each character entry, and the target character is used when ANDing bit maps of a plurality of character entries at the time of retrieval. Of the entries, the character entry with the smallest number of occurrences of the bit is selected, and when the bit string is scanned, the corresponding bits of other character entries are checked in the order of the smallest number of occurrences of the bit only, and all character entries The value of the corresponding bit of is 1
14. The document management apparatus according to claim 13, wherein the full text search is performed by minimizing the access to the bitmap of the character entry by using the document that is the search result as the search result. 15. The variable-length character entry bitmap data is divided into a plurality of fixed-length blocks so that it is not necessary to read the entire block of the character entry when accessing a part of the character entry. Stored in secondary memory,
By having a block table that manages each fixed-length block on the secondary storage, and making it possible to access any fixed-length block from the block table, it is possible to access unnecessary blocks on the secondary storage. 14. The document management device according to claim 13, wherein the full-text search is performed with a reduced number. 16. A document registration means for storing a registered document and registering and recording a value calculated from each character or two consecutive characters contained in the registered document in a character component table, and the character component table. And a document search means for quickly searching for a document corresponding to the search condition, and the structure of the character component table can be changed by designating the character component table entry.
The composition of the character component table is set for each character of the document and two consecutive characters.
A document management apparatus, wherein when a value calculated from a character is recorded, the character search means extracts only a value calculated from two consecutive characters from a search character string. 17. The document search means extracts a value calculated from two consecutive characters from the search character string and a value calculated from one character at the end of the search character string. Item 16. The document management device according to item 16. 18. The document search means extracts a value calculated from two consecutive characters from a search character string and a value calculated from one character at the beginning of the search character string. Item 16. The document management device according to item 16. 19. When using a value calculated from a continuous character string of three or more characters in the structure of the character component table, when the character search unit extracts the character entry from the search character string, 19. The value calculated from one character included in a character string corresponding to a character string entry or two characters included in a character string including characters before and after the character string is not extracted. Document management device. 20. The document search means is capable of processing a search condition in which a single character string or a plurality of character strings are combined by “logical product”, “logical sum” or “logical difference” logical operators. The document management apparatus according to claim 16, wherein the document management apparatus is a document management apparatus. 21. The document management apparatus according to claim 20, wherein when processing two character strings that are combined by "logical difference" in the document search means, a character string on the rear side is not processed. 22. The document registration means, when registering a large number of documents at once, writes the character component table data generated each time one document is registered, directly into the character component table on the secondary storage. 2. The document is collectively registered at high speed by temporarily registering it in the memory, and then writing out the character component table data in the memory to the character component table of the secondary storage collectively. Document management device described. 23. A directory path name of a registered document on a file system registered by the document registration means is managed on a secondary storage, and when referring to document contents, the directory path name is directly based on the registered directory path name. By referring to the original data on the file system, not only the document management system does not need to have the original data in the document management system, but also the user or other application can easily refer to the registered document. 1. The document management device according to 1. 24. When managing the directory path name of the original data of the registered document on the file system registered by the document registration means and referring to the document contents, the file system is directly based on the registered directory path name. By making reference to the original data above, specifying a directory when registering a document, and registering all documents in that directory and its subordinate directories, it is possible to easily specify a document when registering multiple documents. The document management apparatus according to claim 1, wherein the document management apparatus is a document management apparatus. 25. When the directory path name of the original data of the registered document on the file system registered by the document registration means is managed and the document contents are referred to,
Directly refer to the original data on the file system based on the registered directory path name, specify the directory to be registered in advance, always monitor the registration, update, and deletion of documents in that directory, When a document is registered, updated or deleted, the document is automatically registered, updated or deleted in the character component table to reduce the user's time and effort for managing the document. The document management device according to item 1.