KR101846347B1 - Method and apparatus for managing massive documents - Google Patents

Abstract

A method and an apparatus for managing a large document are disclosed. In a distributed file system realizing storage spaces of a plurality of physically separated regional nodes as one file system, the apparatus for managing a document stores a task file gathering texts of a plurality of electronic documents together, distributes and performs a process of extracting a string from the text of the task file through the plurality of regional nodes, and stores a keyword file including a mapping relationship of the string with a file path of an electronic document to which the string belongs.

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention [0002]

The present invention relates to a document management method and apparatus for facilitating classification and search of a large-capacity document.

If a large amount of documents exist, it is difficult to integrate them into one physical storage space and store them. For example, if there is a system that integrates and stores city information, the system needs a very large storage space for accumulating various types of documents produced in various government offices and public institutions in the city on a daily basis. However, since the amount of data that can be stored in one physical storage space (hard disk, etc.) in hardware is limited, there is a limit in realizing integration storage in one space.

A large volume of documents can generally be searched by the document name or some keyword assigned to the document. However, it is difficult to know exactly what the document name or some keyword contains in the document. For example, assuming that a file having a document name of 'Daejeon City Development Status' and 'Seoul City Development Status' among the large-capacity documents stored in the city information integration system exists and information about the development restricted area is included in the document , If the 'development restriction zone' is not assigned as the keyword of the document, it is difficult for the user to search the document storing the information on the 'development restriction zone' realistically unless the user individually grasps the contents of the document.

Alternatively, you can search the entire text of the document to find out if the user has the word you want. For example, in the above example, the entire text included in 'Daejeon Development Status' and 'Seoul City Development Status' document can be searched for individually to find out whether the word 'development restricted zone' exists. However, if the number of documents to be retrieved is one or two, it is practically impossible to retrieve the contents of the entire document. However, if there are tens to hundreds of thousands of documents, it is practically impossible.

Korean Patent Registration No. 10-1562232 entitled " Feature Information Abstraction Technique for Improving Performance of Automatic Document Classification System "

SUMMARY OF THE INVENTION The present invention has been made in view of the above problems, and it is an object of the present invention to provide a method and apparatus for managing a large-capacity document that automatically classifies a large-capacity document or retrieve a large-

According to an aspect of the present invention, there is provided a method of managing a large-capacity document, the method comprising: storing at least one electronic document; Storing a work file in which texts of a plurality of electronic documents are collected as one unit in a distributed file system in which a storage space of a plurality of local nodes physically separated is implemented as one file system; Distributing a process of extracting a character string from the text of the work file through the plurality of local nodes; And storing a keyword file including a mapping relationship between a file path and a character string of the electronic document to which the character string belongs.

According to another aspect of the present invention, there is provided an apparatus for managing a large-capacity document, the apparatus comprising: a plurality of local nodes having physically separated self storage spaces; And a master node managing the storage spaces of the plurality of local nodes as one file system and dividing the data into blocks and storing the divided data in the file system, wherein the master node collects texts of a plurality of electronic documents A keyword file including a working file and a mapping relation between a character string and a file path is stored in the file system and the plurality of local nodes performs a process of extracting a character string from a part of the job file stored in the respective storage spaces, And stores the file path of the electronic document to which the extracted character string belongs in the keyword file.

According to the present invention, the storage space of a large-capacity document can be flexibly increased, and a large-capacity document can be automatically classified and stored. In addition, based on the contents contained in the document, it is possible to quickly search a desired document in a large-size document.

1 is a diagram showing a configuration and an application example of a document management apparatus according to an embodiment of the present invention;
2 is a diagram illustrating an example of storing a file in a distributed file system according to an embodiment of the present invention;
3 is a diagram illustrating an example of executing program code in a distributed file system according to an embodiment of the present invention;
Figure 4 illustrates a method for automatically classifying an electronic document according to an embodiment of the present invention;
5 is a diagram illustrating an example of a method of extracting and storing a character string from a large-capacity electronic document according to an embodiment of the present invention.
6 is a diagram illustrating an example of a method of generating a keyword file for a large-capacity electronic document according to an embodiment of the present invention;
7 is a diagram illustrating an example of a method of updating a keyword file according to an embodiment of the present invention;
8 is a diagram illustrating an example of indexing a keyword file according to an embodiment of the present invention;
FIG. 9 is a flowchart illustrating a method of classifying an electronic document according to an embodiment of the present invention;
10 is a flowchart illustrating a method of generating a keyword file according to an embodiment of the present invention,
11 is a flowchart illustrating a keyword-based electronic document retrieval method according to an embodiment of the present invention.

Hereinafter, a method and apparatus for managing a large-capacity document according to an embodiment of the present invention will be described in detail with reference to the accompanying drawings.

1 is a diagram showing a configuration and an application example of a document management apparatus according to an embodiment of the present invention.

Referring to FIG. 1, the document management apparatus 130 includes a master node 140, a plurality of local nodes 150, 152, and 154, and a monitoring node 160.

The nodes (the master node 140, the local nodes 150, 152, 154, the monitoring node 160, etc.) of the present embodiment are physically separated from each other including a processor capable of executing their own storage space and program code Computing device. The master node 140, the local nodes 150, 152, and 154, and the monitoring nodes 160 are interconnected through a wired / wireless communication network.

The master node 140 manages a distributed file system in which the storage spaces of the physically separated local nodes 150, 152, and 154 are implemented as one file system. In a distributed file system, the storage space of each local node is treated as if it were a single storage space. For example, the storage space of each local node 150, 152, 154 in the distributed file system is handled like a track and a sector of a conventional hard disk, such that a conventional hard disk stores data by dividing storage space into tracks and sectors . When the file is stored in the hard disk, the master node 140 divides the file into the local nodes 150, 152, and 154, and stores the divided files in the local nodes 150, 152, and 154, respectively.

The master node 140 maps and manages a file path indicating a logical storage location of a file in the distributed file system and information indicating which part of the actual file is stored in which local node 150, 152, 154. Therefore, from the point of view of the document collection device, when storing a file or writing program code for that file, you can use the file path of "/ directory / file name" You do not need to know how it is distributed. In the following, the file path is a logical storage location of a file distributed to a plurality of local nodes in the distributed file system, and means a form of "directory + file name".

 The distributed file system of this embodiment can be implemented in various ways, for example, as a HDFS (Hadoop Distribution File System). An example of storing a document in the distributed file system of this embodiment will be described again with reference to FIG.

Each of the regional nodes 150, 152, and 154 includes a processor that not only has its own storage space but also can execute program codes transmitted through the master node 140. For example, when the file path of the document file stored in the distributed file system of this embodiment is '/ Row data / Energy / A.pdf', the data of the actual A.pdf document file is the same as that of the distributed file system It is divided into basic storage units and distributed to each local node. When searching for a specific word in the A. pdf document file, the master node 140 receives the program code for searching for a specific word from the document collecting apparatus 110 and transmits it to each of the regional nodes 150, 152, and 154, The nodes 150, 152, and 154 may transmit the received program code to a portion of the A. pdf document file stored in their storage space, and may transmit the result to the master node 140. Accordingly, the process of searching for a specific word in the A. pdf document file can be processed quickly because it is distributed by the plurality of local nodes 150, 152, and 154. The process of delivering the program code to each of the regional nodes 150, 152, 154 is shown in FIG.

The master node 140 and the local nodes 150, 152 and 154 are connected to each other through wired / wireless communication, so that when a failure occurs in a part of the local nodes 150, 152 and 154, the data stored in the local nodes can not be read. Accordingly, the master node 140 redundantly stores the files in the local nodes 150, 152, and 154 when the files are stored in the distributed file system as shown in FIG.

The monitoring node 160 performs heartbeat communication with the master node 140 and each of the regional nodes 150, 152 and 154 to confirm whether the master node 140 and the regional nodes 150, 152, and 154 are operating normally. The role of the monitoring node 160 may be implemented in the master node 140. Depending on the embodiment, the monitoring node 160 may be omitted.

The document collecting apparatus 110 collects various electronic documents from a plurality of document supplying organizations (100, 102, 104). For example, when the city information integration system is to be implemented, the document collection device 110 collects various types of electronic documents such as education, culture, environment, and traffic from various government offices and public institutions. The document supplying organizations 100, 102, and 104 of the present embodiment may be servers of various government offices or public institutions that provide various kinds of electronic documents.

The document collection device 110 stores the collected documents in the distributed file system via the master node 140. [ Since the storage spaces of the plurality of local nodes 150, 152 and 154 in the distributed file system are treated as one file system, it is possible to easily store and manage a large-capacity document through expansion of the local nodes 150, 152 and 154. The document collecting device stores the electronic document in the logical file path. The master node performs mapping and distribution of the electronic document to the plurality of local nodes. Therefore, when a user writes program code to perform various operations on a specific electronic document, it is not necessary to use a file path of the electronic document and to know which part of the electronic document is stored in which area node. That is, when one file is stored in a plurality of sectors on the hard disk, the file path indicating the storage location of the file is used when the program code for inquiring, storing, or performing various operations is used. It is the same principle as not using.

The user terminal 120 can access the document collection device 110 to search for and search for a desired document. The user terminal 120 refers to various types of terminals capable of wired / wireless communication. For example, there are various types of user terminals such as a computer, a smart phone, and a tablet PC.

Although the present embodiment illustrates a document collecting apparatus 110 and a plurality of document supplying organizations 100, 102, and 104 in order to facilitate understanding of the present invention, the present invention is not necessarily limited thereto. That is, the document management apparatus 130 can be applied to various other fields that require storage of large-capacity documents other than the document collection apparatus 110 shown in this embodiment.

2 is a diagram illustrating an example of storing a file in a distributed file system according to an embodiment of the present invention.

1 and 2, the master node 140 divides the document file 200 into blocks 202, 204, and 206 having a predetermined size and stores the divided document files 200 in the storage spaces 210, 212, and 214 of the local nodes 150, 152, and 154. The size of the block can be variously modified according to the distributed file system. The file path of the document file 200 stored in the distributed file system is composed of a logical storage location of "directory + file name ", and information on which local node the block of the document file 200 is actually stored in (140) stores and manages it along with the file path.

Each block 202, 204, 206 of the document file 200 may be redundantly stored in a plurality of local nodes 150, 152, 154 in order to prepare for the occurrence of a failure in a specific area node. For example, the first block 202 of the document file 200 may be duplicated in the first storage space 210 of the first area node 150 and the Nth storage space 214 of the Nth area node 154 The second block 204 is redundantly stored in the second area node 152 and the first area node 150 and the third block 206 is stored in the second area node 152 and the second area node 150. [ 152). Thus, in the event of a failure in the first local node 150, the master node 140 may read the first block 202 from the Nth local node 154.

In this embodiment, for convenience of description, each block is stored only once, but each block may be stored twice or more according to an embodiment.

3 is a diagram showing an example of executing program code in the distributed file system according to the embodiment of the present invention.

Referring to FIGS. 1 and 3 together, a user may create program code 300 to perform a specific operation on a file stored in a distributed file system.

When the master node 140 receives the program code 300 from the document collection device 110, the master node 140 transfers the program code 300 back to each of the local nodes 150, 1552, and 154. Each local node performs program code 300 for a portion of a file stored in its respective storage space and forwards the result to the master node 140.

For example, when the master node 140 receives the program code for extracting the string of the electronic document from the document collecting apparatus 110, the master node 140 transfers the program code to the local nodes 150, 152, and 154. Each of the regional nodes 150, 152, and 154 transmits a result of executing the program code to a part of the electronic document stored in the respective storage space, to the master node 140. The master node 140 finally generates the execution result of the program code using the program code execution result of each local node and provides it to the document collection apparatus 110. [

In the distributed file system, various methods can be applied to distribute the program code through each local node. For example, the conventional resource management technology YARN (Yet Another Resource Regiator) can be used.

4 is a diagram illustrating a method of automatically classifying an electronic document according to an embodiment of the present invention.

1 and 4, the master node 140 stores at least one electronic document 400, 402, 404 received from the document collection device 110 in a distributed file system. The master node 140 transmits program codes for grasping the classification of the electronic document to the local nodes 150, 152, and 154. Each of the regional nodes (150, 152, 154) executes a program code on a part of the electronic document stored in its storage space to grasp the classification of the electronic document.

The program code for identifying the classification of the electronic document includes (1) a step of extracting a character string (S410), (2) comparing the extracted character string with a word list of each classification in the classification table, and (3) And storing the classification information of the electronic document (S430).

(1) extracting a character string (S410) may be implemented in various ways. For example, the step of extracting a character string (S410) may be performed by extracting a character string composed of a specific part of speech (for example, a noun) or extracting a character string in a space unit by morphological analysis of the text included in the electronic document . If the text for a portion of the electronic document stored in the local node is "at Cincinnati and Home ... held at Miller Park in Milwaukee on the 25th (Korea time)" and the string is extracted in units of spaces, Eames will be "," 25th (Korean time), "Milwaukee", "Miller Park", "Open", "Cincinnati", "at home" etc. or extract the string of nouns through morphological analysis The string is extracted as "Teeth", "25 days", "Korea Time", "Milwaukee", "Miller Park", "Cincinnati", "Home" etc. In addition, Various conventional methods of extracting a character string from a text, such as extracting a specific part of speech (for example, a noun) from a subject or an object, can be applied.

(2) The step of comparing with the classification table (S420) may be implemented by an algorithm for determining which class of the word list belongs to the word list of the classification table 440, in which the character string extracted from the text of each electronic document belongs. A classification table 440 including a word list for each classification is predefined. For example, in the case of the 'environment' classification, words such as 'development restriction zone' and 'Han River water system' may be defined in the classification table 440 as a word list of the classification.

For example, when extracting a string in a space unit, the string may not completely coincide with the 'development restriction zone' of the classification table because it includes an investigation such as 'development restriction zone'. In this case, it is possible to grasp the classification by checking whether or not the word of the classification table is included in the character string, instead of searching the classification table 440 for a word completely matching the character string. That is, since the string 'development restriction zone' includes the 'development restriction zone' existing in the word list of the 'environment' field, the 'environment' classification can be grasped. As another example, when the string is 'Gangneung main site', 'gas station' in the word list of the classification table 440 is included in the character string, so that 'energy' can be grasped by the classification of the character string.

In another embodiment, the classification for each character string can be grasped by referring to the classification table 440 only for the top few strings having a large frequency by using the frequency of the character string extracted from the text. In this case, if the string is extracted in units of spaces, 'development restricted area' and 'development restricted area' are different strings according to the investigation. Therefore, it is possible to use nouns other than the investigation (for example, ') Can be extracted and the frequency can be grasped.

(3) The step of storing the classification according to the electronic document (S430) may be implemented by an algorithm of collecting the identified classification for each character string and storing it in the classification of the corresponding electronic document. Since the classification is identified for each string, a plurality of classifications can be grasped for one electronic document. It is possible to map all of the plurality of classifications identified according to the embodiment to the electronic document 450 and store and manage the same, or map only one class of the maximum frequency to the electronic document according to the frequency of the plurality of classifications, It can be stored and mapped to an electronic document. The master node can collect the results of the string extraction processing distributed through the regional nodes (150, 152, 154), map it to an electronic document and store it as a separate file or additionally store classification information as metadata of the electronic document.

FIG. 5 is a diagram illustrating an example of a method of extracting and storing a character string from a large-capacity electronic document according to an embodiment of the present invention.

1 and 5 together, a large amount of electronic documents 500, 502, and 504 are stored in the distributed file system. The master node 140 generates a work file 510 in which a plurality of electronic documents 500, 502, and 504 are integrated and stores the work file 510 in a distributed file system. For example, if there are 20,000 electronic documents 500, 502, and 504, the master node 140 reads text contained in 20,000 electronic documents 500, 502, and 504 and stores the read text in one work file 510. The work file 510 is stored in the distributed file system.

The file path of each electronic document can be included in the work file 510 together with the text to identify which electronic document the text stored in one work file belongs to. For example, the text read from the first electronic document is stored as one line in the work file and the file path 1 of the first electronic document is inserted before the line. At the end of the text, you can insert an indicator (for example, a line break mark) to distinguish different electronic documents. One work file in which texts of a plurality of electronic documents are integrated into one is divided into a plurality of blocks as shown in FIG. 2 and distributedly stored in a plurality of local nodes.

The master node 140 receives the program code for keyword file generation from the document collection device 110 and transmits the received program code to the local nodes 150, 152, and 154. Each of the regional nodes 150, 152, and 154 receiving the program code for generating the keyword file executes the program code for a part of the work file stored in each of the local nodes 150, 152, and 154.

The program code for keyword file generation can be implemented by an algorithm that extracts a character string, determines the mapping relationship between the file path and the character string to which the character string belongs, and then sorts the mapping relation based on the character string. An implementation example of the program code for keyword file generation will be described again with reference to FIG. The master node 140 collects the information of the mapping relationship between the string and the file path obtained by the local nodes 150, 152, and 154, removes redundancy, and stores the keyword file 520 sorted in the distributed file system .

6 is a diagram illustrating an example of a keyword file generation method for a large-capacity electronic document according to an embodiment of the present invention.

Referring to FIG. 6, the program code for generating a keyword file is composed of (1) a step of extracting a character string, (2) a step of removing redundancy of a character string, and (3) .

(1) The step of extracting a character string may be implemented by extracting a character string consisting of only nouns or extracting a character string on a space-by-segment basis through morphological analysis of text included in the working document 600. The same method as the above-mentioned string extracting method can be applied, except that a step of extracting a character string according to the present embodiment is performed for a work document 600 in which a plurality of electronic documents are integrated.

(2) Removing redundancy of a string may include removing redundancy between strings having the same file path, and removing redundancy between strings having different file paths. For example, in the string list 610 extracted for each file path, when 'string a' is duplicated from the text belonging to the first file path, only one 'string a' is left and the rest is removed. After the redundancy of the strings belonging to the same file path is removed, the relationship between each string and the file path is mapped 620. Then, the redundancy of the same 'string b' existing in the different file paths (the first file path and the second file path) is removed and integrated into one. That is, 'string b' is mapped to the first file path and the second file path.

(3) In the step of sorting based on the string, the mapping relationship between the character string and the file path is sorted in ascending or descending order based on the string (630) and stored in the keyword file. Keyword files are stored in a distributed file system.

7 is a diagram illustrating an example of a method for updating a keyword file according to an embodiment of the present invention.

5 and 6 relate to a case where a character string is extracted using a work file in which a plurality of electronic documents are integrated to generate and store a keyword file. When a new electronic document is stored in the distributed file system after the keyword file is generated, it is preferable to analyze only the newly stored electronic document and add the character string to the keyword file, rather than integrating the entire electronic document again to extract the character string. Fig. 7 relates to a case where a character string of an electronic document newly stored in a distributed file system is reflected in a keyword file.

7, it is assumed that the electronic document 700 is stored in the distributed file system at the file path k. The master node 140 delivers the program code for creating the keyword file to each of the local nodes 150, 152 and 154 where the electronic document is stored, and each of the regional nodes 150, 152 and 154 executes the program code for a part of the electronic document stored in each storage space And the master node 140 generates a final mapping relationship 710 of the string-file path distributed through each local node.

The master node 140 adds the mapping relationship of the string-file path extracted from the electronic document 700 to the existing keyword file. For example, when the string b is extracted from the electronic document and the string b already exists in the keyword file 720, the master node 140 maps the string b of the keyword file 720 to add the file path k do. If the string e extracted from the electronic document does not exist in the keyword file, the master node 140 adds the mapping relationship of the string e-file path k to the keyword file.

8 is a diagram showing an example of indexing a keyword file according to an embodiment of the present invention.

Referring to FIG. 8, it is possible to make an index using a character string of a keyword file as a key so that a keyword file can be easily retrieved. The size of a keyword file storing and storing tens of thousands of electronic document strings with each file path can also be very large. Therefore, it may take a lot of time to search all the keyword files sequentially. Therefore, as in the present embodiment, an index file in which a string of a keyword file is used as a key can be used without directly using the keyword file.

For example, the master node 140 generates an index table 810 in which each string of the keyword file is mapped to an index indicating the position of the file path, and a file path table 820 that stores the file path by index, Save it to the file system.

Upon receiving the search request for the string b, the master node 140 refers to the index table 810 to determine the index 2 for the string b, searches the file path table 820 based on the index 2, The file path 1 and the file path 2 to be mapped to the file path 1 and the file path 2, respectively. The master node 140 may read the electronic document corresponding to the file path 1 and the file path 2 from the storage space of each local node and provide the electronic document to the user terminal 120 through the document collection device 110.

9 is a flowchart illustrating an electronic document classification method according to an embodiment of the present invention.

Referring to FIG. 9, the document management apparatus 130 extracts a character string from the text of the electronic document (S900). For example, the document management apparatus 130 recognizes texts such as numbers, characters, and symbols included in a document created in various forms such as a pdf file, an ms-word file, and a ppt file, and performs a morphological analysis or a spacing unit Extracts a string based on. In the case of an electronic document in which text is stored as an image, a text as an image can be recognized using a conventional character recognition program. In addition, texts such as letters and numbers can be recognized from various types of electronic documents.

The document management apparatus 130 refers to a classification table (440 in FIG. 4) including a list of classified words, and grasps at least one classification to be given to the electronic document (S910). When a character string is extracted from the text of an electronic document in a space unit, the extracted character string may not completely match the word in the word list. In this case, if the word in the word list is included in the extracted string, the classification to which the word belongs is classified as the classification of the electronic document. As another example, the document management apparatus can extract a noun only through morphological analysis or the like, and then search a word list of a classification table matching the noun to grasp the classification. As another example, the document management apparatus 130 may extract a predetermined number of strings based on the frequency of the string extracted from the electronic document, and search the word list of the classification table matching the extracted string to grasp the classification.

The document management apparatus 130 compares at least one sorting information obtained by comparing the character string extracted from the electronic document and the classification table with the electronic document and stores it (S920). The document management apparatus 130 may store classification information in an electronic document, such as storing the mapping relationship between the electronic document and classification information in a separate file, or adding classification information to the meta information of the electronic document.

10 is a flowchart illustrating a method of generating a keyword file according to an embodiment of the present invention.

Referring to FIG. 10, the document management apparatus 130 reads a text of a plurality of electronic documents and creates a work file integrated into one, and stores the created work file in a distributed file system (S1000). The document management apparatus 130 extracts a character string from the text of the job file stored in the distributed file system (S1010). As we have already seen, work files are distributed across multiple local nodes, and the extraction of strings is handled by multiple local nodes. Extracting a string from text can be implemented in a variety of ways as described above.

The document management apparatus 130 removes the redundancy of the extracted character string and obtains a mapping relationship between the character string and the file path of the electronic document to which the character string belongs, and generates and stores a keyword file including the character string and the file path as keys and values ).

11 is a flowchart illustrating a keyword-based electronic document retrieval method according to an embodiment of the present invention.

11, when receiving a search request including a search keyword and a search classification (S1100), the document management apparatus 130 refers to the keyword file and grasps the file path of the character string corresponding to the search keyword (S1110) .

If the string of the keyword file includes a character string extracted in a space unit, the search may be included together with the 'major part'. Search keywords are usually nouns such as "principal", so the search keyword and the string in the keyword file may not match exactly. Therefore, the document management apparatus 130 can determine whether a character string matching the search keyword exists in the character string, rather than determining whether the character string matches the search keyword completely. For example, if a search keyword is 'primary' and a string such as 'primary', 'primary', 'primary' is present in the keyword file, these strings all include 'primary' The device recognizes all of the 'major parts', 'major parts', and 'major parts' as a string corresponding to the search keyword 'major parts', and grasps the file path of the electronic document mapped with these identified strings.

If the file path is identified, the document management apparatus 130 identifies and provides an electronic document having a classification corresponding to the search class of the search request among the electronic documents in the file path (S1120). For example, let's assume that the file path of the string corresponding to the search keyword 'principal' is the file path 1 and file path 2, and the file path 3 of the 'main point' in the 'principal file'. In this case, the search classification is 'traffic', classification information of 'environment' and 'traffic' is given to each electronic document of file path 1 and file path 2, and classification information of 'environment' If so, the document management apparatus 130 extracts and provides the electronic document of the file path 2 to which the classification information of 'traffic' is given.

The present invention can also be embodied as computer-readable codes on a computer-readable recording medium. A computer-readable recording medium includes all kinds of recording apparatuses in which data that can be read by a computer system is stored. Examples of the computer-readable recording medium include ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical data storage, and the like. The computer-readable recording medium may also be distributed over a networked computer system so that computer readable code can be stored and executed in a distributed manner.

The present invention has been described with reference to the preferred embodiments. It will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims. Therefore, the disclosed embodiments should be considered in an illustrative rather than a restrictive sense. The scope of the present invention is defined by the appended claims rather than by the foregoing description, and all differences within the scope of equivalents thereof should be construed as being included in the present invention.

Claims (10)

A method for managing a large capacity document by a document management apparatus including a master node and a plurality of local nodes,
Storing at least one electronic document;
Storing a work file in which texts of a plurality of electronic documents are collected as one unit in a distributed file system in which a storage space of a plurality of local nodes physically separated is implemented as one file system;
Distributing a process of extracting a character string from the text of the work file through the plurality of local nodes; And
Storing a keyword file including a mapping relationship between a file path and a character string of an electronic document to which a character string belongs;
Wherein the step of storing the keyword file comprises:
Extracting a character string from texts of a plurality of electronic documents stored in the job file, removing redundancy of the extracted character string, and determining a file path of the electronic document to which the character string belongs; And
Mapping a string and a file path to each other in a key-value relationship, and storing a keyword file sorted based on a character string. The method of claim 1, wherein storing the electronic document comprises:
Extracting a character string from the text of the electronic document;
Recognizing a classification to which the character string belongs by referring to a classification table to which at least one word is mapped according to the classification; And
And storing classification information of the electronic document. 3. The method of claim 2,
Receiving a search keyword and a search classification from a user terminal;
Identifying a character string corresponding to the search keyword by referring to the keyword file, and determining at least one or more file paths mapped to the character string; And
And searching and providing an electronic document satisfying the search classification among the electronic documents stored in the identified file path. The method as claimed in claim 1,
And extracting a character string based on morpheme analysis or spacing unit for the text of the work file. delete A plurality of local nodes having physically separated self storage spaces; And
And a master node managing the storage spaces of the plurality of local nodes as one file system and dividing data into blocks and storing the divided data in the file system,
Wherein the master node stores a work file in which texts of a plurality of electronic documents are combined and a keyword file including a mapping relationship between a character string and a file path in the file system,
Wherein the plurality of local nodes extract a character string from a part of the job file stored in a storage space of the local node, store the file path of the electronic document to which the extracted character string belongs in the keyword file,
The plurality of local nodes extract a character string from the job file, remove the redundancy of the extracted character string, determine the file path of the electronic document to which the character string belongs, map the character string and the file path in relation to the key and value, And stores the keyword in the keyword file. The method according to claim 6,
The plurality of local nodes extract a character string from a part of the electronic document stored in the respective storage spaces,
Wherein the master node refers to a classification table in which at least one word is mapped according to classification, identifies a classification to which a character string extracted from the electronic document belongs, and stores classification information for each electronic document. 8. The method of claim 7,
The master node, when receiving a search keyword and a search classification from a user terminal, refers to the keyword file to identify at least one file path of a string corresponding to the search keyword, And the electronic document satisfying the classification is read and provided in the file system. delete A computer-readable recording medium storing a program for performing the method according to any one of claims 1 to 4.

Images