KR100290665B1 - Method for storing and managing structured document in relational database - Google Patents

Abstract

The present invention relates to a relational database management system for storing and efficiently searching a structured document written in a general-purpose language for electronic document standardization as a relational database. The present invention relates to parsing an electronic document and generating tokens corresponding to tags, contents, and entities. Making a step; Storing a document number for the electronic document, order information for each token, and type information of a token corresponding to each order information in a document table; Storing the document number, tag type information, relationship information with other tags, and the order information corresponding to a tag for each tag token in a tag table; Storing document numbers, order information corresponding to the contents, and the contents in the contents table for each contents token; Storing the document number, the order information corresponding to the entity, and the name of the entity for each entity token in an entity table; And corresponding to the entity table, storing the document number, the name of the document type definition, the name of the entity, and the content corresponding to the entity in the entity content table, wherein the information obtained through syntax analysis of the SGML document is used as is. In order to maximize the utilization of the model and the performance of the developed system, it is possible to reflect this information and to store / use document order information.

Description

Method for storing and managing structured document in relational database

The present invention relates to a relational database management system for storing and efficiently searching a structured document written in a general-purpose language for electronic document standardization as a relational database.

As the spread of text processing devices using computers, such as various word processors, electronic publishing, and electronic newspaper production systems, has been expanded, a database that can be used to share written information on heterogeneous systems regardless of hardware environment is established. And the importance of search and exchange is increasing day by day.

International standards for exchanging documents and expressing the logical structure of documents are ODA (Office Document Architecture: International Standard (IS) 8613) and SGML (Standard Generalized Markup Language: IS 8879). ODA can construct the logical and layout structure of a document in parallel. In contrast, SGML emphasizes the consistency of markup that represents the logical structure of document information. Therefore, ODA has a limited structure of document structure, which makes it unsuitable for expressing complex document structure. However, SGML has a conceptual logical structure, so any complex document can be created. This flexible programmatic nature allows SGML to specify document content more deeply than any other standard.

Therefore, the use of SGML is becoming a dominant trend, and research has been conducted to process SGML documents using a database. By the way, these are mainly done in object-oriented database. In order to store SGML documents based on the relational database management system most commonly used in the field, a database model for this is needed and also SGML documents stored in the database are inquired. The response speed is very important for this purpose, so it is necessary to design a database model that meets this purpose.

The technical problem to be achieved by the present invention is to store the SGML document as a relational database and to improve the performance of the system for document retrieval function, to minimize the number of query words required during document lookup SGML document to maximize the performance of the development application To provide a relational database management method for.

1 is a diagram illustrating a relationship between objects related to contents of an SGML document related to the present invention.

2 is a diagram illustrating an example of an SGML document as a token by parsing it.

3 shows the structure of the tables used in the database and their relationships according to the present invention.

In order to achieve the above object, the structured document database management method according to the present invention,

In a relational database management system for storing structured documents written in general-purpose languages for electronic document standardization and efficiently inquiring them, the method for storing structured documents in a database,

Parsing the structured document to generate tokens corresponding to tags, content and entities; Storing a document number for the structured document, order information for the respective tokens, and type information of tokens corresponding to the order information in a document table; Storing the document number, tag type information, relationship information with other tags, and the sequence information corresponding to the tag for each tag token in a tag table for each tag token; Storing the content number, the order information corresponding to the content, and the content in a content table for each content token; Storing, for all the entity tokens, the document number, the order information corresponding to the entity, and the name of the entity for each entity token in an entity table; And corresponding to the entity table, storing the document number, the name of the document type definition, the name of the entity, and the content corresponding to the entity in an entity content table.

Another structured document database management method according to the present invention for achieving the above object,

In a relational database management system for storing structured documents written in general-purpose languages for electronic document standardization and efficiently inquiring them, the method for storing structured documents in a database,

Parsing the structured document to generate tokens corresponding to tags, content and entities; The document number for the structured document, the order information for each of the tokens, and the type information if the token is a tag, the relationship information with other tags, the order information of the start and end tags, the content if the token is content, and the token is If it is an entity, storing the name of the entity in a document table; And corresponding to the entity token, storing the name of the document type definition, the name of the entity, and the content corresponding to the entity in an entity content table.

Hereinafter, with reference to the accompanying drawings will be described in detail a preferred embodiment of the present invention.

It is the database schema that describes the logical structure of the database, and the Document Type Definition (DTD) file that describes the logical structure of documents in SGML documents. The DTD describes the logical information of the document in smaller units than the document. Since a document is composed of several pieces of information, it is necessary to distinguish them logically. A logical structure of descriptive markup of DTD is suitable for this.

In general, to store SGML documents in a database, you need to process the following information into a database:

(1) elements within a document-information such as tags, entities, contents, attributes, etc.

(2) Structure relationship information between tags

(3) Order information in document

It is known that such information is efficiently stored in an object-oriented database management system. This is because SGML documents define the internal structure of documents, and this structure can be efficiently processed by using object identifiers (OIDs) provided by most object-oriented database management systems. On the other hand, in order to store the structure of SGML documents in the relational database management system, it is necessary to express the above information in a table structure, which is a logical data storage form of the relational database. There is a need to minimize structured query language (SQL).

In view of the above, the present invention proposes a schema design using document order information to compensate for the structural disadvantage of the relational database management system in terms of system speed in storing SGML documents in the relational database management system.

The present invention is used to store SGML documents in a relational database management system in units of segments based on SGML characteristics as described above. Therefore, it can be applied to SGML related software product development or application such as SGML Repository Manager which stores / manages SGML documents in database.

As described above, the present invention is a database model for storing an SGML document as a relational database.

(1) SGML documents can be stored in a relational database by presenting a relational database model based on document units (hereinafter referred to as 'tokens') generated by parsing SGML documents;

(2) It improves the performance of the system by storing the order information in the document together and using it when searching.

In the present invention, the storage method of the SGML document stores the tokens generated through syntax analysis that verifies whether the SGML document is written according to the document type definition (DTD) in a table in order, and stores the document information according to each token type. do. Here, such a set of tokens is called an ESIS (Element Structure Information Set).

1 illustrates a relationship between objects related to contents of an SGML document related to the present invention. In FIG. 1, a tag represents a structure of an SGML document, and most operations on an SGML document such as a document search are performed based on this. In this case, the range information of the document cannot be dependent on the structure information of the tag in the document search. In this case, the number of required queries is increased, which may cause a decrease in system performance. In order to compensate for this, the present invention uses the order of syntax analysis and extraction in the table design of the relational database as the key of each token, and stores and utilizes the information about the token, thereby dramatically reducing the number of required queries and improving the performance of the system.

If the table is designed based on the above description, it can be expressed as follows.

After parsing the SGML document to generate tokens corresponding to tags, content, and entities, the following tables are created, and FIG. 3 shows the structure of these tables and their relationships.

The SGML document table stores document numbers for SGML documents, order information for each token, and type information of tokens corresponding to each order information. The tag table stores document numbers, tag type information, relationship information with other tags, and order information corresponding to tags for all tag tokens. Here, the relationship information stored in the tag table includes the order information of the first lower tag token, the order information of the upper tag token, and the order information of the next equivalent tag token, and the order information stored in the tag table indicates the start. Order information of the tag token and order information of the tag token indicating the end.

In the content table, document numbers, order information corresponding to the content, and content are stored for each content token. In the entity table, document numbers, order information corresponding to the entities, and names of the entities are stored for each entity token. The entity content table stores the document number, the name of the document type definition, the name of the entity, and the content corresponding to the entity, corresponding to the entity table.

SGML Document Table

(doc_idinteger // document serial number

seq_idinteger // Sequence information (serial number) of each token in the document

type varchar (255) // type of each token)

Tag table

(doc_idinteger // document serial number

Tag_name varchar (255) // tag name

first_child integer // Order information of first child tag

next_sibling integer // order information of next sibling tag

order information for parentinteger // parent tags

Sequence information of start_id integer // start_tag

order information of end_idinteger // end_tag)

Content table

(doc_idinteger // document serial number

seq_idinteger // Order information in document

content varchar2 // document content)

Entity table

(doc_idinteger // document serial number

seq_idinteger // Order information in document

name varchar (255) // entity name)

Entity_Content table

(DTD_name varchar (255) // DTD name

Ent_name varchar (255) // entity name

content varchar2 // entity content)

The following shows an example of an SGML document and explains how this document is stored in each table according to the present invention.

〈! DOCTYPE Memo SYSTEM 'momo.dtd' [

<! ENTITY DOM 'Document Object Model'>]>

<Note>

<Preface>

<Recipient> Hong Gil-dong </ receiver>

<Sender> Samsung </ sender>

<Date> July 24, 1998 </ date>

</ Preface>

<Body>

SRE supports the &DOM; established by the W3C.

</ Text>

〈End Greetings〉

Thank you.

</ End greeting>

<Note>

The SGML document as described above is parsed and represented as a token as shown in FIG. 2. The original number before each token in the figure indicates the serial number (seq_id) of each token.

Table 1 shows what is stored in the SGML document table, with a sequence number (seq_id) appended to each token, and a representation of each type ('start_tag', 'end_tag', 'content', 'entity'). .

Table 2 shows what is stored in the tag table, and the name of each tag token ('note', 'header', receiver ',' sender ', date', 'body', end greeting '), and the children of each tag token, The serial numbers of the parent and the same tag token (first_child, parent, next_sibling) are stored. For example, in the case of the tag token 'header', since the tag token immediately following is the 'receiver' token with the serial number 3, '3' is stored in the 'first_child' item, and the tag token immediately above is the serial number. Since '1' is a 'memo' token, '1' is stored in the 'parent' item, and '13' is stored in the 'next_sibling' item because the tag tokens at the same position are the 'body' token with serial number 13. The serial numbers (start_id and end_id) of the start and end tokens of each tag token are stored.

Document number (doc_id) Serial number (seq_id) Type One One start_tag One 2 start_tag One 3 start_tag One 4 content One 5 end_tag One 6 start_tag One 7 content One 8 end_tag One 9 start_tag One 10 content One 11 end_tag One 12 end_tag One 13 start_tag One 14 content One 15 entity One 16 content One 17 end_tag One 18 start_tag One 19 content One 20 end_tag One 21 end_tag

Table 3 shows what is stored in the contents table, and the contents are stored together with the serial number (seq_id) of the data corresponding to the contents of the document. Table 4 shows what is stored in the entity table, and the entity name is stored along with the serial number (seq_id) of the location where the entity is located. Table 5 shows what is stored in the entity content table, the content of which is stored along with the entity name Ent_name.

doc_id tag_name first_child next_sibling parent start_id end_id One memo 2 - - One 21 One preface 3 13 One 2 12 One receiver - 6 2 3 5 One Caller - 9 2 6 8 One date - - 2 9 11 One main text - 18 One 13 17 One Greeting - - One 18 20

doc_id seq_id content One 4 Hong Gil Dong One 7 Samsung SDS One 10 July 24, 1998 One 14 SRE was enacted by the W3C One 16 Support. One 19 Thank you.

doc_id seq_id name One 15 DOM

DTD_name ent_name content $$ DTD1 DOM Document Object Model

Next, an inquiry about the stored SGML document based on the database model as described above is performed as follows.

When selecting the type of token to search, the serial number of the start token and the end token for the token corresponding to the selected token type is searched in the tag table. The token between the serial number of the start token and the end token is then retrieved from the document table, and for each token found, from the content table if the token type is' content ', and the token type is the entity (' entiry). '), The content is retrieved from the entity table and the entity content table.

For example, the function to look up the "header" in the above document is performed by the following procedure.

a. From the tag table, get the values of the start token serial number ('start_id = 2') and end token serial number ('end_id = 12').

b. From the SGML document table, the tokens in the header are obtained using the serial number obtained in step a.

c. For each token extracted in b, the content is taken from the content table if the type is 'content' and from the entity table and entity content table if the entity is 'entity'.

Performing the above steps a, b, and c can bring about the contents of the preface, ie, Hong Gil-dong Samsung July 24, 1998.

Hereinafter, another embodiment for improving the speed of the system will be described. This embodiment is a method of making the table designed in the previous embodiment into one table. In this case, the storage of the system may be required more than in the previous embodiment. However, the processing speed of the system is improved by reducing the number of SQL queries.

Parse the SGML document to generate tokens corresponding to tags, content, and entities, then create the following table:

The document table contains the document number for the structured document, the order information for each token, the type information if the token is a tag, the relationship information with other tags, the order information of the start and end tags, the content if the token is content, and the token. If this entity, the name of the entity is stored, and the entity content table stores the name of the document type definition, the name of the entity, and the content corresponding to the entity in correspondence with the entity token. Here, the relationship information stored in the document table includes the order information of the first lower tag token, the order information of the upper tag token, and the order information of the next equivalent tag token, and the order information stored in the document table indicates the start. Order information of the tag token and order information of the tag token indicating the end.

SGML Document Table

(doc_idinteger // document serial number

Tag_namevarchar (255) // tag name

typevarchar (255) // type of token

contentvarchar2 // document content

Order information (parental number) of parentinteger // parent tag

seq_idinteger // Token-in-document order information

Order information of end_idinteger // end_tag

first_childinteger // Order information of first child tag

next_siblinginteger // order information of next sibling tag)

Entity Content Table

(DTD_namevarchar (255) // DTD name

Ent_namevarchar (255) // entity name

contentvarchar2 // entity content)

Table 6 shows what is stored in the SGML document table, and Table 7 shows what is stored in the entity content table.

doc_id tag_ name type contents parent_id seq_id end_id first_ child next_ sibling One memo start_tag - - One 21 2 - One preface start_tag - One 2 12 3 13 One receiver start_tag - 2 3 5 - 6 One receiver contents Hong Gil Dong 3 4 - - - One receiver end_tag - 3 5 - - - One Caller start_tag - One 6 8 - 11 One Caller contents Samsung SDS 6 7 - - - One Caller end_tag - 6 8 - - - One date start_tag - One 9 11 - - One date contents July 24, 1998 9 10 - - - One date end_tag - 9 11 - - - One preface end_tag - 2 12 - - - One main text start_tag - One 13 17 - 18 One main text contents SRE was enacted by the W3C 13 14 - - - One main text entity DOM 13 15 - - - One main text contents Support. 13 16 - - - One main text end_tag - 13 17 - - - One Greeting start_tag - One 18 20 - - One Greeting contents Thank you. 18 19 - - - One Greeting end_tag - 18 20 - - - One memo end_tag - One 21 - - -

DTD_name ent_name content $$ DTD1 DOM Document Object Model

Inquiry about the stored SGML document based on the database model as described above is performed as follows.

If you select the type of token you want to retrieve, you can get the serial number of the start token and end token for the token corresponding to the selected token type from the document table, and the content token ('content' ) Is obtained as a single query, but if there is an entity ('entiry') within the search scope, the content is retrieved from the entity content table. As a result, the response speed of the system is increased by reducing the number of queries compared to the method according to the previous embodiment.

As described above, according to the present invention, an element structure information set (ESIS) obtained through parsing of an SGML document, which is essential in the SGML paradigm, is reflected in a database model and additionally By storing and utilizing the sequence information of, it is possible to maximize the utility of the model and the performance of the developed system.

Claims (6)

In a relational database management system for storing structured documents written in general-purpose languages for electronic document standardization and efficiently inquiring them, the method for storing structured documents in a database, Parsing the structured document to generate tokens corresponding to tags, content and entities; Storing a document number for the structured document, order information for the respective tokens, and type information of tokens corresponding to the order information in a document table; Storing the document number, tag type information, relationship information with other tags, and the sequence information corresponding to the tag for each tag token in a tag table for each tag token; Storing the content number, the order information corresponding to the content, and the content in a content table for each content token; Storing, for all the entity tokens, the document number, the order information corresponding to the entity, and the name of the entity for each entity token in an entity table; And And corresponding to the entity table, storing the document number, the name of the document type definition, the name of the entity and the content corresponding to the entity in an entity content table. The method of claim 1, wherein the relationship information with other tags stored in the tag table includes order information of a first lower tag token, order information of an upper tag token, and order information of a next equivalent tag token. How to manage a structured document database. The method of claim 1, wherein the order information corresponding to a tag stored in the tag table includes order information of a tag token indicating a start and order information of a tag token indicating a last. In a relational database management system for storing structured documents written in general-purpose languages for electronic document standardization and efficiently inquiring them, the method for storing structured documents in a database, Parsing the structured document to generate tokens corresponding to tags, content and entities; The document number for the structured document, the order information for each of the tokens, and the type information if the token is a tag, the relationship information with other tags, the order information of the start and end tags, the content if the token is content, and the token is If it is an entity, storing the name of the entity in a document table; And And corresponding to the entity token, storing the name of the document type definition, the name of the entity, and the content corresponding to the entity in an entity content table. 5. The method of claim 4, wherein the relationship information with other tags stored in the document table includes order information of a first lower tag token, order information of an upper tag token, and order information of a next equivalent tag token. How to manage a structured document database. 5. The method of claim 4, wherein the order information corresponding to the tag stored in the document table includes order information of a tag token indicating a start and order information of a tag token indicating a last.