JP3508623B2 - Structured document management system and method, and recording medium - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a document management technique, and more particularly to an apparatus method and a recording medium for managing a structured document using a database.

[0002]

2. Description of the Related Art [Structured document] In a text document, a character string surrounded by symbols "<" and ">", that is, a tag is embedded in the document to form a partial document (referred to as "document element"). SGML (Standardized Gene) is used as a document creation means for creating and displaying documents and data analysis in document element units.
ralized Markup Language), XML (eXtensible M)
arkup Language) and HTML (HyperText Markup)
Language) is known. These SGML, X
A document created by ML or HTML is generally called a "structured document".

Definitions such as tag names of tags and relationships between tags are given by DTD (Document Type Definition), and the structure of the document element of the structured document follows the definition of the tag of this DTD. become.

A document element (elemen identified by a tag
t; also abbreviated as "element") is 0 when viewed from an element.
Or one parent element, that is, an element that contains itself,
It has zero or more child elements, that is, elements contained within itself, and this structure can be represented by a tree structure. In this specification, the tree structure of this document element is referred to as a “logical structure of a structured document”.

[Management of structured document by database]
Managing information about the document structure of structured documents in a database has several advantages. It is possible to take out not only a document but also a partial document (document element) in which the element identified by the tag in the document is a unit to be taken out from the database.

In addition to the full-text search and the search based on the bibliographic information of the document, the search means from the database can perform the search on the condition of the information on the logical structure of the structured document. That is, it is possible to extract a document that satisfies various conditions on the tree structure of an element or a document element thereof.

Such a structured document is stored in a database, an index is created for the document structure and the document itself, and a publication relating to a technique for managing and retrieving specific documents and document elements using this index, for example, Reference is made to the descriptions in JP-A-6-131340 and JP-A-10-240752.

Among them, in the above-mentioned Japanese Patent Laid-Open No. 6-131340, a document component receiving means for receiving a processing request as a document component (that is, an element) from a document processing device together with a registration request, and a logical structure by analyzing the received document component. , A document structure holding unit that holds one or more of the document parts, an index generation unit that generates an index of the specified logical structure, and an index holding unit that holds one or more generated indexes. And a structure of a document parts management device for a structured document having means, and storing the actual document file in a permanent secondary storage device and accessing the index to obtain the actual document. It becomes easy to search on the condition of the internal structure of the document without extracting the file. Is it possible to find the structure of the high-speed.

Further, in the above-mentioned Japanese Patent Laid-Open No. 10-240752, in the structure designation search for a structured document, a highly accurate structure designation search is made possible by designating the condition regarding the appearance position of the logical element in the document. As an issue to be made possible, when registering a document in the database, the logical structure of each registration target document is overlapped, and the structure index that represents the structural elements with the same appearance position in the document by a single metanode is used. By creating and searching for a set of metanodes that satisfy the specified structure condition by referring to the structure index at the time of document search, and searching the character string index using the identifiers of these metanodes as keys,
Disclosed is a method for providing a means for obtaining a document group satisfying a specified condition, thereby enabling a highly accurate structure-specified search on a document database including a set of structured documents.

[0010]

However, the apparatus and method described in the above publications have the following problems.

The method described in Japanese Patent Laid-Open No. 10-240752 is created by creating an index for managing the logical structure of a plurality of documents and superimposing the individual logical structures of the plurality of documents. ing. Specifically, when registering a document in the database, the logical structure of each registration target document is overlaid, and a structure index is created in which structure elements that have the same appearance position in the document are represented by a single metanode. At the time of document retrieval, a set of metanodes that satisfy the specified structural condition is obtained by referring to the structure index, and a character string index is searched by using the identifiers of those metanodes as keys to obtain a document group that satisfies the specified condition. .

However, in the method described in the above-mentioned Japanese Patent Laid-Open No. 10-240752, the structural index created by superposing the logical structures of a plurality of such documents is utilized at the time of retrieval, and the logical structure intended by the user is used. It is unclear whether it can be guaranteed that the documents that meet the conditions are always taken out.

In general, a structured document is a particular DTD.
Accordingly, documents having different DTDs are generally judged to be different types of documents.

Documents having the same element name will hardly exist if the DTDs are different, and even if an element having the same structure and the same name exists in the document, it has the same meaning based on the same usage. It is not always the one with.

The problem of the method described in Japanese Patent Laid-Open No. 10-240752 is that when a plurality of documents having different DTDs are to be registered, structuring elements having the same appearance position in the document are represented by a single metanode. It is actually impossible to create a structured index, and even if a metanode can be created, it is not possible to guarantee semantic uniqueness for the elements of the actual document.

Therefore, it is not possible to guarantee whether the document retrieved by such a search based on the structural index is a document retrieved based on the structural element that the user really intended.

That is, in the method described in JP-A-10-240752, the index is a single DTD,
Alternatively, it is an effective method for documents conforming to extremely similar DTDs, but it is a disadvantageous method for documents having greatly different DTD structures.

On the other hand, the apparatus described in the above-mentioned Japanese Patent Laid-Open No. 6-131340 is provided with index generating means for generating an index having a logical structure, and by accessing the index, without extracting the actual document file. , It becomes easy to search based on the internal structure of the document,
In addition, it is possible to search for a specific structure from a large number of documents at high speed, but this index structure contains registration information, part type, unique information by part type, etc. Has too much information.

In general, index information is loaded and calculated in the memory of a computer, but the memory resources of the computer are limited and the amount of index information that can be loaded at one time is also limited.

The smaller the information amount of the index for one document or one element, the more the calculation process by the index for more documents or elements can be executed on the memory, and as a result, the more documents there are. It is possible to quickly search for a specific structure in the group.

In the document search using the index, first of all, one of the major purposes is to quickly narrow down a target document group from a large number of document groups.

Even if the target document group is narrowed down with respect to some information contained in the index structure, and the information is retrieved and collated from the document, it should have little influence on the entire retrieval performance.

Therefore, the present invention has been made in view of the above problems, and an object thereof is to index a set of a plurality of documents having different structures such as different DTDs, and to the set of a plurality of documents. An object of the present invention is to provide an apparatus and method that uses an index at the time of search.

Another object of the present invention is to provide a method and apparatus for suppressing the amount of information as much as possible from the viewpoint of memory saving of an index and performing a search based on the logical structure of a structured document at a higher speed. Other objects, features, and the like of the present invention will be immediately apparent from the following description.

[0025]

According to the present invention to achieve the above object, a database schema includes a document class for managing documents, a document folder class for managing a set of documents, and a DTD class for managing DTDs. The document class is an element tree index that manages the tree structure relationship of document elements, the document folder class is a document element, and an element-attribute index that creates and manages an index for each attribute. The DTD class is an element name and attribute name. Each has a table for managing as an ID.

In the present invention, when storing a document in the database, the element tree index creating means of the document storing execution means, the element tree index, the element-attribute index creating means the element-attribute index, and the element-attribute ID table. The creating means respectively creates the element-attribute ID table, and gives the element name ID and the attribute name ID as numerical values.

In the present invention, the document retrieval means retrieves a structured document from the database using the element tree index, the element-attribute ID table, and the element-attribute index. When retrieving a document element, a parent element, an ancestor element for the document element, and a relative relationship from a document element in the structured document
The child element, the descendant element, the elder brother element, the younger brother element, the previous element, the next element, or a combination thereof is extracted.

[0028]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described. FIG. 1 is a diagram showing a configuration of an embodiment of the present invention. In one embodiment of the present invention, the database 1
A document class 11 that manages documents and a document folder class 12 that manages a set of documents as the schema classes of
And a DTD class 13 that manages the DTD. The document class 11 has an element tree index 14 for managing the tree structure relationship of the elements of the document, and the document folder class 12 has an element-attribute index 15 for creating and managing indexes for each element and attribute of the document. And the DTD class 13 has an element name and an attribute name
An element-attribute ID table 16 for managing

When storing a structured document (simply abbreviated as “sentence”) in the database 1, the document storage execution unit 2
1 is started, and when executing the search, the search execution unit 22 is started.

The document storage execution unit 21 includes an element tree index creation unit 211, an element-attribute index creation unit 212, and an element-attribute ID table creation unit 213, and when storing a document in the database 1, The element tree index creation unit 211 uses the element tree index 14, and the element-attribute index creation unit 212 uses the element-
The attribute index 15 and the element-attribute ID table creating unit 213 respectively create the element-attribute ID table 16.

The present invention provides a data storage structure for storing documents and elements in a database, provides a memory-saving index structure based on the storage structure, and retrieves elements based on the index structure. Provide a way. Further, the present invention can be used in a system in which a plurality of documents are classified and stored in a database, and a search is performed on a set of the classified documents. In particular, it accepts a search request in document and element units, with the logical structure of the document, that is, the relationship between each element in the document as a constraint condition, and makes it possible to extract the document and element units.

In order to perform the retrieval process at high speed, the memory resource of the computer is not so much consumed by minimizing the logical structure of the document and the data structure of the index that holds the information of the document element and the attribute of the element. To realize the search process.

[Database Schema for Document Management / Search] First, the database schema in the present invention will be described. FIG. 2 is a diagram showing classes related to the schema of the database. The notation shown in FIG. 2 is OM
It conforms to T (Object Modeling Technique).

In the present invention, storing a document in the database means creating a document to be stored by the document storage executing unit 21 and registering it as an instance of each class.

In the present invention, retrieving a document from the database means extracting an instance of each class satisfying the retrieval condition by the document retrieval executing unit 22 and retrieving the instance as a document from the system.

In order to store documents and DTDs in the database 1, the document class 11 is a class of its schema.
And prepare DTD class 13.

Each document and each DT stored in the database 1
D is an instance of document class 11, DTD
It is managed as an instance of class 13.

An instance of the document class 11 is called a "document object". Similarly, an instance of the document folder class 12 in FIG. 2 is called a “document folder object” and an instance of the DTD class 13 is called a “DTD object”.

An ID (identification information) that can be uniquely identified within the class is assigned to each instance of the class.

This ID may be given by the database management system, or may be set uniquely on the application side.

FIG. 3 is a diagram for explaining the present invention and is a diagram showing an example of the relationship between a document object and a DTD object.

In FIG. 3, (document) A111, (document)
B112 and (document) C113 are instances of the document class, and (DTD) X131 and (DTD) Y132 are instances of the DTD class 13. The database 1 manages these individual instances and
It also manages links between instances that indicate which DTD was created based on.

The document class 11 manages all the documents in the database 1, that is, the entire set of document objects. The document class 11 and the document objects 111, 112, 113 are is
A document folder class 12 is prepared as a means for managing a set of document objects, which constitutes the -a relationship.

The document folder class 12 manages a set of arbitrary document objects. In reality, a set of document objects satisfying a certain condition is managed by one document folder class 12 instance.

For example, there is a case in which documents are grouped by author or grouped by age group or publisher to manage the set of documents. In the present invention, the document folder class 12 and the document class 11 form a member-of relationship.

An instance of the document folder class 12,
That is, the document folder object has a structure for constructing a classification hierarchy based on a tree structure.

That is, all the document folder objects can have a relation with 0 or 1 document folder object corresponding to the parent and 0 or more document folder objects corresponding to the children.

In the example of the instance shown in FIG. 4, the (document folder) α121 is the parent and the (document folder) β122 is between the (document folder) α121 and the (document folder) β122, and the (document folder) γ123. (Document folder) γ123 has a parent-child relationship as its child.

Similarly, between (document folder) β122 and (document folder) δ124, (document folder) ε125, (document folder) β122 is a parent, and (document folder) δ124 and (document folder) ε125 are its children. Parent-child relationship, (document folder) γ12
Between (Document folder) ζ 126 and (Document folder) γ 12
3 is a parent, and (document folder) ζ 126 has a parent-child relationship in which it is a child.

As an application example, the document folder objects of the documents collected for each author are used as parents, and the document folder objects of the authors for men and women are created as their children to classify the documents, or the documents are classified according to ages. When classifying, for example, the document folder objects collected in the 1990s are used as parents, and the document folder objects for each year in the 1990s are used as children, and so on.

Although there is no strict rule on the method of actually providing a parent-child relationship, in general, a child-document folder object constitutes a parent document folder object, forming a part-of relationship.

[Element-attribute ID table] When a search condition for a logical structure in a structured document is specified and a specific document or element is extracted, character string data such as an element name and an attribute name included in the document are concretely specified. Must be specified in search conditions.

In order to retrieve the document or element that matches the search condition including these character string data, the character string data of the document object in the database is loaded on the memory of the computer, and the element name and attribute of the search condition are loaded. It is necessary to judge the name. However, the character string data such as the element name and the attribute name is generally much smaller in size (length) than the character string data of the entire document held by the document object. Loading on memory is not efficient. For this reason, an index that has only the information necessary for determining whether or not the search condition is satisfied, such as the element name and attribute name, is created in advance, and at the time of search, only the index is loaded into the memory of the computer for calculation. A method is used.

By the way, when the element name and the attribute name are treated as character string data in the index, the memory of the computer requires a data area corresponding to the character string length. Specifically, many computers have one character for one byte (one character for Kana and Kanji for two bytes). For this reason, if the element name or attribute name in the database is long, that much data on memory required for loading is needed, which is disadvantageous in terms of calculation performance.

Therefore, in the present invention, the element-attribute ID table 16 having a one-to-one relationship with the DTD class 13 is provided in order to save the memory consumption of character strings such as element names and attribute names in the index. There is.

The type and number of element names and attribute names handled in the structured document are limited by the DTD that the structured document follows. Within the scope of a single DTD, element names are uniquely defined, and attribute names are uniquely defined within the range of the elements of the DTD.

An element name ID that is unique to each element name defined in this DTD is given as a numerical value, and a table of the correspondence relationship between the element name and the element name ID is created.
This table is stored in the database as an instance of the element-attribute ID table 16 in a one-to-one relationship with the DTD object.

With respect to the attribute name as well, even if the element name is unique to the element in the DTD object and the same attribute name exists among different element names, an attribute name ID that can identify them is given by a numerical value. That is, the attribute name ID is
A numerical value is assigned so that the attribute name of the element can be determined from the value.

Then, the correspondence between the attribute name and the attribute name ID is created as a table, and the table is used as an instance of the element-attribute ID table 16 created earlier.
Store in database.

FIG. 8 shows an embodiment of the present invention.
It is a figure which shows an example of the instance in the element-attribute ID table 16. Numerical values may be assigned to the element name ID and the attribute name ID as long as the uniqueness in the DTD and the element is guaranteed. Element name ID for DTD
An example of how to give the attribute name ID will be described.

FIG. 5 shows an embodiment of the present invention.
It is a flow chart which shows a processing procedure which gives element name ID. However, the validity of the DTD handled here is guaranteed,
When the DTD description is divided into a plurality of documents by using the DTD internal subset, external subset, external entity, etc., the DTD parser (parser) of the structured document Are merged, and all the DTD information is described in one document. It is also assumed that the parameter entity has already been expanded by the DTD parser. Then, in the processing procedure shown in FIG. 5, the description of such a DTD is continuously read from the beginning to the end.

Referring to FIG. 5, the variable i is initialized in step S101, and a loop process between steps S102 and S107 is performed. In the processing in this loop, first, step S103.
With element type declaration in DTD (element type declaratio
n), and if an element type declaration is found, step S1
In 05, the element ID of the element name of the element declaration is set to the value of the upper digit i
Then, the value of the lower digit is set to 0, and the variable i is incremented by 1 in step S106. If the element type declaration is not found in the determination in step S104,
The process ends. The element type declaration defines the name of the document element and the hierarchical structure (parent-child, sibling relationship, etc.) of the element.

In FIG. 5, the value of the upper digit of the element name ID in step S105 means the value of the digit on the left side between the predetermined digits of the numeral digit, and the value of the lower digit thereof. The value of the right digit. For example, the value "1234
If the digits between the boundaries of "5678" are defined as the 4th and 5th digits, the value of the higher digit is "1234" and the value of the lower digit is "5678".

FIG. 6 is a flowchart showing a processing procedure for giving an attribute name ID.

The loop processing between step S111 and step S122 is performed by the attribute list declaration (attribute-list) in the DTD.
declaration) is processed, the element name ID corresponding to the element name in the attribute list declaration is obtained, and when the attribute list declaration is found, the variable i is initialized and the loop process is started to execute the attribute list declaration. If an attribute name is found by searching for the attribute name in the item, the attribute ID of the attribute name is set as the element name ID for the upper digit value and the lower digit value for i. In addition,
The attribute list declaration in the DTD defines an attribute as additional information of an element, and defines which attribute is attached to which element, an attribute name, a value that can be designated as an attribute, a default value, and the like.

Element name IDs of all element names of the DTD
After is allocated, the description from the beginning to the end of the DTD description is continuously read.

Attribute name ID in step S119 of FIG.
The meaning of the upper digit and the lower digit of is the same as that of the element name ID in step S105 of FIG. However, as a condition, the position of the boundary between the upper digit and the lower digit of the element name ID and the attribute name ID is the same. The number of digits of the numerical value of the element name ID and the attribute name ID to be prepared, and where to divide the high-order digit and the low-order digit, etc., are the element names and attribute names for which the above-mentioned uniqueness should be guaranteed in the DTD of the document. It will be decided by predicting how many will be.

It is assumed that the element name ID and the attribute name ID are 4 respectively.
If the upper digit and the lower digit are 2 bytes each, the number of element names in the DTD of the document that can guarantee the uniqueness is 65535 (2 ¹⁶ −1), which is also the same as the document. The number of attribute names that can guarantee the uniqueness is 65535 among the elements in the DTD. The data amount of 4 bytes corresponds to only 4 characters when converted into a character string. In fact, considering that the character string ends with a null character, the character string is actually three characters. When the element name or attribute name in the DTD is defined with more than three characters, the data area representing the element name or attribute name can be saved.

Next, the structure of the element-attribute ID table 16 in the embodiment of the present invention will be described.

As described above, the element-attribute ID table includes a table of correspondence relationships between element names and element name IDs, and attribute names and attribute name IDs. Therefore, element-attribute I
As the structure of the D table, at least (element name ID,
A data structure having two items of (element name) and (attribute name ID, attribute name) is prepared.

The order of data entry in the element-attribute ID table 16 is sorted in a lexicographical order of element names and attribute names, respectively, to make a list. This is to enable the corresponding element name ID and attribute name ID to be checked earlier by a binary search that compares the dictionary order of the element name and attribute name of each data entry.

The element-attribute ID table creating section 213 creates the element-attribute ID table having the above-mentioned data structure,
The data is stored in the database 1 while maintaining a one-to-one relationship with the DTD object.

FIGS. 7 and 8 are diagrams showing an example in which the element name ID and the attribute name ID are assigned according to the above procedure.
FIG. 7 shows an example of DTD. In FIG.
"! ELEMENT element name content model" is the element type declaration, and "! ATTLIST element name, attribute value candidate, default value" is the attribute list declaration. Also, CDATA means character data. #PCDATA in the element type declaration specifies mixed content.

FIG. 8 shows an example in which element name IDs and attribute name IDs are assigned corresponding to the element names and attribute names in the DTD shown in FIG. The element name ID and the attribute name ID are shown in hexadecimal display. In this example, the value of the element name ID and the value of the attribute name ID are 4 bytes in hexadecimal, and the upper digit and the lower digit are 2 bytes each.

[Element Tree Index] The element tree index 14 is a data structure that expresses the structure in each document, that is, the tree structure of the element, using the element name ID of the element-attribute ID table 16. The element tree index 14 is created and held for each document object.

The element tree index 14 is a list of data entries having a triple data structure of (element name ID, element level, element start position).

This triple data corresponds to the element in the document, and each item of the data has an element name ID corresponding to the element name of the element, a depth from the root element when the element has a tree structure, The value of the start byte position of the description (including the start tag) of the element in the document is stored.

Then, the order of these entries in the element tree index 14 is made equal to the order of appearance of the start tags of the elements in the document. In other words, the appearance order of the start tags is the left depth priority order when the tree structure of the element is formed.

A procedure for creating an element tree index will be described. The element tree index 14 is created by the element tree index creating unit 211.

FIG. 9 is a flowchart showing the procedure for creating the element tree index. The value of each item of the data structure (element name ID, element level, element start position) of the triple of the element tree index is obtained, and this is added to the list of the element tree index 14 as an entry.
However, it is premised that the document handled here is guaranteed to be valid for DTD, and the validity of the DTD itself is also guaranteed. It is also assumed that the omission of tags and omission of attribute values in a document are all complemented by the structured document parser. In the procedure for creating the element tree index shown in FIG. 9, the description of such a document is continuously read from the beginning to the end.

In step S131, the variable i is initialized to 0, and the loop processing between the loop a start end in step S132 and the end end in step S143 is performed.

In step S133, either the start tag or the end tag of the element is searched.

When the tag is found and it is the start tag (steps S134 and S135), the first character position of the start tag is obtained (step S136), the element name of the start tag is obtained (step S137), and the relation with the document object is obtained. Do DTD
Retrieve the element-attribute ID table from the object,
Obtain the element ID corresponding to the element name (step S138),
The element level value is used as the value of the variable i, and the triplet data of the element name ID, the element level, and the element start position is added to the entry of the element tree index (step S140), and then the variable i is incremented by 1 (step S140). S14
1).

On the other hand, in step S135, if the tag is the end tag, one variable is decremented (step S142).

If the tag is not found in step S134, the loop is exited and the process ends.

An example of the element tree index is shown. Figure 7
It is assumed that there is a document as shown in FIG. 10 as a document complying with the DTD. The numbers on the left side of the document do not indicate the line numbers of the document, but the start tag of the element appearing on the right side of the number, the first character position of the string, and the last character position of the end tag. .

The relationship between the elements identified by the tagging in the document shown in FIG. 10 is represented by a tree structure as shown in FIG.

In the tree structure of FIG. 11, the label on the node represents the element name, and the subscript of the node represents the start position of the element corresponding to the label of the node.

The appearance order of the start tags of the elements of the document of FIG. 10 is the left depth priority order of the tree structure of FIG.

According to the order of the elements, a list of triplet data (element name ID, element level, element start position) is created. The result is as shown in FIG.

[Search Based on Logical Structure of Document] In order to search based on the logical structure of a document, the conditions regarding the element name, the attribute name, the attribute value of the document, the contents of the document, and the relation between the elements are satisfied. Including whether to determine.

For example, for the document shown in FIG. 10, the document element name, the attribute name, the attribute value, and the search condition "the element name is D and the attribute a2 value of the element is 2" for the document of the content of the element.
Then, the condition judgment of "a document having an element that contains the character string" string1 "" in the content of the element is TRUE.

Further, in the search condition, the condition “relationship between elements” “attribute a of an element having such an element D as a child element
Adding "a document having an element whose value of 1 is 1" is also within the scope of the search based on the logical structure of the document. The condition determination of the document in FIG. 10 to which this search request is added is TRUE.

As can be seen from the above example, the process of determining whether or not the condition regarding the relationship between elements is satisfied is the extraction of an element that is in a relative relationship from a certain element, such as the parent element or ancestor of an element. It is necessary to take out elements, child elements, descendant elements, elder brother elements, younger brother elements, previous elements, and next elements.

For an explanation of these relative relationships, see
References (ancestor, child, descendant, psibling, f of relative location terms in XML Pointer Language (XPointer)
Descriptions such as sibling, preceding, following) are referred to.

The identification of elements relating to such a relative relationship can be calculated starting from an arbitrary entry of the element tree index.

13 to 18, steps for extracting a parent element, an ancestor element, a child element, a descendant element, a brother element, a brother element, a previous element, and a next element from a certain element using the element tree index, respectively. 21 and 22 are shown as a flow chart.

[Retrieval of Parent Element] Referring to the flowchart of FIG. 13, the procedure for retrieving the parent element will be described.

Referring to FIG. 13, in step S161, variables are initialized. The variables parent and elem are variables for storing the values of the triplet entry of the element tree index, the element name ID, the element level, and the element start position.

The purpose of each variable is that parent is used to store the entry of the parent element, and the variable elem
Is for storing the current entry of the element tree index in the process of calculation.

The variable base is the element that is the starting point of the relative relationship.
Stores the element-level value of the entry (starting element).

The determination as to whether or not there is the previous element in step S162 is whether or not there is an entry before the entry serving as the starting point in the list of entries in the element tree index.

Element tree index 1 shown in FIG.
A description will be given according to the example of No. 4. For example, the entry that is the starting point
Given the third entry (00030000, 2, 30),
"The element one before" means the one before the entry,
This is the second entry (00020000, 1, 20).

Therefore, in the case of this example, it is determined that “the previous element exists”, and the process proceeds to step 63.

In step S163, the entry corresponding to the "previous element" in step S162 is assigned to the variable elem.

In step S164, the elem of step S163.
It is determined whether the element level of is less than the value of base set in the initialization of step 61 by one.

If the result of the determination in step S164 is YES,
The current elem becomes the entry of the parent element to be obtained, and the value of elem is assigned to parent in step S165, and the process ends.

If the decision result in the step S164 is NO,
Returning to step S162, it is determined whether there is an element immediately preceding the elem, and thereafter, the same processing as described above is repeated.

[Retrieval of Ancestor Element] Next, the processing procedure for retrieval of an ancestor element will be described with reference to the flowchart of FIG.

In step S171, necessary variables are initialized. The variables ancestor and elem are variables for storing the value of the entry of the element tree index. The purpose of each variable is to store the entry of the ancestor element that ancestor asks for in the flowchart,
The variable elem is for storing the current entry of the element tree index in the process of calculation.

The variable base stores the element level value of the entry of the starting element. The variable n is the value specified to retrieve the nth ancestor element. The variable i is used as a counter for the iterative process.

From step S172 to step S174,
This is the same as steps S162 to S164 of FIG.

If the decision result in the step S174 is YES,
The current elem becomes the entry for the i th ancestor element.

Then, in step S175, it is determined whether or not the i-th ancestor element elem is the n-th ancestor element to be obtained, and YES.
If so, the value of elem is assigned to ancestor in step S177, and the process ends.

On the other hand, if the decision result in the step S175 is NO, a step S is carried out to find a parent element for the elem.
At 176, the value of bese is newly set to the element level of the elem, and the value of i is incremented by one, and the process returns to step S172 and is repeated.

[Retrieval of Child Element] Next, with reference to the flow chart of FIG. 15, the procedure for extracting a child element will be described.

In step S181, necessary variables are initialized. The variables child and elem are variables for storing the value of the entry of the element tree index. The purpose of each variable is to store the entry of the child element that child wants in this flowchart.
Is for storing the current entry of the element tree index in the process of calculation.

The variable base stores the element level value of the entry of the starting element. The variable n is the value specified to retrieve the nth child element.

The variable i is used as a counter for the repetitive processing.

The “whether there is an element after one” in step S182 means whether or not there is an entry after the entry serving as the starting point in the list of entries in the element tree index.

Explaining with the example of the element tree index shown in FIG. 12, if the entry serving as the starting point is the third entry (00030000, 2, 30), the element after the one is This is the fourth entry (00040000, 3, 40) after the one. Therefore, in the case of this example, it is determined that there is an element after one, ”and the process proceeds to step 83.

At step S183, the entry corresponding to the element immediately after that at step S182 is assigned to the variable elem.

In step S184, the elem of step S183
Element level of bas set in the initialization of step S181
It is determined whether the value is larger than the value of e.

If the result of the determination in step S184 is YES,
It proceeds to step S185.

If the result of the determination in step S184 is NO, the process ends. In step S185, further elem
It is determined whether the element level of is a value one more than the value of base.

If the decision result in the step S185 is YES,
The current elem becomes the i-th child element entry. Then, in step S186, it is determined whether or not the i-th child element elem is the n-th child element to be obtained, and if YES, step S188
Substitute the value of elem for child and finish the process.

If the decision result in the step S186 is NO,
In order to obtain the next child element, the value of i is incremented by 1 in step S187, the process returns to step S182 and the process is repeated again.

[Retrieval of Descendant Element] The processing procedure for retrieval of a descendant element will be described with reference to the flowchart of FIG.

In step S191, necessary variables are initialized. The variables descendant and elem are variables for storing the value of the entry of the element tree index. The purpose of each variable is that the descendant is for storing the entry of the descendant element to be obtained in this process.
elem is for storing the current entry of the element tree index in the process of calculation.

The variable base stores the element level value of the entry of the starting element.

The variable n is a value designated to fetch the n-th descendant element.

The variable i is used as a counter for the repetitive processing.

From step S192 to step S194,
This is the same as steps S182 to S184 in FIG.

If the decision result in the step S194 is YES,
The current elem is the entry for the i th descendant element.

Then, in step S195, it is determined whether or not the i-th descendant element elem is the n-th descendant element to be obtained, and YES.
If so, in step 97, the value of elem is assigned to descendant and the process ends.

If the decision result in the step S195 is NO,
In order to obtain the next descendant element, the value of i is incremented by 1 in step S196, and the process returns to step S192 to repeat the process.

[Retrieval of Brother Element] Next, with reference to the flow chart of FIG. 17, the procedure for extracting the brother element will be described.

In step S201, necessary variables are initialized. The variables psibling and elem are variables for storing the value of the entry of the element tree index.

The use of each variable is that psibling is for storing the entry of the sibling element obtained in this processing procedure, and variable elem is for storing the current entry of the element tree index in the process of calculation. Is.

The variable base stores the element level value of the entry of the starting element.

The variable n is a value designated to take out the n-th brother element.

The variable i is used as a counter for repeating processing.

From step S202 to step S203,
This is the same as steps S162 to S163 in FIG.

At step S204, the elem of step S203.
Element level of bas set in the initialization of step S201
It is determined whether it is the same as the value of e.

If the decision result in the step S204 is YES,
The current elem is the entry for the i-th sibling element.

Then, in step S206, it is determined whether or not the i-th brother element elem is the n-th descendant element to be obtained, and if YES, the value of elem is substituted in psibling in step a8,
The process ends.

If the decision result in the step S206 is NO,
In order to obtain the next brother element, the value of i is incremented by 1 in step S207, the process returns to step S202, and the process is repeated again.

If the decision result in the step S204 is NO,
In step S205, it is determined whether the element level of elem in step S203 is larger than the value of base.

If the decision result in the step S205 is YES,
Returning to step S202, the process is repeated once again. If the determination result is NO, the process ends.

[Retrieval of younger brother element] Next, with reference to the flow chart of FIG. 18, a processing procedure for retrieving a younger brother element will be described.

In step S211, variables are initialized. The variables fsibling and elem are variables for storing the value of the entry of the element tree index.

The usage of each variable is the variable psibling.
Is for storing the entry of the younger brother element obtained in this process, and the variable elem is for storing the current entry of the element tree index in the process of calculation.

The variable base stores the element level value of the entry of the starting element. The variable n is the value specified to retrieve the nth sibling element.

The variable i is used as a counter for the repetitive processing.

From step S212 to step S213,
This is similar to steps S182 to S183 in FIG.

In step S214, the elem of step S213.
Element level of bas set in the initialization of step S211
It is determined whether it is the same as the value of e.

If the decision result in the step S214 is YES,
The current elem is the entry for the i-th sibling element.

Then, in step S216, it is determined whether or not the i-th younger brother element elem is the n-th descendant element to be obtained, and if YES, the value of elem is assigned to fsibling in step S218, and the processing is terminated.

If the decision result in the step S216 is NO,
In order to obtain the next brother element, the value of i is incremented by 1 in step S217, the process returns to step S212 and the process is repeated again.

If the decision result in the step S214 is NO,
In step S215, it is determined whether the element level of elem in step S213 is larger than the value of base.

If the decision result in the step S215 is YES,
The process returns to step S212 to repeat the process again. If the determination result is NO, the process ends.

[Fetching the youngest younger brother element] Next, when there are no more younger brother elements than a certain younger brother element, the younger brother element is called the younger brother element. The "Retrieve youngest brother element" process is shown in "Retrieving descendant elements in right-depth-first order" shown later.
It is a necessary sub-process.

With reference to the flowchart in FIG. 19, the processing procedure for taking out the younger brother element will be described.

In step S221, variables are initialized. The variables fsibling and elem are variables for storing the value of the entry of the element tree index. The use of each variable is to store the entry of the youngest sibling element obtained in this process, psibling, and to store the current entry of the element tree index in the process of calculation, as the variable elem. The variable base stores the element level value of the entry of the starting element.

From step S222 to step S224,
This is the same as steps S212 to S214 in FIG.

If the decision result in the step S224 is YES,
The current elem is the entry of the younger brother element, and step S225
Assign the value of elem to fsibling. However, this fsibli
Since it is unknown at this point whether or not ng is the youngest brother, the process returns to step S222 and the process is repeated once again.
The value of sibling is updated sequentially.

If the decision result in the step S224 is NO,
In step S226, it is determined whether the element level of elem in step S223 is greater than the value of base.

If the decision result in the step S226 is YES,
The process returns to step S222 and the process is repeated again. If the determination result is NO, the process ends.

[Retrieval of descendant elements in right depth priority order]
The process of fetching the descendant elements in the right-depth priority order is a sub-process required for fetching the previous element, which will be described later.

Here, the right depth priority order is the order obtained when the elements are traced in depth priority from right to left in the tree structure shown in FIG.

In the document shown in FIG. 10, the appearance order of the start tags of the elements matches the order obtained when tracing the elements from the left to the right in the depth priority in the tree structure. Therefore,
The order of the elements fetched in the right depth priority order is different from the order of appearance of the start tags of the elements of the document.

With reference to the flow chart shown in FIG. 20, description will be given of a processing procedure for extracting descendant elements in the right depth priority order.

In step S231, variables are initialized. The variables descendant and elem are variables for storing the value of the entry of the element tree index. The purpose of each variable is that the descendant is for storing the entry of the descendant element to be obtained in this process, and the variable elem
Is for storing the current entry of the element tree index in the process of calculation. The variable n is the value specified to retrieve the nth descendant element. The variable i is used as a counter for the iterative process.

In step S232, the subprocess "take out youngest younger brother element" is called, and the entry of the youngest younger brother element of the element elem is taken out.

In step S233, it is determined whether or not the youngest younger brother element can be extracted. If the determination result is YES, the entry of the extracted element becomes the entry of the i-th descendant element.

In step S234, the entry of the fetched element is assigned to elem, and in step S235, it is determined whether or not the i-th descendant element elem is the n-th descendant element to be obtained.

If the decision result in the step S235 is YES, a step S242 substitutes the value of elem for the descendant, and the process ends.

If the decision result in the step S235 is NO, the process advances to a step S236. Further, the determination result of step S233 is NO.
If so, the process ends.

In step S236, the value of i is incremented by one and another iteration processing counter j is newly prepared and initialized to 1.

In step S237, the new variable values i and j are added.
It is a recursive call of "Retrieving descendant elements in right-depth-first order".

In step S238, it is determined whether or not the descendant element has been extracted. If the determination result is YES, the entry corresponding to the element extracted in step S239 is assigned to elem, and elem is the i-th descendant element. Will be the entry.

Then, in step S240, it is determined whether or not the i-th descendant element elem is the n-th descendant element to be obtained, and YES.
If so, the value of elem is assigned to descendant in step S242, and the process ends.

If the decision result in the step S240 is NO,
In order to obtain the next descendant element, the values of i and j are incremented by 1 in step S241, the process returns to the recursive call in step S237, and the same processing is repeated.

If the decision result in the step S238 is NO,
The process proceeds to step S243 to call the sub-process "take out brother element".

In step S244, it is determined whether or not the first brother element for the elem has been taken out. If the decision result is YES, the entry corresponding to the element taken out in step S245 is assigned to elem, and elem is i It becomes the entry of the th descendant element.

Then, in step S246, it is determined whether or not the i-th descendant element elem is the n-th descendant element to be obtained, and YES.
If so, the value of elem is assigned to descendant in step dg, and the process ends.

If the decision result in the step S246 is NO,
In order to obtain the next descendant element, the value of i is incremented by 1 in step S236, the value of j is initialized to 1 again, and step S23
Return to the recursive call of 7 and repeat the same process.

[Removal of Previous Element] Next, with reference to the flow chart of FIG. 21, a processing procedure for extracting the previous element will be described.

At step S251, variables are initialized. The variables preceding and elem are variables for storing the value of the entry of the element tree index.

The usage of each variable is that the preceding is
This is for storing the entry of the previous element obtained in this process, and the variable elem is for storing the current entry of the element tree index in the process of calculation. The variable n is the value specified to retrieve the nth previous element. The variable i is used as a counter for the iterative process.

In step S252, the sub-process "take out older brother element" is called to take out the entry of the first older brother element of the element elem.

In step S253, it is determined whether or not the older brother element has been extracted. If the determination result is YES, step S259
Assign the entry corresponding to the element fetched in step to elem, and enter el
em is the entry of the i-th previous element.

Then, in step S260, it is determined whether or not the i-th previous element elem is the n-th previous element to be obtained, and if YES, the value of elem is substituted into the preceding in step S267, and the processing ends. .

If the decision result in the step S260 is NO, the process advances to a step S261. In addition, the determination result of step S253 is NO.
If so, the process proceeds to step S254.

In step S261, the value of i is incremented by one and another iteration processing counter j is newly prepared and initialized to 1.

At step S262, the new variable value i,
Invokes the subprocess "get descendant elements in right-depth-first order" with j.

In step S263, it is determined whether or not the descendant element has been extracted. If the determination result is YES, step S2
Assign the entry corresponding to the element fetched in 64 to elem,
elem is the entry of the i-th previous element.

Then, in step S265, it is determined whether or not the i-th previous element elem is the n-th previous element to be obtained, and if YES, the value of elem is substituted for the preceding in step S267, and the processing is terminated.

If the decision result in the step S265 is NO,
In order to obtain the next previous element, the values of i and j are incremented by 1 in step S266, and the process returns to step S262 to repeat the same processing.

If the decision result in the step S263 is NO,
The process returns to step S252 to repeat the process again.

In step S254, the subprocess "take out parent element" is called to take out the entry of the parent element of the element elem. In step S255, it is determined whether or not the parent element has been extracted. If the determination result is YES, step S2
Assign the entry corresponding to the element fetched in 56 to elem,
elem is the entry of the i-th previous element.

Then, in step S257, it is determined whether or not the i-th previous element elem is the n-th previous element to be obtained, and if YES, the value of elem is substituted in the preceding in step S267, and the processing is ended.

If the decision result in the step S257 is NO, the value of i is incremented by 1 in a step S258 in order to obtain the next preceding element, and the process returns to the step S252 to repeat the process again.

If the decision result in the step S255 is NO, the process ends.

[Retrieval of Next Element] Next, with reference to the flow chart of FIG. 22, the processing procedure of retrieval of the next element will be described.

In step S271, variables are initialized. The variables following and elem are variables for storing the value of the entry of the element tree index. Describing the usage of each variable, following is for storing the entry of the next element obtained in this process, and variable elem is for storing the current entry of the element tree index in the process of calculation. . The variable n is the value specified to retrieve the nth next element. The variable i is used as a counter for the iterative process.

In step S272, the subprocess "take out younger brother element" is called to take out the entry of the first younger brother element of the element elem.

[0208] In step S273, it is determined whether or not the younger brother element has been extracted. If the determination result is YES, step S279
Assign the entry corresponding to the element fetched in step to elem, and enter el
em is the entry of the i-th next element.

Then, in step S280, it is determined whether or not the i-th next element elem is the n-th next element to be obtained, and if YES, the value of elem is substituted for following in step S287, and the processing ends. .

[0210] If the decision result in the step S280 is NO, the process advances to a step S281. In addition, the determination result of step S273 is NO.
If so, the process proceeds to step S274.

In step S281, the value of i is incremented by one, and another iteration processing counter j is newly prepared and initialized to 1.

At step S282, the new variable value i,
Invoke the subprocess "Extract descendant elements" with j. In step S283, it is determined whether or not the descendant element has been extracted, and if the determination result is YES, it is substituted into the entry elem corresponding to the element extracted in step S284, and elem becomes the entry of the i-th next element.

Then, in step S285, it is determined whether or not the i-th next element elem is the n-th next element to be obtained, and if YES, the value of elem is substituted for following in step S287, and the processing ends.

If the decision result in the step S285 is NO,
In order to obtain the next next element, the values of i and j are incremented by one in step S286, and the process returns to step S282 to repeat the same processing.

If the decision result in the step S283 is NO,
The process returns to step S272 and the process is repeated again.

In step S274, the subprocess "take out parent element" is called, and the entry of the parent element of the element elem is taken out.

In step S275, it is determined whether or not the parent element has been extracted. If the determination result is YES, step S276
Assign the entry corresponding to the element fetched in step to elem, and enter el
em is the entry of the i-th next element.

Then, in step S277, it is determined whether or not the i-th next element elem is the n-th next element to be obtained, and if YES, the value of elem is substituted for following in step S287, and the processing ends.

If the decision result in the step S277 is NO,
In order to obtain the next next element, the value of i is incremented by one in step S278, and the process returns to step S272 to repeat the process. If the decision result in the step S275 is NO, the process ends.

[Element-attribute index] The element-attribute index 15 is created and held for each document folder object, and the element name, the attribute name, and the attribute value are selected from the document set held by the document folder object. , Used to search for documents or elements.

The element-attribute index is (DTDI
D, element-attribute name ID, attribute value, document ID, element start position) is a list of data entries having a five-element data structure.

The element-attribute name ID will be described. The element-attribute name ID is an element name ID or an attribute name ID. As described above, the element name ID is the only ID in the DTD object. The attribute name ID is also unique to the element name in the DTD object, and is an ID that can identify the same attribute name even if different element names exist.

When an element-attribute index 15 entry is created, if no element in the document has an attribute,
Create one entry for that element and
The value of the attribute name ID is used as the element name ID of the element.

When the element in the document has one or more attributes, the entries corresponding to the attributes are created, and the value of the element-attribute name ID corresponding to each attribute is set to the attribute name ID of the attribute. And Since the element having the attribute name ID can be determined from the attribute name ID, it is possible to specify the element corresponding to the entry from the entry having the attribute name ID.

[DTDID and Document ID] The DTDID and document ID will be described. DTDID and Document I
D is an object ID uniquely identifiable in each class, which is assigned to each instance of the DTD class and the document class in the database. This number may be given by the database management system, or may be given uniquely on the application side.

The values of the DTDID and the document ID in the entry of the element-attribute index 15 are the element-attribute name ID.
It is the document ID of the document having the element corresponding to, and the DTD ID of the DTD to which the document complies.

[Attribute Value] The attribute value will be described. The attribute value is literally the value of the attribute. The value of the attribute value in the entry of the element-attribute index 15 becomes the attribute value of the attribute corresponding to the element-attribute name ID.

[Element Start Position] The element start position will be described. The element start position is the element tree index 1
Same as element start position in 4 and description of element in document
The value of the start byte position (including the start tag). Element −
The value of the element start position in the entry of the attribute index becomes the element start position of the element corresponding to the element-attribute name ID.

A procedure for creating the element-attribute index 15 will be described. The element-attribute index creation unit 212 creates the element-attribute index.

23 and 24 are flowcharts showing the procedure for creating the element-attribute index 15, and the data structure of the five-piece element-attribute index 15 structure.
(DTDID, element-attribute name ID, attribute value, document ID,
The procedure of obtaining the value of each item of (element start position) and adding the created entry to the list of the element-attribute index while maintaining the order sorted by the value of (DTDID, element-attribute name ID) is shown. There is.

However, it is assumed that the document handled here is guaranteed to be valid for DTD, and the validity of the DTD itself is also guaranteed. Also omit tags in the document,
It is assumed that the omission of attribute values, etc. are all complemented by the structured document parser. Then, in the processing shown in FIGS. 23 and 24, it is assumed that the description of such a document is continuously read from the beginning to the end.

Referring to FIG. 23, steps S301 and S3
When the document object is added to the document folder object in 02, the DTD object related to the document object is also added in steps S303 and S304.

Then, based on the information of the element-attribute ID table having a one-to-one relationship with the DTD object, in the following steps, an element-attribute index entry for each element of the added document object is created / added. I will do it.

In step S312, the sub process shown in FIG. 25 is called. Based on (DTDID, element name ID) given in step S312 of FIG. 23, in step S331 of FIG. 25, a list of entries having that (DTDID, element name ID) is searched.

In step S331, a sub process whose flow chart is shown in FIG. 26 is called.

From the element-attribute index list, the same DTDI is set by using the start position and end position for all entries of the element-attribute index and the DTDID value given in step S331 of FIG.
Find the list of entries with a value of D.

In step S332 of FIG. 25, it is determined whether the entry list has been found.

If found, step S35 in FIG.
Since the start position and end position of the entry list are returned in step 4 and step S361, the same element-attribute name ID value is set for the entry list of the start position and end position in step S333 of FIG. Find the list of entries you have.

In step S334, it is determined whether or not the entry list is found. If found, the insertion position on the list is returned in step S355 of FIG. 26, so that the insertion position is determined in step S335 of FIG. return.

If the target entry list is not found in steps S332 and S334 of FIG. 25, the insertion position on the list is returned in step S348 of FIG. 26. Therefore, the insertion position is returned in step S335. To do.

Similarly in step S320 of FIG.
Calling 5 subprocesses. However, the data to be given is (DTDID, attribute name ID), and the subsequent processing is the same as in the case of (DTDID, element name ID) described above.

By the above processing, the entry of the element-attribute index 15 is created, and the order of the entries is sorted by the value of (DTDID, element-attribute name ID).

The purpose of sorting the entries of the element-attribute index 15 by the value of (DTDID, element-attribute name ID) is to search for the entry (DTDID,
This is because it enables a binary search with the element-attribute name ID) as a key.

FIG. 26 shows the actual binary search procedure.
The sub-process whose flowchart is shown in FIG.

FIG. 27 is a diagram showing an entry search procedure in the element-attribute index. Step S37
In 1, the sub process shown in FIG. 26 is called.

A list of entries having the same DTDID value is selected from the element-attribute index list by using the start position and end position of all the element-attribute index entries given in step S371, and the DTDID value as the target data. look for.

In step S372, it is determined whether the entry list has been found.

If found, step S35 in FIG.
Since the start position and the end position of the entry list are returned in step 4 and step S361, the entry list having the same element-attribute name ID is further targeted to the entry list of the start position and the end position in step S373. Find the list.

In step S374, it is determined whether or not the entry list is found. If found, the above-mentioned D
Similar to the case of TDID, the start position and end position of the corresponding entry list are returned.

The loop processing from step S375 to step S379 is applied to each of the returned entries.

With respect to each entry, the condition judgment for the attribute value is performed in step S376, and the entry which is YES as a result of the judgment in step S377 is returned in step S378.

There are various possible conditions for the attribute value in step S376. For example, arithmetic determination when the attribute value is a numeric value, character string matching when the attribute value is a character string, Partial match and the like.

In the embodiment of the present invention, the array tree index creation unit 21 of the document storage execution unit 21.
1, the processing steps of the element-attribute index creation unit 212, the element-attribute table creation unit 213, and the search execution unit 22 are executed by a program executed by a computer. For example, FIG. 5, FIG. 6, FIG. 9, FIG.
According to the flowchart shown in 7 etc., create a program in the desired programming language and create an execution module,
The present invention is implemented by reading the program from a computer-readable recording medium that records the source program or the execution module or a communication medium that can be communicatively connected to the computer, and installing and executing the program in the computer. be able to.

[0254]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS As one embodiment of the present invention, an example of document retrieval based on the document storage structure of the present invention will be described.

[Search for Document and Its Element Using Element-Attribute Index] As shown as an instance diagram in FIG. 28, there is one document folder object, and the document folder object is three document objects A, B, C.
Document objects, and document objects individually
Assume that it is associated with objects X and Y.

The document IDs of the document objects A, B, and C are 1, 2, and 3, respectively, and the DTDIs of the DTD objects X and Y are set.
D is 1 and 2, respectively.

The document folder object also manages DTD objects X and Y.

FIG. 29 is a diagram showing an example of the contents of the document objects A, B, and C. The numbers on the left side of FIG. 28 do not represent the line numbers of the document, but the first character position of the start tag and the last character position of the end tag of the elements appearing on the right side of the numbers, respectively. I shall.

FIGS. 30 and 31 are views showing the element-attribute ID tables managed by the DTD objects X and Y, respectively.

When the above document objects A, B, and C are registered in the document folder object, the document folder object follows the procedure shown in FIGS.
2. Create the element-attribute index shown in 2.

Example of Search Request The procedure of the search process of the search execution unit 22 for the element “patient”, which is a descendant element of the element “disease record” and the attribute “pID” of which is 1000 or more, will be described.

Step 1: Retrieval of element "Patient" whose attribute "pID" value is 1000 or more:

A) The document folder object loads the element-attribute ID table of the DTD objects X and Y managed by itself, and the DTD IDs of X and Y.

B) From the element-attribute ID table of the DTD objects X and Y, the element name ("" used in the search request)
"Patient", "disease record"), element name ID of attribute name ("pID"),
Check the attribute name ID.

DTD object X (DTDID is 1)
For:-"Patient" = 00030000, "pID" = 00030001, "Disease Record" = 00
040000-For DTD object Y (DTDID is 2):-"Patient" = 00040000, "pID" = 00040001, "Disease record" = 00
050000-Since the upper 2 bytes of the ID of "patient" and "pID" have the same value, it can be seen that the attribute "pID" is the attribute of the element "patient".

C) In the element-attribute index held by the document folder object, (DTDID, "pI
The data satisfying the condition that the attribute name ID of D ") and the attribute value is 1000 or more are searched according to the search search procedure shown in FIG.

At this time, the search condition of step S376 of FIG. 27 is "attribute value is 1000 or more". As a result, the following entries are returned.

-(1, 00030001, 1200, 1, 20) -(1, 00030001, 1200, 1, 40) -(2,00040001,1500,3,40)

That is, in FIG. 33, the elements marked with "○" correspond to the search results. The part marked with "x" represents the element "patient" that did not satisfy the attribute value condition of 1000 or more.

Step 2. Element "patient" is element "disease record".
Check if it is a descendant element of: a) Load the document object from the document ID of each data, and load the element tree index of each document object.

FIG. 34 shows the element tree index of the document object of document A (document ID is 1).
FIG. 9 is a diagram showing an element tree index of a document object of document C (document ID is 3).

These element tree indexes are shown in FIG.
It is created in advance by the element tree index creation procedure shown in.

B) Based on the element tree indexes of FIGS. 34 and 35, the attribute "pID" obtained in step 1c)
Calculates whether or not the element "patient" with 1000 or more is a descendant element of the element "disease record". This calculation process corresponds to the search based on the logical structure of the document.

This calculation is performed by extracting the descendant elements shown in FIG. 16 to the descendant elements of the element "disease record" in step 1
Check whether the element "patient" obtained in c) exists.

This checking method is performed by the third data item "element start position" in the element tree index.
And the value of the fifth data item “element start position” of the element-attribute index is the same value if it is the same element, so in the “element start position” of the descendant element of the element “disease record” Can be determined by checking whether or not the value of the "element start position" of the entry of the element-attribute index obtained in step 1 c) is included.

As a result of the above, the following entries become descendant elements of the element "disease record".

-(1, 00030001, 1200, 1, 40) -(2,00040001,1500,3,40)

That is, in FIG. 36, the element of the part marked with "○" corresponds to the final search result.

C) From the document object of each document ID and its element start position, the content of the element which is the search result is extracted and returned.

[0280]

As described above, the present invention has the following effects.

The first effect of the present invention is that in the search for a set of a plurality of structured documents having different structures such as different DTDs, the search based on the structure of each document can be correctly executed.

Conventionally, it is possible to perform a structure-based search for one document, and for a structure-based search for a plurality of documents, for example, Japanese Patent Laid-Open No. 10-240752 discloses a structure that represents the structure of a plurality of documents. Although a search using an index has been proposed, as described above, in the method described in the above-mentioned Japanese Patent Laid-Open No. 10-240752, the structure index does not always correctly represent the structure of each document. However, even if a search is performed using the structure index, there is no guarantee that a search result that correctly reflects the logical structure condition of each document will be returned.

According to the present invention, the DTD in which the document structure information is described is linked and managed for each document, and in the search based on the document structure, the DTD information for each document is referred to. , Realizes a search based on the correct logical structure of each document.

The second effect of the present invention is to improve the search performance by providing a method of creating and using an index for use in the search based on the structure.
That's what it means.

In a search request based on the logical structure of a document, the element name or attribute name of the document is often designated. Since element names and attribute names are usually represented by character strings, the above-mentioned JP-A-6-131
In the device described in Japanese Patent No. 340, a character string of the element name or the attribute name is directly handled as an index, but the problem of handling a character string as an index is that the character string is generally variable length, and the document element It is impossible to predict in advance the required index capacity, and if the character string is long, the index capacity is required accordingly.

According to the present invention, the DTD class is provided with an element-attribute ID table as the first index, and attention is paid to the fact that element names and attribute names are uniquely determined within a certain DTD range. Element-attribute I by giving the element name ID and the attribute name ID corresponding to the name and the attribute name
It is held and managed as a D table, and element name IDs and attribute name IDs are given as numerical values. For example, when given as int type (integer type) data in a 32-bit calculator, the capacity required for the element name ID and attribute name ID is a fixed length of 32 bits, and if 32 bits are converted to half-width characters
It's 4 characters long, so you can save memory of index.

The memory saving of the index means that information on more documents can be loaded into the memory of the computer at one time, and as a result, the document search performance is improved.

An element tree index is provided as a second index, and the element tree index is an index representing the relationship between elements in each document, and for each element (element name ID, element level, element start position). It manages the structure of the document by giving the triplet data of. Since the capacity required for each entry of the element tree index can be expressed numerically for each item, if given as int type data in a 32-bit calculator, 96 bits and 12 bytes are provided for each entry. . That is,
The element tree index is also a memory-saving index.

Also, by using the element tree index, extraction of a relative element such as a parent element, an ancestor element, a child element, a descendant element, a brother element, a brother element, a previous element, and a next element from a certain element. Since it is possible, it can be said that the 12-byte structure is a necessary and sufficient amount of information as an index for responding to a search request based on the structure.

In the present invention, an element-attribute index is provided as the third index. The element-attribute index is an index that holds element and attribute information for a plurality of documents as an index of the document folder class of the present invention. The structure of the index is (DTDID, element-attribute name ID, attribute value, document I
D, element start position), and the element name ID or attribute name ID is used in the second item "element-attribute name ID",
The size is kept small and the length is fixed rather than using the character string as it is.

Also, from the first item "DTDID" to D
By referring to the element-attribute ID table of the TD object, it is possible to determine the search condition based on the element name and attribute name defined in each DTD.

As described above, since each entry has a relationship with the information regarding DTD, it is possible to handle a plurality of documents having different DTDs by a single index.

It is also possible to specify a document object from the document ID of the element-attribute index, perform condition determination based on the structure using the element tree index, and immediately extract the corresponding element from the element start position. .

The entries of the element-attribute index are in the order sorted by (DTDID, element-attribute name ID).

As a result, it is possible to search for an entry by a binary search using (DTDID, element-attribute name ID) as a search key, and it is possible to retrieve a target document or element at a higher speed.

[Brief description of drawings]

FIG. 1 is a diagram showing a configuration of an embodiment of the present invention.

FIG. 2 is a diagram showing classes related to a database schema according to the embodiment of the present invention.

FIG. 3 is a diagram showing an example of a link relationship between a document instance and a DTD instance according to an embodiment of the present invention.

FIG. 4 is a diagram showing an example of a classification hierarchy based on a tree structure of a document folder object according to the embodiment of the present invention.

FIG. 5 is a single DTD according to an embodiment of the present invention.
It is a figure which shows an example of the unique allocation method of element name ID in the range of.

FIG. 6 is a diagram showing an example of a unique allocation method of attribute name IDs within a single DTD range according to an embodiment of the present invention.

FIG. 7 is a diagram showing an example of a DTD according to an embodiment of the present invention.

8 is a diagram showing an element-attribute ID table holding an element name ID and an attribute name ID assigned to the DTD of FIG. 7. FIG.

FIG. 9 is a flowchart showing a procedure for creating an element tree index according to the embodiment of the present invention.

FIG. 10 is a diagram showing an example of a document according to the embodiment of the present invention.

11 is a diagram showing a tree structure notation for elements of the document of FIG.

12 is a diagram showing an element tree index for the document of FIG.

FIG. 13 is a flowchart showing a procedure for extracting a parent element according to the embodiment of the present invention.

FIG. 14 is a flowchart showing a procedure for extracting an ancestor element in the embodiment of the present invention.

FIG. 15 is a flowchart showing a procedure for taking out child elements according to the embodiment of the present invention.

FIG. 16 is a flowchart showing a procedure for extracting a descendant element according to the embodiment of the present invention.

FIG. 17 is a flowchart showing a procedure for taking out a brother element in the embodiment of the present invention.

FIG. 18 is a flowchart showing a procedure for taking out the younger brother element in the embodiment of the present invention.

FIG. 19 is a flowchart showing a procedure for taking out the youngest brother element in the embodiment of the present invention.

FIG. 20 is a flowchart showing a procedure for extracting descendant elements in the right depth priority order according to the embodiment of the present invention.

FIG. 21 is a flowchart showing a procedure for taking out the front element according to the embodiment of the present invention.

FIG. 22 is a flowchart showing a procedure for taking out a next element according to the embodiment of the present invention.

FIG. 23 is a flowchart (part 1) showing a procedure for creating an element-attribute index in the embodiment of the present invention.

FIG. 24 is a flowchart (part 2) showing the procedure for creating an element-attribute index according to the embodiment of the present invention.

FIG. 25 is a flowchart showing a procedure for obtaining an entry insertion position in the element-attribute index according to the embodiment of the present invention.

FIG. 26 is a flow chart showing a procedure for extracting target data from data on a list and a procedure for calculating a data insertion position according to an embodiment of the present invention.

FIG. 27 is a flowchart showing an entry search procedure in the element-attribute index according to the embodiment of the present invention.

FIG. 28 is a diagram showing an instance showing an example of a relationship between each object of a document folder, a document, and a DTD according to an embodiment of the present invention.

FIG. 29 is a document object according to an embodiment of the present invention.
It is a figure which shows A, B, and C.

FIG. 30 is a diagram showing an element-attribute ID table of a DTD object X according to an embodiment of the present invention.

FIG. 31 is a diagram showing an element-attribute ID table of a DTD object Y according to an embodiment of the present invention.

FIG. 32 is a diagram showing an element-attribute index according to an embodiment of the present invention.

FIG. 33 is a diagram showing a search result of an element “patient” having an attribute “pID” value of 1000 or more in one embodiment of the present invention.

FIG. 34 is a document object according to an embodiment of the present invention.
It is a figure which shows the element tree index of A.

FIG. 35 is a document object according to an embodiment of the present invention.
It is a figure which shows the element tree index of C.

FIG. 36 is a diagram showing a search result of an element “patient” which is a descendant element of the element “disease record” in one example of the present invention.

[Explanation of symbols]

1 database 11 documents 12 document folders 13 DTD 14 element tree index 15 element-attribute index 16 elements-attribute ID table 21 Document Search Execution Unit 211 Element tree index creation part 212 element-attribute index 213 element-attribute ID table creation unit 22 Search execution unit 111, 112, 113 documents 131, 132 DTD 121-126 Document Folder

─────────────────────────────────────────────────── ─── Continuation of front page (51) Int.Cl. ⁷ Identification code FI G06F 12/00 546 G06F 12/00 546T (56) References Takuya Kitano, Misa Namiuchi, XML document based on semi-structured data model Storage and retrieval and its implementation method, IPSJ Research Report (99-DBS-117), January 23, 1999, Vol. 99, No. 6, p. 31- 38 Takuya Kitano, Misa Namiuchi, Prototype of XML document management system using semi-structured data model, Proc. Of the 57th National Conference of Information Processing Society of Japan (Late 1998) (3), 1998 10 5th p.m. 283-284 (58) Fields investigated (Int.Cl. ⁷ , DB name) G06F 17/28-17/30 G06F 12/00

Claims (11)

(57) [Claims] 1. A database for storing a structured document, and a structured document that has been input is analyzed and stored in the database.
Document storing and executing means for storing, and means for searching documents from the database and outputting search results
And a search means for outputting to the database, as a class in the schema, the document class that manages a document, the document of DTD; and DTD class that manages the (Document Type Definition document type definition), and the document A document folder class for managing a set of DTDs, and a document tree class representing a tree structure relationship of document elements in the document class as a first index, and the document as a second index. Folder class manages index for each element of document and attribute
Element for - with the attribute index, as a third index, the said DTD class, elements for managing the element and attribute names as ID - attribute I
D table, which is a numerical element name ID corresponding to the element name and the attribute name of the element in the document,
An element-attribute ID table having an attribute name ID is provided, and the document storage executing means is provided for each element-attribute ID table of the database.
A means for creating the element tree index for each document using a prime ID, and an element name defined in the DTD of the input structured document
And element name ID and attribute name ID respectively
And means for creating the element-attribute ID table, and storing the element and attribute information for the input structured document.
A means for creating the element-attribute index forming an index to be held is provided, and the input structured document is stored in the element tree index.
And the element-attribute ID table, and the element-attribute
Using an index, in the form based on the logical structure,
Said retrieval means and stored in the database and retrieves the document from the database,
When performing a search for the input search request based on the logical structure of the document, the target structuring is performed using the element tree index, the element-attribute ID table, and the element-attribute index .
A structured document management device characterized by searching a document and outputting the search result to the output means . 2. The retrieval unit extracts a document element as a hierarchical relationship from a certain document element in a structured document, and shows a parent element, an ancestor element, a child element, a descendant element, and an older element for the document element. brother element before the element, taken out in either or a combination of the following elements, a structured document management apparatus according to claim 1, wherein a. Wherein said retrieval means, documents and Upon retrieval of the element, DTD performs a search for a plurality of structured documents logical structure are different from each other, such as different structured document according to claim 1, wherein the Management device. 4. A method for creating the element tree index.
The column is an index representing the relationship between elements in each document , and the structure of the index is such that for each element (element name ID, depth from the root element when the element is represented by a tree structure). The element tree having three sets of item data (element level indicating the
Create an index and store it in the database,
The structured document management device according to claim 1, wherein 5. A method for creating the element-attribute index.
A column is an index that holds information on elements and attributes for a plurality of documents, and includes (DTDID, element-attribute name I
D, attribute value, the document ID, said element consisting of five sets of elements starting position) - Creates an attribute index, the database
The structured document management apparatus according to claim 1 , wherein the structured document management apparatus stores the structured document management apparatus. 6. A procedure for creating the element-attribute ID table.
The stage is for creating an element-attribute ID table having correspondence between element name ID and element name and correspondence between attribute name ID and attribute name.
Or, wherein only become such element name ID for each element name defined in the DTD is given as a number, to create a correspondence between the element name and the element name ID as a table, the table, It is stored in the database as an instance of the element-attribute ID table in a one-to-one relationship with an object that is an instance of the DTD class, and the attribute name is unique to the element in the DTD object and different from each other. Even if the same attribute name exists between element names, an attribute name ID that can identify them is given as a numerical value, and the correspondence relationship between the attribute name and the attribute name ID is created as a table, and the table stores the element- The structured document according to claim 1 , wherein the structured document is stored in the database as an instance of an attribute ID table. Management device. 7. A schema of a database, a document class for managing documents, a DTD class for managing DTD (Document Type Definition) of the documents, and a document folder class for managing a set of the documents and DTDs. And, as a first index, the document class is provided with an element tree index representing a tree structure relationship of elements of the document, and as a second index, the document folder class is provided for a plurality of documents. It is an index that holds information on elements and attributes, (DTDID, element-attribute name ID,
Attribute value, document ID, ing than five sets of elements starting position) is needed
An element-attribute I for managing element names and attribute names as IDs is provided in the DTD class as a third index.
D table, which is a numerical element name ID corresponding to the element name and the attribute name of the element in the document,
Element having an attribute name ID - structured document management apparatus comprises a database having the attribute ID table, when storing the input <br/> documents in the database, said element (a) - Create an attribute ID table In doing so, an element name ID that is unique to each element name defined in the DTD of the input document is given as a numerical value, and a correspondence relationship between the element name and the element name ID is created as a table. , The created table, the DT
It is stored in the database as an instance of an element-attribute ID table in a one-to-one relationship with an object that is an instance of D class, and the attribute names are unique to the elements in the DTD object and different from each other. Numerical values are given to the attribute name IDs so that they can be identified even if the same attribute name exists between them, and the correspondence relationship between the attribute name and the attribute name ID is created as a table,
Storing the created table in the database as an instance of the element-attribute ID table, and (b) in creating the element tree index, the first character position of the start tag of the element of the input document. , The last character position of the end tag, the element name in the start tag as the element start position, the element-attribute ID table is retrieved from the DTD object related to the document object of the document, and the element name of the element Next, the ID is calculated, and then the element level indicating the depth from the root element when the element is represented by a tree structure is calculated.
D, element level, element start position) is added, and the created element tree index is added.
(C) Each item of the data structure (DTDID, element-attribute name ID, attribute value, document ID, element start position) of the quintuplet for storing the element-attribute index Value is calculated, the created entry is added to the list of the element-attribute index while maintaining the order sorted by DTDID and the element-attribute name ID value, and the created element is added.
A step of storing an attribute index in the database, the element tree index for performing a search based on the logical structure of the document from the database in response to an input search request , The element-attribute ID table,
Targeted structuring using the element-attribute index
A structured document management method comprising the steps of searching a document and outputting the search result from an output means . 8. A parent element, an ancestor element, a child element, or a descendant as an element having a relative relationship with a certain element using the element tree index when performing a search from the database based on the logical structure of the document. Element, brother element,
The structured document management method according to claim 7 , wherein any one of a younger brother element, a previous element, a next element, or a combination thereof is taken out. 9. When creating the element-attribute index, when a document object is added to an object that is an instance of a document folder class, a DTD object related to the document object is also added at the same time, and the added document is added. Create and add an entry for the element-attribute index for each element of the object, find the start tag of the element of the document, find the first character position of the start tag and the element name of the start tag, DTD associated with the document object An element having a one-to-one relationship with an object-an element ID corresponding to the element name is obtained from an attribute ID table, an attribute in the start tag is obtained, and if there is no attribute, the element is (DTDID, element name ID)- The insertion position for the attribute index is obtained, and (DTDI
D, element-attribute name ID, null, document ID, element start position) is added, and if there is an attribute in the start tag, the attribute value of the attribute is changed from the element-attribute ID table to the attribute name. The corresponding attribute ID is obtained, the insertion position with respect to the element-attribute index is obtained by (DTDID, attribute name ID), and the (DTDI
8. The structured document management method according to claim 7 , wherein an entry of D, element-attribute name ID, null, document ID, element start position) is added. 10. A DTD is used for searching a document and its elements.
8. The structured document management method according to claim 7 , wherein a search is performed for a plurality of structured documents having different logical structures such as different. 11. A schema of a database, a document class for managing a document, a DTD class for managing a DTD (Document Type Definition) of the document, and a document folder class for managing a set of the document and the DTD. And, as the first index, the document class is provided with an element tree index representing the tree structure relationship of the elements of the document, and as the second index, the document folder class is provided with the element-attribute. The index is an index that holds information on elements and attributes for multiple documents,
(DTDID, element-attribute name ID, attribute value, document ID,
Five sets than ing elements of the element starting position) - attribute index
As a third index, the DTD class has an element-attribute I for managing element names and attribute names as IDs.
D table, which is a numerical element name ID corresponding to the element name and the attribute name of the element in the document,
Element having attribute name ID-In a structured document management device provided with a database having an attribute ID table, in storing an input document in the database, (a) for each element name defined in the DTD A unique element name ID is given as a numerical value, a correspondence relationship between the element name and the element name ID is created as a table, and the created table is in a one-to-one relationship with an object that is an instance of the DTD class. And element-attribute I
Stored in the database as an instance of the D table, the attribute names are unique to the elements in the DTD object, and even if the same attribute name exists between different element names, these can be identified. An element-attribute index in which a numerical value is given to the attribute name ID, a correspondence relationship between the attribute name and the attribute name ID is created as a table, and the created table is stored in the database as an instance of the element-attribute ID table. (B) The first character position of the start tag and the last character position of the end tag of the document element are obtained, the element name in the start tag is obtained as the element start position, and the document object of the document is obtained. The element-attribute ID table is taken out from the related DTD object, the element name ID of the element is obtained, and then continued. Te, the element obtains a element level which represents the depth from the root element when expressed in a tree structure, (element name ID, element level, element start position) add to <br/> entries consisting triples (C) When a document object is added to an object that is an instance of the document folder class, a DTD object related to the document object is also added at the same time, and the element tree index is created. Create and add an element-attribute index entry for each element,
The start tag of the document element is searched for, the first character position of the start tag and the element name of the start tag are obtained, and the element name is obtained from the element-attribute ID table having a one-to-one relationship with the DTD object related to the document object. Element ID corresponding to
If there is no attribute in the start tag, the insertion position for the element-attribute index is calculated by (DTDID, element name ID), and the insertion position is set to (DTDI
D, element-attribute name ID, null, document ID, element start position) is added, and if there is an attribute in the start tag, the attribute value of the attribute is changed from the element-attribute ID table to the attribute name. The corresponding attribute ID is obtained, the insertion position with respect to the element-attribute index is obtained by (DTDID, attribute name ID), and the (DTDI
D, element-attribute name ID, null, document ID, element start position), element-attribute index creation processing, and (d) in response to the input search request When conducting a search based on the logical structure,
Target using the element tree index, the element-attribute ID table, and the element-attribute index
A structured document is searched, and at that time, as elements having a relative relationship from a certain element using the element tree index, a parent element, an ancestor element, a child element, a descendant element, an older brother element, a younger brother element, a previous element, A process for extracting any one of the following elements or a combination thereof and outputting the search result to the output means; and causing the computer constituting the structured document management device to execute the processes (a) to (d) above. A computer-readable recording medium in which a program is recorded.