JP2001282856A - Index generation method, index display system, index retrieval method and index generation device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method capable of generating an index for structured documents, which is smaller in size than conventional one and enables efficient retrieval. SOLUTION: In a tree, composed of plurality of vertices and plurality of edges, which represent a XML document of an object of index generation, names of starting points and names of endpoints of edges, excluding edges which satisfy a predetermined condition, are interchanged arbitrarily, and succeedingly, plurality of edges which share a common starting point and have a same name of starting point and a same name of endpoint are merged into one recursively. Then, root vertices of data structure of indices for plurality of XML documents are merged into one. Also, regarding to this data structure in which root vertices are merged into one, plurality of edges which share common starting point and have a same name of starting point and a same name of endpoint are merged into one recursively. These indices thus generated are stored in increments of edge. On this occasion, every edge having same name of starting point is stored in a continuous region in a storage device.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an index creating method and an index creating apparatus for creating an index, which is an effective data structure, for quickly extracting necessary information from information contained in a structured document. The present invention relates to an index display method for displaying corresponding information in response to a browsing request, and an index search method for searching for corresponding information in response to an index search request.

[0002]

2. Description of the Related Art Methods for extracting necessary information from data include, for example, methods such as "scan", "search", "hashing", and "in decking".

[0005] Scanning is a method of examining data one by one and extracting data that matches required information, and is the most primitive method. The search is a method of arranging data in advance according to a certain rule and searching for data using this rule, and includes a block search, a binary search, and the like. Hashing is a method of using a function called a hash function to find the location of the content to be searched from the content of the content.

On the other hand, in decking prepares a table called an index separately from the data to be searched,
This is a method of searching for data using this table. The conventional index is a table that records a set of information contents to be searched and the location of the contents.
M and B-TREE. An index is rarely simply a set of information content and its location, and is often characterized by specifying the content of information and quickly obtaining the location of the content. (For example, a tree structure with small variations in the depth from the root to each leaf and the size of each vertex is used).

[0005] Relational databases are one area in which index research is progressing. The relational database is a database that stores a table having a vertical axis called a row and a horizontal axis called a column (each row is called a tuple). For example, FIG.
FIG. 7 shows an example of a table. In the table, if an index is created for a specific column, for example, a column called “lead inventor”, it is possible to perform a search for this column at high speed. FIG.
For example, in response to a search request that “the first inventor searches for a tuple that is Toshiki Kizu”, it is possible to quickly find a tuple that satisfies the condition.

When the search target is a relational database, the conventional indiction technology is effective, but the search target is not always always a relational database.

In recent years, the standard document format exchanged via a communication path such as the Internet or an intranet has been changed to HT.
There are structured documents such as ML documents and XML documents (XML
For documents, see, for example, “Extended Mar
kup Language (XML) 1.0 "(W3C
Recommendation 10-Februa
ry-1998)). It is assumed that there is a demand for obtaining necessary information from a structured document such as an HTML document or an XML document. In this case, the conventional indiction technology is not effective.

[0008] The difference between a search in a relational database and a search in a structured document results from the difference between the structure of the table and the structure of the structured document.

FIGS. 28 and 29 show examples of structured documents. The specific example shown in FIG. 28 is described in “Application No. 8-23.
2515 ”is expressed in an XML document format. The specific example shown in FIG. 29 expresses "predetermined information relating to application number Japanese Patent Application No. 8-47526" in an XML document format. These two XMs
The L document corresponds to the two “predetermined information related to the patent application” illustrated in FIG. 27 (however, the included information amounts are slightly different examples).

FIG. 27 is compared with FIG. 28 with respect to “information on first inventor”.
In contrast to the column, which has a flat structure in which the value of each tuple corresponds, in FIG. 28, a “name” element exists as a child element of the “first inventor” element, and a “name” element exists. "Last name" element and "First name" element are present in the child element of the element, "Gitsu" character string is present in the child character string of the "Last name" element, and "Toshiki" character string is present in the child element of the first name element Has a hierarchical structure.

For example, in the case of searching for "information on the first inventor", in FIG. 27, "the value of the column of"
Search for tuples that are "x"]. In other words, it is sufficient to prepare an index for storing a tuple corresponding to the value of the column whose column name is "lead inventor".

On the other hand, an XML document does not have a single search type because it has a hierarchical structure. The search pattern for obtaining a structured document including this character string from the character string “Gizu” includes a search “Search for an XML document whose last name is Kizu” and a search pattern “The last name of the name is Gitsu. XM
Search for "L document", search for "XML document whose first inventor's name is Gizu",
"X in which the last name of the first inventor in the published patent gazette is Kizu
Search for ML document ".

Therefore, in the same way as in the relational database, the value of the hierarchy (corresponding to the column value in the relational database) is determined for each hierarchy (corresponding to the column name in the relational database).
Considering a system that prepares an index that stores XML documents (corresponding to tuples in a relational database) corresponding to and, in order to obtain search results at high speed for all of the preceding search requests, an index of the hierarchy "last name" and a hierarchy "name" , Surname ”index, hierarchy“ first inventor, first name,
An index of "last name" and an index of hierarchy "public patent gazette, first inventor, first name, last name" must be prepared.

In addition to the above search request, if a search "search for an XML document in which the first inventor's name includes the last name" is desired, a high-speed search must be performed first. The prepared four types of indexes cannot be used. In order to realize this search at a high speed, a new index of the hierarchy "first inventor, name" must be prepared.

Furthermore, in this method, an index is required only for the type of the hierarchy to be searched, which is not practical in terms of the storage size of the index.

[0016]

As described above, in the conventional indiction technology, it is difficult to create an index that can effectively respond to a search request when a structured document is targeted. Even with an index that could handle the request to some extent, there were problems with data size and search speed.

The present invention has been made in view of the above circumstances, and has an index creation method and an index creation method for creating an index of a structured document, which enables a more efficient search with a smaller data size than before. It is an object to provide an apparatus, an index display method for displaying corresponding information in response to an index browsing request, and an index search method for searching for corresponding information in response to an index search request.

[0018]

According to the present invention, vertices other than the highest vertex (root) have only one upper vertex as a parent, and vertices other than the lowest vertex have one or more lower vertices as children. , A plurality of vertices that form a parent-child relationship hierarchically, and a plurality of edges that hold information (eg, the name of the start point and the name of the end point) related to a pair of upper and lower vertices in the parent-child relationship 1 represented by the structure consisting of
An index creation method for creating an index from two structured documents, in which information of a plurality of sides having the same vertex on the upper side and having a predetermined relationship is obtained in one target structured document. It is determined whether or not there are a plurality of sides to be held (for example, sides sharing the upper vertex and having the same name of the start point and the end point). A vertex located on the lower side of is integrated as one vertex to perform a sharing process of sharing the plurality of sides into one, and the structure of the structured document after performing the sharing process is retained. An index is created. The above is a case where an index of only a single structured document is created. When creating an index for a plurality of structured documents, the highest vertex of the structure representing the plurality of target structured documents is integrated into one to form a new structure, and the plurality of target In each of the structured documents and the new one structure, it is determined whether or not there are a plurality of sides having the same vertex on the upper side and holding a plurality of pieces of information having a predetermined relationship. If the plurality of sides exist, a vertex located on the lower side of the plurality of sides is integrated as one vertex to perform a sharing process of sharing the plurality of sides into one. Creates an index that holds the structure of the structured document after the conversion process. The above-described sharing process (herein, referred to as a first sharing process) for each of a plurality of target structured documents, and the sharing process (here, a second sharing process) for the new one structure are similarly performed. The order of the process of integrating the top vertices can be set in various ways, for example, the first sharing process, the integration of the top vertices, and the first sharing process. Or the integration of the top vertex, the first sharing process, the second
May be performed in this order, or the first sharing process and the second sharing process may be performed collectively after the top vertices are integrated. Alternatively, when creating an index for a plurality of structured documents, the sharing process is not performed for each of the plurality of target structured documents, and the highest vertex of the plurality of structured documents is set to 1 Into one new structured document, perform the sharing process of the edge belonging to the original structured document of the new structured document, and perform the sharing process. Create an index that holds the structure of the structured document of.

Further, the present invention provides a hierarchical structure in which vertices other than the highest vertex have only one upper vertex as a parent and vertices other than the lowest vertex have one or more lower vertices as children. Index creation for creating an index of a structured document represented by a structure including a plurality of vertices forming a parent-child relationship and a plurality of edges holding information relating to a pair of upper and lower vertices in the parent-child relationship The apparatus further comprises means for creating an index holding a structure in which the plurality of sides having a certain relationship are integrated into one while maintaining the parent-child relationship in the structure of the target structured document. Features.

It should be noted that the present invention relating to the apparatus is also realized as an invention relating to a method, and the present invention relating to a method is also realized as an invention relating to an apparatus.

Further, the present invention relating to an apparatus or a method is provided for causing a computer to execute a procedure corresponding to the present invention (or for causing a computer to function as means corresponding to the present invention, or for causing a computer to correspond to the present invention). The present invention is also realized as a computer-readable recording medium in which a program for realizing the function of performing the above is recorded.

According to the present invention, an index is created which maintains a parent-child relationship in the structure of a target structured document and holds a structure in which a plurality of sides having a certain relationship are combined into one. This makes it possible to quickly search for necessary information from information included in the structured document. In addition, the index requires less storage area for the index than a conventional index having the same search function, which also contributes to an improvement in search speed.
In addition, storing portions having the same structure of the index in a continuous area on the index storage device contributes to further improvement in search speed. Further, according to the present invention, when performing a search, it is possible to interactively perform a search request several times without knowing the structure of the structured document. When the structure of a structured document is known, a higher-speed search can be performed by specifying the structure in one search request. Thus, the same index can support both search forms.

[0023]

Embodiments of the present invention will be described below with reference to the drawings.

First, the basic configuration of the present invention or its embodiment will be described.

The present invention targets a document represented by a set of data called "vertices" and a set of data called "edges".

Here, each vertex holds a name, and each side is composed of two vertices, a vertex called a "start point" and a vertex called an "end point". With one exception,
All vertices shall be the end points of only one side.

For example, in an XML document, a vertex corresponding one-to-one to an element, an attribute name, an attribute value, a character string, and an exceptional vertex prepared separately from them are referred to as a vertex. A set of element vertices-element vertices; a set of element vertices-attribute name vertices; a set of element vertices-character string vertices; a set of attribute name vertices-attribute value vertices; A document that can be expressed by a set of these vertices and edges as a set obtained by combining a set of exception vertices and a root element vertex.

The end point viewed from the start point is called a "child", and the start point viewed from the end point is called a "parent".

Except for one exceptional vertex, all vertices are the end points of only one edge, meaning that all vertices have only one parent, except for one exceptional vertex. I do.

The exceptional vertex is called a "root".

The above-described data structure composed of a set of edge sets of a vertex set is called a rooted tree, and a document that can be expressed by the rooted tree is called a “structured document”.

The present invention relates to creation, storage, browsing, and retrieval of an index, which is a data structure for quickly extracting necessary information from a plurality of structured documents.

The index expresses all structured documents as rooted trees and combines all roots of the rooted tree into one vertex to create a new large rooted tree (including only one structured document). There is also an index).

The greatest feature of the rooted tree of the index is that "the hierarchy of the original structured document is retained".

For example, a hierarchy such as "find a structured document in which the child of A is B, the child of B is C, and the child of C is D" is obtained. It is possible to find a structured document that matches the search request by tracing the hierarchy of the root tree of the index from parent vertices to child vertices or from child vertices to parent vertices. is there. At this time, since the root tree of the index retains the hierarchy of the original structured document, if a side of the index that does not match the search request appears during the search, it is not necessary to perform the search ahead from this side. Absent. In the previous example, there is an edge where the child of A is B, and B
If there is no edge whose child is C, there is no need to proceed with the search before this edge.

However, an index in which the roots of the rooted trees of all the structured documents are combined into one vertex is not sufficient in terms of the search speed and the size of the data structure.

Therefore, starting from this index, a plurality of documents having a fixed relationship within one structured document, or over a plurality of structured documents, or within one structured document and over a plurality of structured documents. Sides, for example,
The same hierarchy in a single structured document is obtained by performing (recursively) an operation of combining a plurality of sides holding the same start point name and the same end point name into one side while sharing the start point. It is possible to combine a plurality of sides holding information and a plurality of sides holding the same hierarchical information in a plurality of structured documents.

The root tree of the index finally obtained: (1) Holds the hierarchical structure held by the structured document. (2) There is a characteristic that there is only one side having a certain relationship (for example, a side sharing a start point and holding the same start point name and the same end point name).

This has the following two effects.

(1) When a search is performed using the index before the transformation, the search corresponding to a plurality of sides is performed. When the search is performed using the index after the transformation, only one side is searched. The search speed is improved because the corresponding search is suppressed.

(2) Since the plurality of sides of the index before the transformation corresponds to only one side in the index after the transformation, the size of the data structure can be reduced.

When the finally obtained rooted tree of the index is stored in the continuous area of the storage device, the index name can be stored in the continuous area of the storage device for each side having the same starting point name. Means that it is possible to start the search from vertices other than the root when performing the search, which brings about the effect of improving the search speed.

In the following, the present embodiment will be described in detail by taking an XML document as an example of a structured document.

FIG. 1 shows a configuration example of a structured document / index processing system according to an embodiment of the present invention.

As shown in FIG. 1, this structured document /
The index processing system includes an XML document input device 1,
Index creation device 2, index storage device 3, index reading device 4, index browsing device 5,
A search input / output device 6 is provided.

The XML document input device 1 is a device for inputting an XML document to be added to an index.

XML input from XML document input device 1
The L document is transmitted to the index creation device 2.

The index creation device 2 creates an index data structure from the XML document transmitted from the XML document input device 1 and stores it in the index storage device 3.

The index reading device 4 receives an instruction from the index browsing device 5 or the search input / output device 6,
The index is read from the index storage device 3,
The data is transmitted to the index browsing device 5 and the search input / output device 6.

The index browsing device 5 receives a user's index browsing request, instructs the index reading device 4 to read an index, and displays the index transmitted from the index reading device 4.

The search input / output device 6 receives the search request from the user, transmits the search request to the index reading device 4, and displays the search result transmitted from the index reading device 4.

In the configuration shown in FIG. 1, there is also a configuration in which the index browsing device 5 and / or the search input / output device 6 are not provided.

In the configuration shown in FIG. 1 or in the configuration shown in FIG. 1 but without the index browsing device 5 and / or the search input / output device 6, all or one of the structured documents that are the basis of the index are Some configurations further include a structured document storage device for storing a unit, and some configurations do not store any structured documents.

This system can be configured as a stand-alone device using one or a plurality of computers, and can also be configured as a server-client system via a network using a plurality of computers. It is.

The above transmission includes a case where data is actually transmitted and a case where a pointer for accessing data is transmitted.

FIG. 2 shows a specific example of the structured document.
The XML document illustrated in FIG. 8 and the XML document illustrated in FIG. 29 will be described as examples. Note that the name of the structured document in FIG. xml, and the name of the structured document in FIG. xml.

(E) is an element, (AN) is an attribute name, (AV) is an attribute value, (S) is a character string, (WE) is an arbitrary element, (WAN) is an arbitrary attribute name, WA
V) means an arbitrary attribute value, and (WS) means an arbitrary character string.

First, a procedure for creating an index from an XML document will be described.

Here, X. xml (FIG. 28) will be described as an example.

First, from the XML document input device 1 to the X.X. xm
Enter l.

X. The xml is transmitted from the XML document input device 1 to the index creation device 2.

The following procedure is performed in the index creation device 2.

First, an XML document is represented by a tree.

FIG. 2 shows an XML document X. An example of a tree expressing xml is shown.

Each branch in FIG. 2 is given a two-stage name such as, for example, Root Patent Publication (E) and Publication Patent Application (E) Application Number (AN).

The former means a side having "root" as a starting point name and "open patent publication (E)" as an end point name, and the latter refers to "open patent publication (E)" as a starting point name and an "application number (AN)". )] ”Means the side with the end point name.

In the following, the side (root, published patent publication (E)) is referred to as the side (published patent publication (E), application number (A)).
N)), and conversely, the side (open patent publication (E), application number (AN)) is a child (supplied) side of the side (root, open patent publication (E)). Call. In other words, if there is a relationship that the end point of one side is the start point of the other side,
The one side is a parent side of the other side, and the other side is a child side of the one side.

In the following, a list of sides having a parent-child relationship is called a path. For example, in the tree of FIG.
(Root, published patent publication (E)) (published patent publication (E), application number (AN)) (application number (AN), Japanese Patent Application No. 7-23)
2515 (AV)) is an example of a path.

Next, in the target tree, the start point name and the end point name of the sides other than the side satisfying the predetermined condition are replaced with "arbitrary (*)" (names are omitted).

Here, as an example, in the tree of FIG. 2, a side whose starting point name is “name of invention (E)”, a side whose starting point name is “first inventor (E)”, and “an inventor” (E)]
, A side having a starting name of "name (E)", a side having a starting name of "name (E)", and a side having a starting name of "surname (E)". The names of the start point and end point of the removed side are replaced with “arbitrary (*)”. However, *
E, AN, AV, or S shall be entered, and "root"
Shall not be replaced.

FIG. 3 illustrates a tree resulting from replacing the tree of FIG. 2 as described above.

Next, edges are shared (recursively) in the target tree.

In the tree of FIG. 3, there are eight sides (arbitrary (E), arbitrary (E)) starting from the end point of the side (root, arbitrary (E)). These eight sides are combined into one to create a new tree.

FIG. 4 exemplifies a tree obtained as a result of grouping edges in the tree of FIG.

In the tree of FIG. 4, there are two sides (arbitrary (E), arbitrary (S)) starting from the end point of the side (arbitrary (E), arbitrary (E)), and the side (arbitrary (E) E), optional (E))
There are four sides (inventor (E), name (E)) starting from the end point of. Combine each side into one (end point is 1
Into a new tree).

FIG. 5 exemplifies a tree obtained as a result of grouping edges in the tree of FIG.

In the tree of FIG. 5, the sides (inventor (E),
There are four sides (name (E), last name (E)) and four sides (name (E), first name (E)) starting from the end point of the name (E). These four sides are combined into one to create a new tree.

FIG. 6 exemplifies a tree obtained as a result of grouping edges in the tree of FIG.

In this example, since there are no more edges to be grouped, the tree shown in FIG. xml (FIG. 28) is a data structure of the index.

The created index is transmitted from the index creation device 2 to the index storage device 3 and stored in the index storage device 3.

The created index is stored in the index storage device 3 in units of sides. At that time, the name of the start point of the side (root, arbitrary (E), arbitrary (AV), name of invention (E), first inventor (E), inventor (E), name (E), last name (E ), Name (E)),
The data may be stored in a continuous area of the index storage device 3.

FIG. 7 shows an example in which the tree of FIG. 6 is stored in the index storage device 3.

In FIG. 7, each of the areas enclosed by squares represents a continuous area on the index storage device 3, and the sides having a parent-child relationship are connected by a dotted line between the end point of the parent side and the start point of the child side. . Further, each side is accompanied by the name of the start point and end point of the side, and the XML to which the side originally belonged.
The name of the document is appended.

Next, a description will be given of the processing procedure of each operation described in the examples up to this point in the case where an index is created from an XML document with no index created.

FIG. 8 shows an example of the processing procedure in this case.

(1) An XML document whose index is to be created is input from the XML document input device 1 (step S1).

(2) The XML document input from the XML document input device 1 is transmitted to the index creation device 2 (step S2).

(3) It is determined whether or not a side whose name should be omitted when an index is created from the XML document in the XML document (step S3). If such a side is included, the process proceeds to (4). If such a side is not included, the process proceeds to (5). Whether to omit the names of the sides of the XML document is determined by the names of the sides (a pair of the start point name and the end point name). If there are a plurality of sides having the same name in the same document, Either keep the name for, or omit the name for all edges. For example, in FIG.
Although there are four sides named (Publication Patent Publication (E) and First Inventor (E)), the index retains some of these sides and omits some of them. Is not allowed (all four side names are omitted in this example). In addition, (First inventor (E),
There are four sides named (name (E)), but some of these sides are retained in the index,
It is not permissible to omit some names (in this example, 4
All the names of the two sides are preserved).

(4) For all sides to be omitted, the start point and end point names are replaced with different names (step S).
4). Specifically, when the start point (end point) name ends with (E) (that is, when the corresponding vertex is an element), it is replaced with “arbitrary (E)”, and the start point (end point) name changes to (A).
N) (that is, when the corresponding vertex is an attribute name), it is replaced with “arbitrary (AN)”, and when the start (end) name ends with (AV) (that is, the corresponding vertex is an attribute value). If there is), replace with "arbitrary (AV)", and if the start (end) name ends with (S) (that is, if the corresponding vertex is a character string), replace with "arbitrary (S)" .

(5) When creating an index from an XML document, it is determined whether to share a side in the XML document (step S5). If the side is to be shared, proceed to (6); otherwise, proceed to (7).

(6) If there are a plurality of sides sharing the same start point and having the same start point name and the same end point name, those sides are put together into one (step S6). As a result of unifying the edges (by unifying the end points into one), a new start point may be shared and a plurality of sides having the same start point name and the same end point name may appear. Put it in a book and put the edges together until there are no more edges to put together.

(7) The data structure of the index created by the index creation device 2 is transferred to the index storage device 3 and stored in the index storage device 3 (step S7).

When storing an index in the index storage device 3, the index is stored in units of sides of the index.

The information held by each side is as follows. (A) Start point name (b) End point name (c) Pointer to child side (if necessary) (d) Pointer to parent side (if necessary) (e) Corresponds to this side (if necessary) However, in order to hold the tree structure of the index, it is necessary to hold both or one of the information (c) and (d).

The index created here contains only one XML document, but one index contains only one XML document. There is a configuration in which information of a plurality of XML documents can be mixed in one index.
When there is always one XML document included in one index, or when XML documents included in one index
If there is one or a plurality of L documents but information of another XML document is not newly added to the index, it is not necessary to hold the name of the XML document corresponding to each side. However, when it is assumed that information of a plurality of XML documents coexists in one index, which XML document each side included in the index corresponds to is stored as information (e). There is a need.

Note that (e) may hold information other than the name of the XML document, such as information indicating a method of accessing the XML document, such as a URL.

As described above, when storing sides, sides having the same start point name may be stored in continuous areas of the storage device.

Next, a description will be given of a procedure for adding information of another structured document to an index of a structured document in a state where the index is stored in the index storage device 3.

Here, X. xml (FIG. 28) is stored in the index storage device 3, and Y. xml (FIG. 29) will be described as an example.

First, from the XML document input device 1 to Y. xm
Enter l.

Y. The xml is transmitted from the XML document input device 1 to the index creation device 2.

The following procedure is performed in the index creation device 2.

First, an XML document is represented by a tree.

FIG. 9 shows an XML document Y. An example of a tree expressing xml is shown.

Next, in the tree representing the target XML document, the names of the start point and the end point of sides other than sides having predetermined characteristics are replaced with “arbitrary (*)” (names are omitted. ).

In the tree of FIG. 9, similar to the previous example,
A side starting from "Name of Invention (E)", a side starting from "First Inventor (E)", a side starting from "Inventor (E)", and a "Name (E)" The name of the start point and the end point of the sides other than the side starting from “, the side starting from“ name (E) ”and the side starting from“ surname (E) ”are“ arbitrary (*) ”. Replace with Also, as in the previous example, * represents any of E, AN, AV, and S.
No replacement is made for "root".

FIG. 10 illustrates a tree resulting from replacing the tree of FIG. 9 as described above.

Next, edges are shared (recursively) in the tree representing the target XML document.

That is, in the tree of FIG. xm
As in the case of creating an index from l, if there is a plurality of sides having the same start point name and the same end point name, the operation of combining those sides into one is performed recursively. Repeat to create a new tree.

FIG. 11 exemplifies a tree obtained as a result of grouping (recursively) edges in the tree of FIG.

The tree shown in FIG. 6 is a data structure of an xml index.

Next, the previously created and stored X.D. xml
The index shown in FIG. 28 is stored in the index storage device 3.
(The data structure read is as shown in FIG. 6), and the "root" vertex of the tree of FIG. 11 and the "root" vertex of FIG. 6 are combined into one to create a new data structure.

FIG. 12 shows the tree obtained as a result.

Further, in the tree of the new data structure that has been put together, sharing of edges is performed (recursively).

In the tree of FIG. 12, when there are a plurality of sides having the same start point and the same start point name and the same end point name, the operation of combining those sides into one is recursively repeated. To create a new tree.

FIG. 13 illustrates a tree obtained as a result of (recursively) grouping the edges in the tree of FIG.

The created index (an index to which information corresponding to a new structured document is added) is
The data is transmitted from the index creation device 2 to the index storage device 3 and stored in the index storage device 3.

As with the previous example, when storing,
The data is stored in the index storage device 3 in the unit of a side. At this time, the starting point name of the side (root, arbitrary (E), arbitrary (AN), title of the invention (E), first inventor (E), inventor ( (E), name (E), last name (E), first name (E)) may be stored in a continuous area of the index storage device 3 for each side having the same name.

FIG. 14 shows an example in which the tree of FIG. 13 is stored in the index storage device 3.

In the same manner as in the previous example, in FIG. 14, the area within the range enclosed by the square represents the continuous area on the index storage device 3, and the side having the parent-child relationship is the end point of the parent side and the side of the child side. The starting point is connected with the dotted line. In addition, on each side,
The name of the XML document to which the side originally belonged is added together with the start point name and end point name of the side.

Next, a description will be given of the processing procedure of each operation described in the examples up to this point in a case where information of an XML document is newly added to an index in a state where an index has already been created.

FIG. 15 shows an example of the processing procedure in this case.

(1) An XML document whose index is to be created is input from the XML document input device 1 (step S101).

(2) The XML document input from the XML document input device 1 is transmitted to the index creation device 2 (step S102).

(3) It is determined whether or not a side whose name should be omitted when an index is created from the XML document in the XML document (step S103). If such a side is included, the process proceeds to (4). If such a side is not included, the process proceeds to (5). Whether to omit the names of the sides of the XML document is determined by the names of the sides (a pair of the start point name and the end point name). If there are a plurality of sides having the same name in the same document, Either keep the name for, or omit the name for all edges. The side held in the index stored in the index storage device 3 is the XM to be newly added to the index.
The L document must be held, and the side omitted in the index stored in the index storage device 3 must be omitted in the XML document to be newly added to the index. For example, in FIG. 9, there are four sides named (Publication Patent Publication (E), Lead Inventor (E)), but these sides are omitted in the index shown in FIG. All sides must be omitted. Also, there are four sides named (First Inventor (E), Name (E)), but since these sides are stored in the index shown in FIG. 7, all of these sides are stored. There must be.

(4) For all sides to be omitted, the start point and end point names are replaced with different names (step S).
104). Specifically, when the start point (end point) name ends with (E) (that is, when the corresponding vertex is an element)
Is replaced with "arbitrary (E)", and if the start point (end point) name ends with (AN) (that is, if the corresponding vertex is an attribute name), it is replaced with "arbitrary (AN)" and the start point ( When the (end point) name ends with (AV) (that is, when the corresponding vertex is an attribute value), it is replaced with “arbitrary (AV)”, and when the start point (end point) name ends with (S) (that is, the corresponding If the vertex is a character string), "arbitrary (S)"
Replace with

(5) When creating an index from an XML document, it is determined whether to share a side in the XML document (step S105). If the side is to be shared, proceed to (6); otherwise, proceed to (7).

(6) If there are a plurality of sides sharing the same start point and having the same start point name and the same end point name, those sides are put together into one (step S106). As a result of grouping edges, a new start point may be shared, and there may be multiple edges with the same start point name and the same end point name. Combine the edges until there are no more.

(7) The index is read from the index storage device 3, and the “root” vertex of the data structure of the read index and the “root” vertex of the tree obtained in the previous step are combined into a single tree. Is created (step S107).

(8) When creating an index from a plurality of XML documents, it is determined whether to share a side among the plurality of XML documents (step S108). If the side is to be shared, the process proceeds to (9). If the side is not shared, the process proceeds to (10).

(9) If there are a plurality of documents that share the same start point, have the same start point name, the same end point name, and belong to different structured documents, those sides are combined into one (step S).
109). As a result of grouping edges, a new start point may be shared, and there may be multiple edges with the same start point name and the same end point name. Combine the edges until there are no more.

(10) The data structure of the index created by the index creation device 2 is transferred to the index storage device 3 and stored in the index storage device 3 (step S110).

When storing an index in the index storage device 3, the index is stored in units of sides of the index.

The information held by each side is as follows. (A) Start point name (b) End point name (c) Pointer to child side (if necessary) (d) Pointer to parent side (if necessary) (e) Name of XML document corresponding to this side As described above, in order to hold the tree structure of the index, it is necessary to hold both or one of the information (c) and (d).

In this case, since information of a plurality of XML documents is mixed in one index, which XML document each side included in the index corresponds to is stored as information of (e). Need to be kept.

As described above, (e) may hold information other than the name of the XML document, such as information indicating a method of accessing the XML document, such as a URL.

Further, as described above, when storing sides, sides having the same start point name may be stored in continuous areas of the storage device.

[0138] Here, a case has been described where information of another XML document is added to an index containing information of one XML document to form an index containing information of two XML documents. When adding information of another XML document to an index including information of the document, the same processing as in adding information of another XML document to an index including information of one XML document may be sequentially performed. . Instead of adding XML document information to the index one by one, an index including information of a plurality of XML documents and a plurality of XML documents are added.
The index including the information of the L document can be processed and combined as described above.

Similarly, it is possible to collectively collect information of three or more single XML documents into one index, or collectively collect three or more indexes including information of a plurality of XML documents. It is.

Next, a variation of the index creation processing of the present embodiment will be described.

First, in the case where information of only one XML document is included in one index (information of a plurality of XML documents is not combined), in the example of the procedure of FIGS. Although the sharing and the sharing of the side are performed, the name may be omitted and the sharing of the side may not be performed. Alternatively, the name may not be omitted and the sharing of the side may be performed. .

Further, one index contains a plurality of XMs.
If the information of the L document is included, in the procedure examples of FIGS. 2 to 6 and FIGS. 9 to 13, the names are omitted, the sides are shared, and a plurality of XMLs are used in each XML document.
Although the sharing of the side is performed between the documents, when sharing the side between a plurality of XML documents, the XML
One or both of abbreviation of a name and sharing of a side in a document may not be performed.

Next, a procedure for browsing the contents of an XML document by using the index stored in the index storage device 3 via the index reading device 4 will be described.

Here, using the index illustrated and illustrated in FIG. 14 created and stored as in the previous example, the XM
The browsing procedure will be described with reference to the case of browsing the contents of an L document as an example.

As a specific example, a case where a search is made while browsing an index for an XML document storing predetermined information related to a patent application whose “name of invention” is “continuous data server device and control command transmitting device” Is assumed.

First, with respect to the index browsing device 5,
Enter the index browsing request.

The input browsing request is transmitted to the index reading device 4.

The index reading device 4 stores the side starting from “root” as the index name.
From the upper area, all sides starting from “root” are read and transmitted to the index browsing device 5.

In the case of a system configuration in which there can be a plurality of indexes, when there are a plurality of “root” vertices,
There are a method of performing the above-described processing for all “root” vertices, and a method of performing the above-described processing for some “root” vertices selected based on a predetermined standard.

The index browsing device 5 displays the received side.

When a plurality of sides are received, a method of displaying all the received sides collectively, and a method of selecting one or a group (for example, a group of sides having the same starting point) selected based on a predetermined reference There is a method of displaying in order for each.

FIG. 16 shows an example of a state in which the sides to be received at this time are displayed. The meaning of the display in FIG. 16 is that the names of the XML documents including the path (root, arbitrary (E)) are “X.xml” and “Y.xml”. In FIG. 16, (root, arbitrary (E))
The part displayed as is selectable.

Here, when (root, arbitrary (E)) is selected,
A request to read the child side of the side (root, arbitrary (E)) is transmitted from the index browsing device 5 to the index reading device 4.

The index reading device 4 reads the child side of the side (root, arbitrary (E)) from the index storage device 3 and transfers it to the index browsing device 5.

The index browsing device 5 displays the received side.

FIG. 17 shows an example of a state where the side to be received at this time is displayed. The meaning of the display in FIG. 17 is (root, arbitrary (E)) (optional (E), arbitrary (AN))
The names of the XML documents including the path "X.xml" and "Y.xml" are the names of the XML documents including the path (root, arbitrary (E)) (arbitrary (E), arbitrary (E)). Are “X.xml” and “Y.xml”. In FIG. 17, portions indicated as (arbitrary (E), arbitrary (AN)), (arbitrary (E), arbitrary (E)) can be selected.

When (arbitrary (E), arbitrary (E)) is further selected, the side (arbitrary (E), arbitrary (E),
An optional (E) child side read request is transmitted.

The index reading device 4 reads the child side (arbitrary (E), arbitrary (E)) from the index storage device 3 and transfers it to the index browsing device 5.

The index browsing device 5 displays the received side.

FIG. 18 shows an example of a state in which the sides to be received at this time are displayed. The meaning of the display in FIG. 18 is (root, arbitrary (E)) (optional (E), arbitrary (E))
The name of the XML document including the path (name of invention (E), image transmission method (S)) is “X.xml”,
The name of the XML document including the path of (root, arbitrary (E)) (optional (E), arbitrary (E)) (name of invention (E), data transfer method (S)) is "Y.xml". , (Root, any (E)) (any (E), any (E)) (any (E),
The names of the XML documents including the path “arbitrary (S)” are “X.xml” and “Y.xml”, and are (root, arbitrary (E)) (arbitrary (E), arbitrary (E)) (first invention) (E), the name of the XML document including the path of the name (E)) is “X.xml” and “Y.xml”,
The names of the XML documents including the paths (root, arbitrary (E)) (arbitrary (E), arbitrary (E)) (inventor (E), name (E)) are “X.xml” and “Y.xml”. That is, it is.

As a result, the name of the XML document storing the predetermined information related to the patent application whose “name of invention” is “continuous data server device and control command transmitting device” is “X.xml”. I understood that.

Next, a description will be given of the processing procedure of each operation shown in the examples up to now when browsing the index.

FIG. 19 shows an example of the processing procedure in this case.

(1) For the index browsing device 5,
Input an index browsing request (Step S20)
1).

(2) An index browsing request is transmitted from the index browsing device 5 to the index reading device 4 (step S202). The index browsing request includes a request for specifying the parent side and a request for not specifying the parent side. Specifying the parent side in the index browsing request means that the child side of this parent side is being browsed, and not specifying the parent side means that the side without the parent side, that is, the "root" vertex It means that you are trying to browse the starting side. Here, the parent side is not specified.

(3) The index reading device 4 receives the index reading request and determines whether or not the parent side has been designated (step S203). If the parent side has not been specified, the process proceeds to (4). If the parent side has been specified, the process proceeds to (5).

(4) The index reading device 4 reads out all sides starting from “root” from the index storage device 3 and sends them to the index browsing device 5 (step S204). As described above, in a case where a configuration is adopted in which the names of the starting points of the sides are stored in a continuous area of the index storage device 3 for each side,
The side starting from “root” is stored in a continuous area of the index storage device 3. Next, proceed to (6).

(5) The index reading device 4 reads all child sides of the side designated as the parent side from the index storage device 3 and sends them to the index browsing device 5 (step S205). Next, proceed to (6).

(6) The index browsing device 5 receives the side transmitted from the index reading device 4 and displays it (step S206).

(7) It is determined whether a new browsing request has been input to the index browsing device 5 (step S20).
7). If an input is made, the process proceeds to (3); otherwise, the process proceeds to (7).

Next, X is searched for using the index stored in the index storage device 3 via the search input / output device 6.
A procedure for searching the contents of the ML document will be described.

Here, using the index illustrated and illustrated in FIG. 14 created and stored as in the previous example, XM
The search procedure will be described by taking the case of searching for the contents of an L document as an example.

As a specific example, “name” of “lead inventor”
It is assumed that an XML document that stores predetermined information related to a patent application whose “surname” is “Gizu” is searched.

First, for the search input / output device 6, (first inventor (E), name (E)) (name (E), surname (E))
A search path of (last name (E), Kizu (S)) is input.
The input search path is transmitted to the index reading device 4.

Here, a case is considered where a search request is input in a path expression. However, a system configuration in which a search request is input by another method is also conceivable. For example, a natural language input system such as "Search for an XML document containing information that the first inventor's first name is Kizu" is an example. In that case, the search request input in the search input / output device 6 is converted into a path expression.

The index reading device 4 determines whether the side having the first inventor (E) as the start point name and the name (E) as the end point is to be held in the index or to be omitted. To determine if they are.
Since the side having the first inventor (E) as a starting point and the name (E) as an end point is to be retained, the first inventor (E)
Storage device 3 that stores the side starting from
From the upper area, the side having the first inventor (E) as the start point name and the name (E) as the end point name is read. The side read in the case of the example of FIG. 14 is only one side as shown by the hatched side in FIG.

Next, of the child sides of the read sides, all sides whose end points are the last name (E) or the arbitrary (E) are read. The side read in the case of the example of FIG. 14 is only one side as shown by the hatched side in FIG.

Finally, of the child sides of the read sides, all sides whose end points are Gizu (S) or arbitrary (S) are read. The side read in the case of the example of FIG.
As shown by the hatched side in FIG. 2, there is only one side.

As a result, the name of the XML document storing the predetermined information related to the patent application in which the “last name” of the “first inventor” is “Kitsu” is “X.xml”. You can know that there is.

Next, as another specific example, a case where the “lead inventor” searches for an XML document storing predetermined information related to the patent application “Kizu” will be described.

Here, between the "first inventor" element and the "Gizu" character string, any name may be used, but only those having two elements are searched.

For the search input / output device 6, (First inventor (E), (WE)) ((WE), (WE)) ((W
E), a search path of Kizu (S)) is input. The input search path is transmitted to the index reading device 4.

As described above, (WE) means an arbitrary element. (WAN) means an arbitrary attribute name, (WAV) means an arbitrary attribute value, and (WS) means an arbitrary character string.

First, the index reading device 4 checks whether or not any of the sides having the first inventor (E) as the starting point name is omitted in the index. Since there is no side to be omitted among the sides starting from the first inventor (E), from the area on the index storage device 3 where the side starting from the first inventor (E) is stored, The side holding the end point name starting from the inventor (E) and ending at (E) is read. The side read in the case of the example of FIG. 14 is only one side as shown by the hatched side in FIG.

Next, all the sides holding the end point name ending in (E) and the sides having the end point name (WE) among the child sides of the read sides are read. The sides read in the case of the example of FIG. 14 are two sides as shown by hatched sides in FIG.

Finally, of the child sides of the read sides, all sides whose end point names are Gizu (S) or (WS) are read. The side read in the case of the example of FIG.
As shown by the hatched side in FIG. 2, there is only one side.

As a result, the “leading inventor” stores in XML the predetermined information related to the patent application “Kizu”.
It can be known that the name of the document is “X.xml”.

Regarding the case of performing a search using an index, the processing procedure of each operation shown in the examples up to this point will be described.

FIG. 24 shows an example of the processing procedure in this case.

(1) A search path is input to the search input / output device 6 (step S301).

(2) A search path is transmitted from the search input / output device 6 to the index reading device 4 (step S3).
02).

(3) The recheck flag is set to false.

Thereafter, a loop process is performed.

(4) Among the sides included in the search path, those whose corresponding sides have not been read from the index storage device 3 are searched in order from the top (step S304). If such a side exists (let this side be E) and this is the first side of the search path, go to (5), and if such a side exists (let this side be E), If it is not the first side, the process proceeds to (6), and if there is no such side, the process proceeds to (8) (steps S341 and S342).

(5) The index reading device 4
The starting point and the ending point are checked (step S305). Here, processing according to the start point and end point of E is performed as follows.

(5-1) When the start point name of E ends with (E) When the start point of E ends with (E), the start point of E is divided into six parts according to the branch related to the end point of E as follows. Are divided into two according to the branch regarding the holding target / omission target. Note that there are various methods for determining whether the object is to be held or omitted, for example, by providing a table in which conditions of the side to be held are registered and referring to the table.

(5-1-1) When the end point name of E ends with (E) It is checked whether the side having the same start point name and end point name as that of E is to be held in the index or is to be omitted. (5-1-1-1) When it is a holding object The sides that hold the same start point name and end point name as E are read from the index storage device 3, and “edge set” is set to these edges. (5-1-1-2) When it is an object to be omitted: Read all (arbitrary (E), arbitrary (E)) sides from the index storage device 3 and set “edge set” to these sides. Set the recheck flag to true.

(5-1-2) When the end point name of E ends with (AN) It is checked whether a side having the same start point name and end point name as that of E is to be held in the index or is to be omitted. (5-1-2-1) When it is a holding object All sides holding the same start point name and end point name as E are read from the index storage device 3, and “edge set” is set to these sides. (5-1-2-2) Case of Omission From index storage device 3 (arbitrary (E), arbitrary (A
N)) Read out all sides and set “edge set” to these sides. Set the recheck flag to true.

(5-1-3) When the end point name of E ends in (S) It is checked whether a side having the same start point name and end point name as E is to be held in the index or to be omitted. (5-1-3-1) When it is a holding object All sides holding the same start point name and end point name as E are read from the index storage device 3, and "edge set" is set to these sides. (5-1-3-2) When it is a target to be omitted: All (arbitrary (E), arbitrary (S)) sides are read from the index storage device 3, and these sides are set in the "edge set". Set the recheck flag to true.

(5-1-4) When the end point name of E is (WE): It is checked whether or not there is any side having the same start point name as that of the index, which is to be omitted in the index. (5-1-4-1) When Existing: All sides holding the same start point name as E and the end point name ending with (E) and (arbitrary (E), arbitrary (E)) sides from the index storage device 3 Read out and set “edge set” to these edges. Set the recheck flag to true. (5-1-4-2) When it does not exist All sides holding the same start point name as E and the end point name ending with (E) are read from the index storage device 3, and "edge set" is set to these sides. .

(5-1-5) When the end point name of E is (WAN) It is checked whether or not any of the sides having the same start point name as E in the index is omitted. (5-1-5-1) When it exists The same starting point name as E from the index storage device 3, (AN)
Then, all the sides holding the end point name and the (arbitrary (E), arbitrary (AN)) side are read, and an "edge set" is set to these sides. Set the recheck flag to true. (5-1-5-2) When it does not exist The same starting point name as E from the index storage device 3, (AN)
All the sides holding the end point names ending with are read out, and “edge set” is set to these sides.

(5-1-6) In the case where the end point name of E is (WS): It is checked whether or not any of the sides having the same start point name as E in the index is omitted. (5-1-6-1) When Existing: All sides holding the same start point name as E and the end point name ending at (S) and (arbitrary (E), arbitrary (S)) sides from the index storage device 3 Read out and set “edge set” to these edges. Set the recheck flag to true. (5-1-6-2) When it does not exist All the sides holding the same start point name as E and the end point name ending with (S) are read from the index storage device 3, and "edge set" is set to these sides. .

(5-2) When the start point name of E ends with (AN) When the start point of E ends with (AN), E
Are divided into two by the branch on the end point of
Divided into two.

(5-2-1) When the end point of E ends with (AV) It is checked whether a side having the same start point name and end point name as that of E is to be held in the index or to be omitted. (5-2-1-1) When the side is a holding target All sides holding the same start point name and end point name as E are read from the index storage device 3, and "edge set" is set to these sides. (5-2-1-2) Case of Omission From index storage device 3 (any (AN), any (A
V)) Read out all sides and set "side set" to these sides. Set the recheck flag to true.

(5-2-2) When the end point name of E is (WAV) It is checked whether or not any of the sides having the same start point name as E in the index is omitted. (5-2-2-1) When it exists: the same start point name as that of E from the index storage device 3, (AV)
Then, all the sides holding the end point name ending in and the (arbitrary (AN), arbitrary (AV)) sides are read out, and an “edge set” is set to these sides. Set the recheck flag to true. (5-2-2-2) When it does not exist The same start point name as that of E from the index storage device 3, (AV)
All the sides holding the end point names ending with are read out, and “edge set” is set to these sides.

(5-3) When the start point name of E ends with (WE) When the start point of E ends with (WE), as follows:
Are divided into six by the branch regarding the end point of
Divided into two.

(5-3-1) When the end point name of E ends with (E) It is checked whether or not any of the sides having the same end point name as that of the index is to be omitted in the index. (5-3-1-1) When it exists: All the sides holding the same end point name as E and the (arbitrary (E), arbitrary (E)) sides are read from the index storage device 3,
An "edge set" is set for these edges. Set the recheck flag to true. (5-3-1-2) When it does not exist All the sides holding the same end point name as E are read from the index storage device 3, and "edge set" is set to these sides.

(5-3-2) When the end point name of E ends with (AN) It is checked whether or not any of the sides having the same end point name as E in the index is to be omitted. (5-3-2-1) When it exists: All the sides holding the same end point name as E and the (arbitrary (E), arbitrary (AN)) side are read from the index storage device 3 and the “edge set” is read out from these. Set to side. Set the recheck flag to true. (5-3-2-2) When it does not exist All the sides holding the same end point name as E are read from the index storage device 3, and the "edge set" is set to these sides.

(5-3-3) When the end point name of E ends in (S) It is checked whether or not there is any side having the same end point name as that of the index in the index. (5-3-3-1) When Existing: All sides holding the same end point name as E and (arbitrary (E), arbitrary (S)) sides are read from the index storage device 3, and
An "edge set" is set for these edges. Set the recheck flag to true. (5-3-3-2) When it does not exist All sides holding the same end point name as E are read from the index storage device 3, and "edge set" is set to these sides.

(5-3-4) In the case where the end point of E is (WE) It is checked whether or not any of the sides having the end point ending in (E) is to be omitted in the index. (5-3-4-1) When it exists The side holding the end point name ending at (E) and the (arbitrary (E), arbitrary (E)) side are all read from the index storage device 3, and the “edge set” is read. Are set on these sides. (5-3-4-2) When it does not exist All the sides holding the end point names ending in (E) are read from the index storage device 3, and the "edge set" is set to these edges.

(5-3-5) When the end point name of E is (WAN) It is checked whether or not any of the sides having the end point ending in (AN) is to be omitted in the index. (5-3-5-1) When it exists The side holding the end point name ending with (AN) and the (arbitrary (E), arbitrary (AN)) side are all read from the index storage device 3, and the “edge set” is read. Are set on these sides. (5-3-5-2) When none exists: All sides holding the end point names ending with (AN) are read from the index storage device 3, and an "edge set" is set to these edges.

(5-3-6) When the end point name of E is (WS): It is checked whether or not there is an omission target in the index among the sides having the end point ending in (S). (5-3-6-1) When it exists The side holding the end point name ending at (S) and the (arbitrary (E), arbitrary (S)) side are all read out from the index storage device 3 and “edge set”. Are set on these sides. (5-3-6-2) When the side does not exist All the sides holding the end point names ending in (S) are read from the index storage device 3, and the "side set" is set to these sides.

(5-4) When the starting point name of E is (WAN) When the starting point of E ends with (WAN), as follows:
E is divided into two by a branch related to the end point, and each is divided into two by a branch related to a holding target / omission target.

(5-4-1) In the case where the end point name of E ends with (AV) It is checked whether or not any of the sides having the same start point name as E in the index is omitted. (5-4-1-1) When it exists: All the sides holding the same end point name as E and the (arbitrary (AN), arbitrary (AV)) sides are read out from the index storage device 3, and the "edge set" is Set to side. Set the recheck flag to true. (5-4-1-2) When it does not exist All sides holding the same end point name as E are read from the index storage device 3, and "edge set" is set to these sides.

(5-4-2) In the case where the end point name of E ends with (WAV) It is checked whether or not any of the sides having the end point name ending with (AV) is omitted in the index. . (5-4-2-1) In the case where there is, all of the side holding the end point ending at (AV) and the (arbitrary (AN), arbitrary (AV)) side are read out from the index storage device 3, and the "edge set" is read. Set these sides. (5-4-2-2) When it does not exist All the sides holding the end point names ending with (AV) are read from the index storage device 3, and the "side set" is set to these sides.

After the process (5), the process proceeds to (7).

(6) The index reading device 4
The end point is checked (step S306). Here, processing according to the end point of E is performed as follows.

(6-1) When the end point name of E ends with (E) Of the child sides included in the “edge set”, the side having the same end point name as E and the end point name are arbitrary (E) Are read out, and an “edge set” is newly set for these edges. If there is an edge whose end point name is arbitrary (E) in the “edge set”, the recheck flag is set to true.

(6-2) When the end point name of E ends with (AN) Of the child sides included in the “edge set”, the side having the same end point name as E and the end point name are arbitrary (AN) Are read out, and an “edge set” is newly set for these edges.
If there is an edge whose end point name is arbitrary (AN) in the “edge set”, the recheck flag is set to true.

(6-3) When the end point name of E ends with (AV) Of the child sides included in the “edge set”, the side having the same end point name as E and the end point name are arbitrary (AV) Are read out, and an “edge set” is newly set for these edges.
If there is an edge whose end point name is arbitrary (AV) in the “edge set”, the recheck flag is set to true.

(6-4) When the end point name of E ends in (S) Among the child sides of the sides included in the “edge set”, the side having the same end point name as E and the end point name are arbitrary (S) Are read out, and an “edge set” is newly set for these edges. If there is an edge whose end point name is arbitrary (S) in the “edge set”, the recheck flag is set to true.

(6-5) When the end point name of E is (WE) Among the child sides of the sides included in the “edge set”, the side having the end point name ending with (E) and the end point name are arbitrary (E ) Are read out, and an “edge set” is newly set for these edges.

(6-6) When the end point name of E is (WAN) Of the child sides included in the “edge set”, the side having the end point name ending with (AN) and the end point name are arbitrary (AN ) Are read out, and an “edge set” is newly set for these edges.

(6-7) When the end point name of E is (WAV) Of the child sides included in the “edge set”, the side having the end point name ending with E (AV) and the end point name are arbitrary ( AV) are read out, and an “edge set” is newly set for these edges.

(6-8) When the end point name of E is (WS) Among the child sides of the sides included in the “edge set”, the side having the end point name ending with (S) and the end point name are arbitrary (S ) Are read out, and an “edge set” is newly set for these edges.

After the process (6), the process proceeds to (7).

(7) It is determined whether the “edge set” is an empty set (step S307). If the “edge set” is an empty set, an XM holding information matching the search path
This means that there is no L document. In this case,
The fact that there is no XML document is transmitted to the search input / output device 6, and the process ends (step S309). If the “edge set” is not an empty set, the process returns to (4).

(8) Check the recheck flag. (8-1) When the recheck flag is true The side set in the “side set” is a set of sides corresponding to the search path. XML documents corresponding to these sides are
It is checked whether each document matches the search path, and the matching one is transmitted to the search input / output device 6 (step S30).
8). (8-2) When the recheck flag is false The side set in the “side set” is a set of sides corresponding to the search path. The XML document corresponding to these sides is transmitted to the search input / output device 6 (step S308).

In this embodiment, the method of sequentially examining the sides included in the search path from the beginning has been described. However, the method of examining the search path in order from the end or the method of examining the search path from the middle. In addition, there are various system configurations such as a method of checking XML documents one document at a time when XML document candidates are narrowed down during the search using the index, and a combination of these.

In the following, a case will be described in which the present system is realized by a client-server system.

FIG. 25 shows an example of the configuration of a client server system. In this example, an index creation device 2, an index storage device 3, an index reading device 4, and an XML document storage device 7 are mounted on the server 10 side, and an XML document input device 1, an index browsing device 5, a search All or a part of the input / output device 6 is mounted. The client 20 can communicate with the server 10 via a network 30 such as the Internet or an intranet. Client 20
Of the desired XM provided by the server 10
It is possible to receive the services of creating and storing an index for L documents, browsing the index, searching the index, and acquiring and browsing the XML document.

FIG. 26 shows a configuration example of a client server system. In this example, an index creation device 2, an index storage device 3, and an index reading device 4 are mounted on the server 11 side, and an XML document input device 1, an index browsing device 5,
All or part of the search input / output device 6 is implemented, and XML
The document storage device 7 is provided by one or more other servers 12. The client 20 can communicate with the server 11 or the server 12 via the network 30. The user of the client 20 can receive services provided by the server 11 for creating and storing an index for a desired XML document, browsing the index, and searching for the index.
An XML document acquisition / browsing service can be received.

In addition to the above, various system configurations are possible.

Note that each of the above functions can be realized as software.

The present embodiment is also directed to a computer which records a program for causing a computer to execute predetermined means (or for causing a computer to function as predetermined means or for causing a computer to realize predetermined functions). It can also be implemented as a readable recording medium.

In the above, an XML document has been described as an example of a structured document. However, the present invention can of course be applied to an HTML document and other structured documents.

The present invention is not limited to the above-described embodiments, but can be implemented with various modifications within the technical scope.

[0238]

According to the present invention, an index is created which holds a structure in which a plurality of sides having a certain relationship are combined into one while maintaining a parent-child relationship in the structure of the target structured document. It is possible to create an index of a structured document that has a smaller data size and enables more efficient search than before.

[Brief description of the drawings]

FIG. 1 is a diagram showing a configuration example of a structured document / index processing system according to an embodiment of the present invention.

FIG. 2 is a diagram showing an example of a tree expressing the XML document of FIG. 28;

FIG. 3 is a diagram showing an example of a tree after the tree of FIG. 3 has been replaced;

FIG. 4 is a diagram illustrating an example of a tree obtained as a result of combining edges in the tree of FIG. 3;

FIG. 5 is a diagram showing an example of a tree obtained as a result of combining edges in the tree of FIG. 4;

FIG. 6 is a diagram showing an example of a tree obtained as a result of combining edges in the tree of FIG. 5;

FIG. 7 is a view for explaining a storage format of the tree in FIG. 6 in the index storage device;

FIG. 8 is a flowchart illustrating an example of a processing procedure when an index is created from an XML document;

9 is a diagram showing an example of a tree expressing the XML document of FIG. 29;

FIG. 10 is a diagram illustrating an example of a tree after the tree of FIG. 9 has been replaced;

FIG. 11 is a diagram showing an example of a tree obtained as a result of recursively collecting edges in the tree of FIG. 10;

12 is a diagram illustrating an example of a tree obtained as a result of combining the tree of FIG. 11 with the tree of FIG. 6;

FIG. 13 is a diagram illustrating an example of a tree obtained as a result of recursively combining edges in the tree of FIG. 10;

FIG. 14 is a view for explaining a storage format of the tree in FIG. 13 in the index storage device;

[FIG. 15] A new XML is added to an already created index.
Flowchart showing an example of a processing procedure when adding document information

FIG. 16 is a diagram showing a display example of information about a side obtained by an index browsing request for a root vertex.

17 is a diagram showing a display example of information on a side obtained by an index browsing request for a child side of the side in FIG. 16;

18 is a diagram showing a display example of information on a side obtained by an index browsing request for a child side of the side in FIG. 17;

FIG. 19 is a flowchart illustrating an example of a processing procedure when browsing an index;

FIG. 20 is a diagram for describing a search using an index.

FIG. 21 is a diagram illustrating a search using an index.

FIG. 22 is a diagram illustrating a search using an index.

FIG. 23 is a diagram illustrating a search using an index.

FIG. 24 is a flowchart illustrating an example of a processing procedure when a search is performed using an index.

FIG. 25 is a diagram for describing a system configuration;

FIG. 26 is a diagram for describing a system configuration;

FIG. 27 illustrates an example of a relational database.

FIG. 28 shows an example of a structured document.

FIG. 29 shows another example of a structured document.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... XML document input device 2 ... Index creation device 3 ... Index storage device 4 ... Index reading device 5 ... Index browsing device 6 ... Search input / output device 7 ... XML document storage device 10-12 ... Server 20 ... Client 30 ... Network

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Toshiki Gitsu 1st address, Komukai Toshiba-cho, Saiwai-ku, Kawasaki-shi, Kanagawa Inside the Toshiba R & D Center (72) Inventor Seiji Maeda Komukai, Sai-ku, Kawasaki-shi, Kanagawa No. 1 Toshiba-cho, Toshiba R & D Center (72) Inventor Hirokuni Yano No. 1, Komukai Toshiba-cho, Koyuki-ku, Kawasaki-shi, Kanagawa Prefecture Toshiba R & D Center (72) Inventor Hiroshi Yao, Kawasaki, Kanagawa 1F, Komukai Toshiba-cho, Sachi-ku F-term in Toshiba R & D Center (reference) 5B075 ND02 ND34 NK10 NK21 NR02 5B082 AA11 EA01

Claims (10)

[Claims] 1. A vertex other than the highest vertex has a single upper vertex as a parent, and a vertex other than the lowest vertex has one or a plurality of lower vertices as children to form a hierarchical parent-child relationship. An index creation method for creating an index from a structured document represented by a structure including a plurality of vertices to be processed and a plurality of edges holding information related to a pair of upper and lower vertices in a parent-child relationship. In the structured document to be processed, it is determined whether there are a plurality of sides having the same vertex on the upper side and holding a plurality of pieces of information having a predetermined relationship, and Exists, a vertex located on the lower side of the plurality of sides is integrated as one vertex to perform a sharing process of sharing the plurality of sides into one, and after performing the sharing process, Structured document structure Indexing method characterized by creating a lifting index. 2. A vertex other than the highest vertex has a single upper vertex as a parent, and a vertex other than the lowest vertex has one or more lower vertices as children to form a hierarchical parent-child relationship. An index creation method for creating an index from a plurality of structured documents represented by a structure including a plurality of vertices to be processed and a plurality of edges holding information related to a pair of upper and lower vertices in a parent-child relationship. The top vertices of the structures representing the plurality of target structured documents are integrated into one new structure, and each of the plurality of target structured documents and the new one structure It is determined whether there are a plurality of sides having the same vertex on the upper side and a plurality of sides holding information having a predetermined relationship, and if the plurality of sides exist, the About multiple sides A vertex located on the lower side of is integrated as one vertex to perform a sharing process of sharing the plurality of sides into one, and an index holding the structure of the structured document after performing the sharing process An index creation method characterized by creating an index. 3. A vertex other than the highest vertex has a single upper vertex as a parent, and a vertex other than the lowest vertex has one or a plurality of lower vertices as children to form a parent-child relationship hierarchically. An index creation method for creating an index from a plurality of structured documents represented by a structure including a plurality of vertices to be processed and a plurality of edges holding information related to a pair of upper and lower vertices in a parent-child relationship. Then, the top vertices of the structures representing a plurality of target structured documents are integrated into one new structure, and the new one structure has the same vertex on the upper side. It is determined whether there are a plurality of sides corresponding to the structured document and a plurality of sides holding information in a predetermined relationship, and if the plurality of sides exist, the plurality of sides are determined. Located below Vertices are integrated as one vertex to perform the sharing process of sharing the plurality of sides into one, and to create an index holding the structure of the structured document after performing the sharing process. Index creation method to be characterized. 4. The edge according to claim 2, wherein the side also holds information for specifying a structured document to which the side belongs.
Index creation method described in. 5. The index creation method according to claim 1, wherein the sharing process is repeatedly performed recursively until the edge cannot be shared. 6. Prior to the sharing process performed for the first time, whether information other than information satisfying a predetermined condition exists among information relating to the upper and lower vertices respectively held by the plurality of sides. 6. The index creation method according to claim 1, wherein it is determined whether or not the information is present, and if the information is present, a replacement process for replacing the information with a predetermined abbreviation is performed. 7. When storing the created index in a storage device, a part corresponding to a plurality of sides holding the same information as information related to the upper-side vertex is stored in the index. 7. The storage device according to claim 1, wherein the data is stored in a continuous area of the storage device.
Index creation method described in section. 8. When a browsing request is received for an index created by the index creation method according to any one of claims 1 to 7, a plurality of apexes having the same vertex as the upper vertex are received. An index display method characterized by displaying the index for each part corresponding to a side. 9. A portion corresponding to the side in the index when a search request specifying a structure is received for an index created by the index creation method according to any one of claims 1 to 7. An index search method comprising: searching for a part holding a corresponding structure by referring to information held in the index. 10. A vertex other than the highest vertex has a single upper vertex as a parent, and a vertex other than the lowest vertex has one or a plurality of lower vertices as children to form a hierarchical parent-child relationship. An index creation device for creating an index of a structured document represented by a structure including a plurality of vertices to be processed and a plurality of sides holding information related to a pair of upper and lower vertices in a parent-child relationship. An index comprising: means for creating an index holding a structure in which a plurality of the sides having a certain relationship are integrated into one while maintaining the parent-child relationship in the structure of the target structured document. Creation device.