JP3287307B2 - Structured document search system, structured document search method, and recording medium storing structured document search program - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a structured document search technique, and more particularly to a structured document search technique capable of processing a compound query using a document structure of a structured document and a character string included in the document as search conditions. The present invention relates to a system and a structured document search method.

[0002]

2. Description of the Related Art Document elements constituting a structured document are decomposed and stored in a database, and the texts are associated with the document elements together (hereinafter referred to as "document structure information"), so that the document structure is included in the document structure. A search system that specifies a character string to be searched as a search condition has been realized.

As an example of this type of conventional structured document retrieval system, "Procedures of First International Conference on the Application of Database Technologies and There" was published in 1994.・ Integration (Proceedings of 1st International Conferenceon
the Application of Database Technologies & their
Integration (ADTI'94)) on pages 272 to 283 of "Database Systems for Structured Documents".
Reference is made to a paper by R. Sacks-Davis et al. entitled "Tructured Documents".

[0004] A relational database or an object-oriented database is used as a database for storing document elements. In particular, when an object-oriented database is used, the relation between document elements is directly expressed as the relation between data in the database. It is possible to cruise the document structure according to the structure from the beginning of the document to find a document element that includes the character string specified in the search condition, or to search for a specific structure, for example, "5 or more document elements <author> Including document element <paper> ”can be detected. This example is described in the above reference 1.

Conventionally, two methods have been known as a search method for a database in which document elements are decomposed and managed.

One of the methods is to perform a condition search by starting with a document element that appears first in an original document, sequentially visiting child elements of the document element in an inclusive relationship, and executing a condition determination. . Here, this method is referred to as a “top-down method”.

As another method, a set in which document elements are classified for each document element name (tag name) is created in a database in advance, and a search for a character string that meets a condition is searched from the classification set. There is a way to do that. This method is called "bottom-up method".

There is also known a dedicated document search system that realizes high-speed document search by using a general full-text search (full-text search) engine without using a database.

[0009] As an example of combining an object-oriented database and a document retrieval system, "Proceedings of the Twentys VLD Conference (Proceedi
ngs of the 20th VLDB Conference), pages 740 to 749, "Integrating a Structured Text Retrieval System.
With an object-oriented database system (Integrating a Structured-Text Re
trieval System with an Object-Oriented Database Sy
stem) ”by TW Yan et al. (referred to as“ Reference 2 ”).

[0010]

In the search by the top-down method, since the inclusion relation (parent-child relationship) between document elements appears in the data structure in the database, the search condition includes the condition relating to the document structure. In this case, it is easy to obtain and process information on the document structure. However, this top-down method has a problem that search efficiency is poor.

The reason is that the condition determination of the document element is performed in the order of the structure from the first document element of the document. Therefore, it is not possible to narrow down the set to be searched by the document element name or the like, and there is a possibility that an unnecessary document element is visited. Because it is high.

On the other hand, in the search by the bottom-up method, since a set is generated in advance for each document element name, narrowing down by the document element name included in the search condition is performed in advance. Since different document elements are not subject to the condition determination, there is a problem that the search efficiency is good, but the condition determination based on the document structure is difficult.

The reason is that the document elements are stored as a classification set independently of the document structure, and information on the document structure is missing.

Furthermore, in the case of a search method using a full-text search engine, high-speed search is possible for an arbitrary character string. However, similar to the bottom-up method, structural information is lost. It is not possible to answer a query including a condition relating to the document structure by itself.

Further, in a system in which a database storing information of a document structure is combined with a full-text search engine, a database access for searching for structure information becomes a bottleneck, which causes performance degradation.

SUMMARY OF THE INVENTION Accordingly, the present invention has been made in view of the above problems, and has as its object to provide a structured document search system and method which can answer a query using the document structure of a structured document as a search condition. Is to do.

Another object of the present invention is to provide a structured document search system capable of quickly answering a query in which an arbitrary character string is included in a document as a search condition (for example, a matching condition of a character string included in the document). And a method.

Still another object of the present invention is to provide a structured document capable of quickly answering a compound query including a search condition relating to a document structure of a structured document and a matching condition (search condition) of a character string included in the document. A search system and method are provided.

[0019]

A structured document retrieval system according to the present invention that achieves the above object receives a structured document input from an input unit and stores a full text of the structured document in a text storage unit. Document registration means,
The input text of the structured document is decomposed into document elements, and a unique ID (identification number) is assigned, and the text element corresponding to each of the document elements and the relationship between the document elements are stored in a document part storage unit. A full-text index is generated in units of the input structured document and the document elements constituting the structured document, and the text for the document element includes A full-text index generating means for adding a document element ID, a parent document element ID including the document element, and an ID of an original document as text, and generating a full-text index for the text in that state; Full-text search execution means for performing a search on text and document structure using the full-text index generated by the full-text index generation means. Search means, query analysis means for receiving a search request for a structured document input from the input means, and analyzing query conditions, and based on the structure information of the document stored in the document part storage section, only conditions relating to the structure And a query execution unit that executes a search for a document or a document element that matches a search condition using the full-text search unit or the structure condition determination unit according to the analysis of the query analysis unit.
From the document element ID obtained as a result of the search by the full-text search unit or the structure condition determining unit, information on the document element corresponding to the corresponding document element ID is extracted from the document parts storage unit. A document element retrieval unit for controlling the text storage unit to retrieve the text of the original structured document from the text storage unit and display and output the text to the output unit, and when a compound query including a plurality of conditions is given, The query execution means executes a search for each of the plurality of conditions, one at a time,
In the third and subsequent condition searches, a new query condition is created in which the search result of the previously executed condition is embedded as a condition to narrow down the set of search targets, and the ID of a certain document element and its parent document element or When a search condition related to the document element such as a child document element is given as an input, a document element condition is determined by searching a document structure in the document part storage unit and determining whether the document element satisfies the condition. It further comprises a judgment means.

[0020]

[0021]

[0022]

[0023]

Embodiments of the present invention will be described below. First, the essential parts of a preferred embodiment of the present invention will be outlined.

In the first preferred embodiment of the structured document retrieval system of the present invention, the text of each document element includes a document element ID, a parent document element ID including the document element, and an original document element ID. A full-text index generating means (132 in FIG. 1) for generating a full-text index in this form, and a document element decomposing / storing means (FIG. 1) for storing document structure information. Of 112)
When the query for the structured document includes the matching condition of the character string and the condition related to the document structure, the search is performed by the full-text search execution means (131 in FIG. 1).
If only the condition related to the document structure is included in the inquiry, the structure condition determination means (116 in FIG. 1) searches using only the information in the document part storage unit (122 in FIG. 1).

In the structured document search system according to the second embodiment of the present invention, a certain document element is used as a starting point to determine the truth of a plurality of query conditions for all related document elements. A document element condition judging means (218 in FIG. 10) is provided, and when the number of condition search results of a part of the compound query is sufficiently small, a document element ID obtained as a result of the search and a document part storage unit (FIG. 10) Of 12
Using the information of 2), the operation is performed to determine all the other unprocessed conditions at once for all the document elements related to the document element obtained as a result of the search.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

[First Embodiment] FIG. 1 is a diagram showing a configuration of a first embodiment of the present invention. Referring to FIG.
The first embodiment of the present invention includes a data processing device 100 that operates under program control, and an input device 1 such as a keyboard.
50, a storage device 120, and an output device 140.

The storage device 120 includes a text storage section 121, a document parts storage section 122, and a full-text index storage section 123.

The data processing device 100 includes a structured document management means 110 and a full text search means 130.

The structured document managing means 110 includes a structured document registering means 111, a document element decomposing / storing means 112, an inquiry processing means 113, and a document (element) extracting means 117.

The query processing means 113 includes a query analysis means 114
And a query execution unit 115 and a structure condition determination unit 116.

The full-text search means 130 is a full-text search execution means 131
And a full-text index generating means 132.

These means operate as follows.

The structured document management means 110 stores and manages the structured document and its structure information in the storage device 120, processes an inquiry (search request) for the stored structured document, and matches the condition. The document is returned to the output device 140.

The structured document registration means 111 is provided with the input device 150
, And stores the entire text in the storage device 121. Then, in order to extract and manage the structure information, the entire text is passed to the document element decomposition / storage means 112.

The document element decomposing / storing means 112 decomposes the text of the structured document into document elements, and associates the text corresponding to each document element with the relation between the document elements.
It is stored in the document part storage unit 122 of the storage device 120. here,
A unique ID (identification information) is assigned to each piece of document element information to be stored.

The query processing means 113 receives a search request for the structured document input from the input device 150, and the query analysis means 114 analyzes the query condition. Then, the query execution unit 115 uses the full-text search unit 130 and the structure condition determination unit 116 to search for a document or document element that matches the search condition.

When a compound query composed of a plurality of conditions is given, the query execution means 115 executes a search for each of the individual conditions. A new query condition that narrows down the search target set based on the search result executed in (1) is generated.

The structure condition judging means 116 judges only the condition relating to the structure based on the structure information of the document stored in the document part storage section 122.

The document (element) fetching unit 117 is a document returned from the document part storage unit 122 or the full-text search executing unit 131.
The corresponding text is extracted from the ID or the document element ID and sent to the output device 140.

The full-text search means 130 generates a full-text search index, and performs a high-speed search for documents containing an arbitrary character string by full-text search.

The full-text search execution means 131 uses the full-text index stored in the full-text index storage 123 to
Perform a search on any string, and any combination thereof.

The full-text index generating means 132 generates a full-text index in units of a structured document and its constituent document elements. Here, the ID of the document element, the ID of the parent document element containing the document element, and the ID of the original document are added to the text for the document element as text, and the full-text index is added to the text in that state. Is generated.

The storage device 120 stores the text body and structure information of the structured document, and the full-text index used by the full-text search means 130.

The text storage section 121 stores the text body of the structured document.

The document part storage unit 122 stores the structure information of the structured document.

The full-text index storage unit 123 stores a full-text index used by the full-text search means 130.

The input device 150 sends the input structured document or an inquiry for the structured document to the structured document management unit 110.
Pass to.

The output device 140 displays the structured document or document element on a screen or the like in an appropriate format.

Next, FIGS. 2, 3 and 4 are flow charts showing the processing flow of the first embodiment of the present invention. The entire operation of the first embodiment of the present invention will be described in detail with reference to FIGS.

Prior to the search for a structured document, a full-text index is generated according to the procedure shown in FIG.

The entire structured document input from the input device 150 to the structured document management unit 110 is stored in the text storage unit 121 by the structured document registration unit 111 (FIG. 2).
Step A1).

Next, the document element is extracted by the document element decomposing / storing means 112, and a unique ID is given (step A2 in FIG. 2).

The document element decomposing / storing means 112 generates document structure information in the document part storage unit 122 from the information obtained in the process of extracting the document elements (step A3 in FIG. 2). The full-text index is generated by the full-text index generation unit 132 using the document element of (1) as one unit (step A4 in FIG. 2).

At this time, in addition to the text corresponding to the document element, the ID of the document element, the ID of the parent document element containing the document element, and the ID of the original document are added as text, and the text in that state is added. A full-text index is generated for.

Next, a search process according to the first embodiment of the present invention will be described with reference to the flowcharts of FIGS.

First, the query analysis unit 113 analyzes a search request for a structured document input from the input device 150 (step B1 in FIG. 3), and determines whether the search request is a compound query including a plurality of search conditions. (Step B2 in FIG. 3).

If the result of the determination in step B2 of FIG. 3 indicates that the query is a compound query, the query execution means 115 determines that the inclusion relation of the document element is more outward, that is, the occurrence location in the structured document is closer to the beginning of the document. One search condition for a document element close to is retrieved (step B3). This is for acquiring the parent document element ID to be specified when the subsequent search condition is evaluated. For example, taking the tree structure of a document element as an example, this corresponds to narrowing down the target set from a portion closer to the root on the tree structure toward a leaf portion.

On the other hand, if it is not a compound query, step B3 is not performed.

Next, the query execution means 115 determines whether or not the search condition to be processed includes a character string (keyword) included in the document element as a condition (step B4 in FIG. 3).

If the keyword condition is included in the search condition, it is determined whether or not there is a result obtained by executing the process first (step B5 in FIG. 3).

In the case of evaluating the first condition in the compound query, there is no processing executed first, so this judgment is false. When the second and subsequent conditions are evaluated, there is a result of the previously executed processing.
, A condition that the document element ID of the previously executed search result is used as the parent document element ID is added to the search condition to be processed (step B6 in FIG. 3).

Since the parent document element ID is added to the full-text search text in step A4 of FIG. 2 as a text, in step B6 of FIG. 3, a kind of narrowing-down process using this is realized. Equivalent to.

Then, for the newly generated search condition, the full-text search execution means 131 causes the full-text index storage unit 123
Performs a full-text search for the keyword based on the full-text index stored in
The ID is taken out (step B7 in FIG. 3).

If the keyword is not included as a condition, the processing of steps B4 to B7 in FIG.
The structure condition determination means 116 refers to the structure information in the document part storage unit 122, searches for a document element that matches the search condition, and extracts the ID of the document element that matches the condition (step B in FIG. 3).
8).

When the search process or the condition determination process is completed, the query execution means 115 determines whether or not there is any unprocessed search condition in the query (step B9 in FIG. 4). If there is any unprocessed one, the process returns to step B3 in FIG. 3 and the same processing as described above is repeatedly executed for the unprocessed condition.

If it is determined in step B9 in FIG. 4 that all the search conditions have been processed, the document (element) extracting means 117 obtains a text corresponding to the ID from the obtained document element. Or, a structured document body including the document element is acquired (step B10 in FIG. 4).

Finally, the document (element) extracting means 11
7 sends the acquired text to the output device 140,
140 presents this to the user in an appropriate format (step B11 in FIG. 4).

Next, the operation and effect of the first embodiment of the present invention will be described.

In the first embodiment of the present invention, the target of the full-text index is not only the text but also the ID of the document element and the document ID of the parent element containing the document element.
And the document ID of the structured document containing the document element are added in text format, and "text" + "document element ID" +
Since it is configured to generate a full-text index in the form of "structured document ID", for a compound query whose parent document element is specified in the condition related to the document structure,
Even without referring to the structural information stored and managed separately,
By performing a full-text search, a condition can be determined.

Further, in the first embodiment of the present invention, when the search condition does not include the document matching condition and is composed only of the condition for the document structure, the full-text search is not used and the holding is performed. Since it is configured to make a condition judgment from the structured information that has been performed, the condition judgment can be executed independently of the full-text search function.

[Embodiment 1] Next, in order to describe the first embodiment of the present invention in further detail, a description will be given of an embodiment to which a specific example is applied. Note that the configuration and processing flow of an example of the present invention are the same as those of the above-described first embodiment of the present invention.

In this embodiment, an SMGL (Standard Generalized Markup L), which is one of the structured documents, in which the standards of ISO8879 and JIS X 4151 are established, will be described below.
anguage) is used as an example.

As shown in FIG. 5, "recipe", "titl"
An SGML document as shown in FIG. 6 is input from the input device 150 by a DTD (Document Type Definition) composed of five document elements of “e”, “step”, “item”, and “exp”. Then, the entire text in FIG. 6 is stored in the text storage unit 121 by the structured document registration unit 111 (step A1 in FIG. 2).

Further, the document elements are decomposed and extracted in document element units by the document element decomposing / storing means 112, and a unique ID is assigned to each document element (step A2 in FIG. 2). The information indicating the association is stored in the document part storage unit 12.
2 (step A3).

Further, a full-text index generating means
With 132, a full-text index is generated in the full-text index storage unit 123 for each document element.

At this time, the text of the document element includes FIG.
As shown in (1), in addition to the ID of the document element itself, the ID of the parent document element containing the document element and the ID of the original document are inserted as character strings. The full-text index generation unit 132 is configured as shown in FIG.
As shown in (1), a character string of “text” + “own ID” + “parent ID” + “document ID” is regarded as one “document”, and a full-text index is generated (step A4 in FIG. 4). In the example shown in FIG. 8, the text is “<item> Fry the meat </ item>”
Is pid = 8, parent ID is pid = 6, pid = 1, and document ID (doc_i
In d), did = 100.

Here, a compound query including the character string matching condition and the document structure condition as shown in FIG.
It is assumed that the input is made by the input device 150.

In the query processing means 113, first, the query analysis means 114 converts the query into the following conditions: (1) "a character string of" fried "is included in <step>" (condition 1); Include “butter” in <exp> included ”(condition 2) (steps B1-B2 in FIG. 3).

Then, condition 1 for <step>, which is a document element including <exp> and having an outer inclusion relation outside, is selected as the first processing target (step B3 in FIG. 3). If there is no inclusion relationship between the document elements specified in the condition, any may be selected.

The query execution means 115 determines that the condition 1 includes a character string matching condition (step B4 in FIG. 3), and further determines that there is no search result performed prior to this processing. Then (step B5 in FIG. 3), the inquiry is transferred to the full-text search execution means 131.

The full-text search execution means 131 refers to the full-text index in the full-text index storage unit 123 and executes “<step
A search is made for a document element that includes the character strings “>” and “fried” at the same time (step B7 in FIG. 3).

As a result, the document element shown in FIG. 16 is obtained as satisfying the conditions.

Thereafter, since the unprocessed condition 2 exists, the process starting from step B3 in FIG. 3 is started again.

The condition 2 includes a keyword and the result of the search for the condition 1 exists. Therefore, the query execution unit 115 embeds the search result of the condition 1 in the condition 2 to generate a new search condition ( Step B4-B in FIG.
6).

The new search condition generated at this time is as follows:
"The character strings"<exp>"and" butter "are included at the same time, and" pid = (document element ID of the search result of condition 1) "
Search for document elements that simultaneously include the character string "(condition 2-1)."

The latter half of the new search condition 2-1 (after “and”) is a search condition newly added as a condition for narrowing down the search target set.

When there are a plurality of document elements obtained as a result of Condition 1, for each of them, “pid = (document element
ID) ”is generated, and the process is executed for each condition (step B7 in FIG. 3).

In this case, among the document elements represented by the tree structure in FIG. 7, the document elements corresponding to IDs 6 and 9 satisfy the conditions, so that (1) the character strings “exp” and “butter” Is searched for and the document element that contains the character string "pid = 6" at the same time. "(2)" The character strings "exp" and "butter" are included at the same time and "pid = 9" Search for document elements that include character strings simultaneously "is generated.

Depending on the function of the full-text search means 130, these conditions may be evaluated simultaneously.

As the processing result of the search condition 2-1, ID = 8
The document element with ID = 11 is obtained.

Since this is the processing result of the original query, the document (element) fetching unit 117 stores the information of the corresponding document element from the obtained document element ID in the document part storage unit 122
Then, the entire text of the original structured document is extracted from the text storage unit 121 using the original document ID, and the extracted content is displayed by the output device 140 (steps B10-B11 in FIG. 4).

On the other hand, as shown in FIG. 10, when a query that does not include a keyword but includes only a condition relating to a structure is given (in the example of FIG. 10, <rec> having four or more <step> s).
ipe>), the query execution unit 115 delegates the processing to the structure condition determination unit 116, and the structure information determination unit 116 searches only the document part storage unit 122 and stores the stored <reci
From among the <pe>, search for those having four or more document elements of <step>, retrieve them by the document (element) retrieval means 117, and pass them to the output device 140 from the document (element) retrieval means 117 Output in a simple form.

[Second Embodiment] Next, a second embodiment of the present invention will be described.

FIG. 11 is a diagram showing the configuration of the second embodiment of the present invention. Referring to FIG. 11, a second embodiment of the present invention is shown.
In this embodiment, the inquiry processing means 213 has a document element condition determination means 218 in addition to the configuration of the inquiry processing means 113 in the first embodiment described with reference to FIG.

The document element condition judging means 218 operates as follows.

When given as input the ID of a certain document element and the search condition associated therewith, the document element condition determination means 218 searches the document structure in the document part storage unit 122 and finds that the given document element is It is determined whether the condition is satisfied. The search condition here relates to not only the document element to which the ID is given, but also all the document elements related to the document element, such as its parent document element or child document element. And handle this.

Next, FIGS. 12 and 13 show the second embodiment of the present invention.
9 is a flowchart showing a processing flow of the embodiment.
The entire operation of the second exemplary embodiment of the present invention will be described in detail with reference to FIGS.

Steps B1-B8 in FIG. 12 and FIG.
B10-B11 correspond to FIGS. 3 and 4 respectively.
Since the processing is the same as that shown in FIG.
In the first embodiment, if it is determined in step B9 of FIG. 4 that there is another unprocessed condition, a new search condition expression is generated and the search process by the full-text search unit 130 is continued. In the second embodiment of the present invention, it is determined whether the number of document elements (IDs) obtained as a result of the search processing in step B7 or B8 is one (step C1 in FIG. 13).

If the number of document elements is one, the ID
Is sent to the document element condition determining means 218, and the document element condition determining means 218 executes the condition determination using the information in the document part storage unit 122 (FIG. 13 Step C2).

In the second embodiment of the present invention, the processing of the document element condition judging means 218 is executed when only one search result is obtained according to each condition. If the speed is sufficiently high, the number of search results may be increased to a plurality, as long as the speed is not slower than the process of the full-text search means 130.

Next, the operation and effect of the second embodiment of the present invention will be described.

In the second embodiment of the present invention, when the number of search results based on some conditions in a compound query is one or sufficiently small, the search means is changed from the full-text search means 131 to the document element condition determination means 218. And the remaining search conditions are evaluated collectively for the document structure storage unit 122, so that the full-text search by the full-text search execution unit 131 is performed for each condition constituting the compound query. It is possible to avoid the processing to be performed once and to save the processing cost in the data processing device 200.

[Embodiment 2] Next, in order to describe the second embodiment of the present invention in more detail, an embodiment to which a specific example is applied will be described. Note that the configuration and processing flow of an example of the present invention are the same as those of the above-described second embodiment of the present invention.

As shown in FIGS. 5, 6, 7 and 8, when a structured document is registered, it is assumed that a compound query as shown in FIG. 14 is given.

Since there is no inclusive relation between the document elements of <item>, <exp>, and <title>, the condition that the first “<item> includes the character string“ Careeroo ”” is processed (FIG. 12). Steps B1-B3).

It is assumed that, as a result of the search by the full-text search execution means 131, only one document element satisfying the condition (ID = n-2) is obtained. In the present embodiment, this is detected (step C1 in FIG. 13), and the document element ID and all the remaining conditions include a character string "ketchup" in <exp> and a character string "delicious" in <title>. (Condition 3) to the document element condition determination means 218.

The document element condition judging means 218 obtains information of the document element corresponding to ID = n−2 from the document part storage unit 122, and traces all the document elements related to the document element with the information as a starting point. The condition 3 is determined for them (step C2 in FIG. 13).

Since the document element of ID = n-2 does not satisfy the condition 3, the document (element) extracting means 117 cannot extract the information of the document element (step B10 in FIG. 13), and the output device 140
Is displayed to indicate that there is no document satisfying the condition (step B11 in FIG. 13).

Next, a third embodiment of the present invention will be described.

Referring to FIG. 15, the third embodiment of the present invention includes a recording medium 400 on which a structured document search program is recorded. The recording medium 400 may be a magnetic disk, a semiconductor memory, or another recording medium.

The structured document search program is executed by the recording medium 40
Read from 0 to the data processing device 300, the data processing device
Control the operation of 300. The data processing device 300 performs the following processing under the control of the structured document search program, that is, the data processing device 100 in the first and second embodiments.
And the same processing as the processing by 200 is executed.

First, when a structured document is provided to the structured document management means 210 by the input device 150, information representing the document body and the document structure is generated in the storage device 120. The information representing the document structure is: It is stored in the document part storage unit 122. In this state, when a query for a structured document is given by the input device 150, the structured document management means 210 analyzes the query, and the query processing means 213 or the full text retrieval means 130 The document is searched. The retrieved document element or structured document is output to the output device 140 in an appropriate format.

[0114]

As described above, according to the present invention, the following effects can be obtained.

A first effect of the present invention is that a compound query composed of a plurality of search conditions including a condition relating to the document structure of a structured document and a condition relating to a character string included in a document element is processed only by the full-text retrieval means. Therefore, the processing speed is improved as compared with the case where the structure information of the document element is directly accessed and the condition is determined.

The reason is that, in the present invention, the full-text index is generated so that the information of the inclusion relation between the document elements and the structured document ID is included in the full-text index used by the full-text search means. It is.

The second effect of the present invention is that, when the number of search results based on some conditions in a compound query is one or sufficiently small, the remaining search conditions can be evaluated collectively.
That is, the processing speed is improved.

The reason is that, in the present invention, when it is detected that the number of search results is one or sufficiently small and unprocessed query conditions still remain, the search means is changed from full-text search to This is for changing to a means for directly executing the condition determination on the structure information of the document.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a first exemplary embodiment of the present invention.

FIG. 2 is a flowchart for explaining an operation of the first exemplary embodiment of the present invention, and is a flowchart showing an operation of a data registration process.

FIG. 3 is a flowchart illustrating an operation of the first exemplary embodiment of the present invention, and is a flowchart (part 1) illustrating a search operation;

FIG. 4 is a flowchart illustrating an operation of the first exemplary embodiment of the present invention, and is a flowchart (part 2) illustrating a search operation;

FIG. 5 is a diagram for specifically explaining the operation of the first embodiment of the present invention, and is a diagram illustrating an example of a DTD of an SGML document as an example of a processing target.

FIG. 6 is a diagram specifically illustrating an operation of the first exemplary embodiment of the present invention, and is a diagram illustrating an example of an SGML document that is an example of a processing target.

FIG. 7 is a diagram for specifically explaining the operation of the first exemplary embodiment of the present invention, and is a diagram showing structure information of an SGML document as an example of a processing target.

FIG. 8 is a diagram for specifically explaining the operation of the first exemplary embodiment of the present invention, and is a diagram illustrating an example of a text for which a full-text index is generated.

FIG. 9 is a diagram for specifically explaining the operation of the first exemplary embodiment of the present invention, and is a diagram illustrating an example of a search condition.

FIG. 10 is a diagram for specifically explaining the operation of the first exemplary embodiment of the present invention, and is a diagram illustrating an example of a search condition.

FIG. 11 is a block diagram illustrating a configuration of a second exemplary embodiment of the present invention.

FIG. 12 is a flowchart for explaining the operation of the second exemplary embodiment of the present invention, and is a flowchart showing a search operation (part 1);
It is.

FIG. 13 is a flowchart for explaining the operation of the second exemplary embodiment of the present invention, and is a flowchart showing a search operation (part 2);
It is.

FIG. 14 is a diagram for specifically explaining the operation of the first exemplary embodiment of the present invention, and is a diagram illustrating an example of a search condition.

FIG. 15 is a block diagram showing a configuration of a third exemplary embodiment of the present invention.

FIG. 16 is a diagram for specifically explaining the operation of the first exemplary embodiment of the present invention, and is a diagram illustrating an example of a search result.

[Explanation of symbols]

 100, 200, 300 Data processing device 110, 210 Structured document management means 111 Structured document registration means 112 Document element decomposition / analysis means 113, 213 Query processing means 114 Query analysis means 115, 215 Query execution means 117 Document (element) Extraction unit 218 Document element condition determination unit 120 Storage device 121 Text storage unit 122 Document part storage unit 123 Full-text index storage unit 130 Full-text search unit 131 Full-text search execution unit 132 Full-text index generation unit 140 Output device 150 Input device 400 Recording medium

──────────────────────────────────────────────────続 き Continuation of front page (56) References JP-A-7-44579 (JP, A) JP-A-6-28403 (JP, A) JP-A 8-241332 (JP, A) JP-A-5-241332 158984 (JP, A) JP-A-8-255155 (JP, A) JP-A-6-301721 (JP, A) JP-A-4-217073 (JP, A) JP-A-7-225771 (JP, A) JP-A-7-319918 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) G06F 17/30 G06F 12/00 JICST file (JOIS)

Claims (1)

(57) [Claims] 1. A structured document registering means for receiving a structured document input from an input means and storing the full text of the structured document in a text storage unit, and storing the input text of the structured document in a document A document element disassembling / storing means for disassembling each element, assigning a unique ID (identification number) thereto, and storing a relation between the text corresponding to each of the document elements and the document element in a document part storage unit; A full-text index is generated in units of the structured document and the document elements constituting the structured document,
At this time, the text for the document element includes the ID of the document element and the ID of the parent document element containing the document element.
A full-text index generating means for generating a full-text index for the text in which the ID of the original document is added, and a text and document structure using the full-text index generated by the full-text index generating means. A full-text search executing means for executing a search relating to the structured document; a query analyzing means for receiving a search request for the structured document input from the input means and analyzing a query condition; Based on the structure information of the obtained document, a structure condition determining unit that determines only a condition related to a structure, and matches the search condition using the full-text search unit or the structure condition determining unit according to the analysis of the query analysis unit. A query execution unit for executing a search for a document or a document element; and the full-text search unit. Alternatively, information of a document element corresponding to the corresponding document element ID is extracted from the document component storage unit from the document element ID obtained as a result of the search by the structure condition determination unit, and further stored in the text based on the original document ID. And a document element extracting means for controlling to extract and output the text of the original structured document from the section to the output means, and when a compound query composed of a plurality of conditions is given, the query executing means Then, a search is performed for each of the plurality of conditions one by one. At this time, in the second and subsequent condition searches, a new query condition that embeds the search result of the previously executed condition as a condition and narrows the search target set is used. The ID of a document element and a search condition related to the document element such as a parent document element or a child document element starting from the document element are given as inputs. And a document element condition determining unit that searches a document structure in the document part storage unit and determines whether the document element satisfies the condition. apparatus.