JP2962287B2 - Structured document search device and machine-readable recording medium recording 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 retrieval technique for a structured document in which one document is composed of a plurality of logical structures (zones), and more particularly, retrieves only a zone designated by a user. The present invention relates to a technology for searching a structured document for an object.

[0002]

2. Description of the Related Art In recent years, SGML (Standard Generalized)
Structured documents, such as Markup Languages), in which one document is composed of a plurality of zones, have been frequently used. Along with this, performing a search (zone search) targeting only a specific zone of a structured document as a search target has become an important function in performing a highly accurate search.

As a conventional technique for performing a zone search, for example, a search condition is set to “a document including a keyword between a start tag and an end tag of a zone to be searched” and character string collation is performed for the entire text. In some cases, a document that satisfies the search condition is searched by performing the search.
However, in this method, since the character string matching is performed on the entire text, the search time becomes extremely long.
In order to solve such a problem, a technique has been proposed in which an unnecessary portion of a text (a zone other than a zone designated as a search target) is skipped and a search is performed (for example, Japanese Unexamined Patent Application Publication No. No. 8-16600). According to this technique, the range in which character string matching is performed is reduced, so that the search time can be shortened as compared with the above-described conventional example. However, in the search for large-scale data of gigabyte class, even if the character string collation range can be reduced to, for example, about 1/10 by skipping unnecessary parts, character string collation can be performed for 100 megabyte class text. You don't want fast searches because you need to do it.

As described above, it is difficult to perform a high-speed search by a method of performing a zone search by character string matching.
It is conceivable to perform a zone search using a full-text index created to enable high-speed search of large-scale data. The full-text index is a key information part in which a key character string is stored as key information, and for each key character string stored in the key information part, a document identifier of a document in which the key character string exists, and a position in the document as positional information. And a position information section stored as a. The following three methods (A) to (C) can be considered as a method of performing a zone search using such a full-text index.

(A) A method in which information about a zone is stored in a key information section. (B) A method in which information about the zone is provided in the position information section. (C) A method of creating an index for a zone different from the full-text index.

[0006] In the method (A), key information composed of a pair of a key character string and a zone name of a zone in which the key character string exists is stored in the key information part of the full-text index. At this time, for key character strings existing in a plurality of zones, key information including pairs of key character strings and zone names is stored for each zone. The position information section stores the document identifier of the document corresponding to each keyword and the position in the document. At the time of search, a document including the keyword is found in the zone of the zone name by searching the full-text index using the zone name and the keyword designated by the user as keys.

In the method (B), the position information stored in the position information section is provided with the zone name in addition to the document identifier and the position in the document as the position information. At the time of the search, first, a full-text index is searched using the keyword specified by the user as a key, thereby obtaining all the position information of the document including the keyword, and thereafter, the position information specified by the user from the position information is obtained. A zone search is performed by selecting location information including a zone name.

In the method (C), apart from the full-text index, a zone index in which the start position and end position of each zone are stored for all documents to be searched is created. Then, at the time of retrieval, first, by searching the full-text index, the position information of the document including the keyword specified by the user is obtained. Thereafter, the zone index is searched, and the start position and the end position of the zone specified by the user in the document are obtained. Further, based on the position in the document in the position information and the acquired start position and end position of the zone, it is determined whether or not the keyword specified by the user exists in the zone specified by the user of the document. By performing a check, a zone search is performed (for example, see JP-A-8-314).
966).

[0009]

However, the method (A) described above has a problem that the size of the full-text index increases because the number of key information increases in proportion to the number of zones. Further, there is a problem that the search speed is reduced due to an increase in the size of the full-text index.

Further, the method (B) has a problem that the size of the full-text index increases because the capacity of the position information part of the full-text index increases. Further, there is a problem that the retrieval speed is reduced because the amount of reading from the position information section is increased.

Further, the method (C) requires a zone index storing the start position and end position of each zone of each document to be searched, so that the index size becomes large. is there. Furthermore,
For each corresponding document found by searching the full-text index, it is necessary to search the index for the zone and check whether the keyword specified by the user exists in the zone specified by the user. However, there is a problem that the search speed is reduced due to overhead.

It is therefore an object of the present invention to reduce the index size and increase the search speed in zone search using a full-text index.

[0013]

In order to achieve the above object, a structured document retrieval apparatus according to the present invention comprises: a document storage means for storing a plurality of structured documents each comprising a plurality of zones;
A zone information table in which information indicating the position of each zone in the zone position conversion document is stored, and a position in each of the zones in the structured document stored in the document storage means indicated by the contents of the zone information table In-document position conversion means for creating a zone position conversion document moved to
The key character string, the document identifier of the structured document in which the key character string exists, and the key character string in the zone position conversion document based on the zone position conversion document created by the intra-document position conversion means. An index creating means for creating an index in which the positions in the document are stored in correspondence with each other; a search condition input means for receiving a search condition expression including a zone name and a keyword of a zone to be searched; The index is searched using the keyword in the received search condition expression as a key, and based on the document identifier of the structured document in which the keyword is obtained as a result, the position in the document, and the contents of the zone information table, A keyword for finding a document identifier of a structured document in which the keyword exists in a zone indicated by a zone name in the search condition expression It is obtained by a search means.

In this configuration, as a preparation for performing a zone search, the in-document position conversion means indicates each zone in the structured document stored in the document storage means by the contents of the zone information table. A zone position conversion document that has been moved to a position is created, and an index creation unit generates a key character string and a structured document in which the key character string exists based on the zone position conversion document created by the in-document position conversion unit. Then, an index is created in which the document identifier of the key character string and the position of the key character string in the document in the zone position conversion document are stored correspondingly.

When a user inputs a search condition expression including a zone name and a keyword of a zone to be searched at the time of zone search, the search condition input means accepts the search condition expression and the keyword search means accepts the search condition expression in the search condition expression. An index is searched using a keyword as a key, and a document identifier of a structured document in which the keyword is obtained as a result,
Based on the position in the document and the contents of the zone information table,
A document identifier of a structured document in which the keyword exists in a zone indicated by a zone name in the search condition expression is obtained.

[0016]

Embodiments of the present invention will now be described in detail with reference to the drawings.

FIG. 1 is a block diagram of an embodiment of the present invention. A document storage means 1, a document position conversion means 2, a zone information table 3, an index creation means 4, an index 5, a keyword search means 6, a search condition input means 7, a keyword search result storage means 8, a logical condition analysis means 9, and a search result output means 10.

The document storage unit 1 stores a plurality of structured documents to be searched. FIG. 2 shows the document storage unit 1.
FIG. 4 is a diagram showing an example of the contents of the above. In this embodiment, for the sake of simplicity, the document storage means 1 stores the document identifiers ID1 and ID1.
It is assumed that two structured documents 21 and 22 of ID2 are stored. Each of the structured documents 21 and 22 is composed of a zone having a zone name “whole”, “name of invention”, “abstract”, “purpose”, and “composition”, and each zone has a zone start tag and a zone end. It is divided by tags and the like.

The zone information table 3 stores information indicating the position of each zone in the zone position conversion document created by the in-document position conversion means 2. FIG. 3 is a diagram showing an example of the contents of the zone information table 3, in which a zone name,
Zone position information indicating where in the zone position conversion document the zone with the zone name is to be located is stored. In the example of FIG. 3, the zones of the zone names “entire”, “name of the invention”, “abstract”, “object”, and “structure” are respectively referred to as “first character to 2000th” in the zone position conversion document.
"The first character to the 500th character", "the 50th character"
This means that the characters are arranged in the first to 2,000-th characters, the 501st to 1000th characters, and the 1001 to 2000th characters.

The in-document position conversion means 2 refers to the zone information table 3 and, for the structured document stored in the document storage means 1, converts the character string of each zone into the zone position information in the zone information table 3. Has a function of creating a zone position conversion document moved to the position indicated by. Therefore, the character string of each zone existing at a different position in each structured document is present in the same range in the zone position conversion document created by the in-document position conversion unit 2.

The index creation means 4 has a function of creating an index 5 based on the zone position conversion document corresponding to each structured document created by the in-document position conversion means 2. The index 5 has a key information section 51 and a position information section 52 as shown in FIG. The key information section 51 stores key information such as N character sets and words. The position information section 51 stores the document identifier of the structured document in which the key information exists and the zone position conversion document in which the key information exists. Is stored. Here, the location information "i-
"j" represents the j-th character of the zone position conversion document corresponding to the structured document having the document identifier IDi. Therefore, in the first area of the example of FIG. 4, the character "sentence" exists in the structured document 21 of the document identifier ID1, and the first character, the 504th character in the zone position conversion document 21 'corresponding thereto. And that the character "sentence" exists in the structured document 22 of the document identifier ID2, and in the corresponding zone position conversion document 22 ', exists in the first character and the 501st character. ing.

The search condition input means 7 has a function of receiving a search condition expression input by a user, a function of using the logical condition analysis means 9 to decompose the search condition expression into search items,
It has a function of passing a search item to the keyword search means 6 and the like. The search condition expression input by the user includes one or more search items including a pair of a zone name and a keyword to be searched. When a plurality of search items is included,
Each search item is linked by logical operation symbols such as AND and OR. FIG. 5 shows one of the search condition expressions input by the user.
FIG. 4 is a diagram showing an example, in which two search items are logical operation symbols A;
Connected by ND. This search condition expression indicates a search for a structured document that includes the keyword “search” in the zone with the zone name “name of the invention” and the keyword “index” in the zone with the zone name “summary”.

The keyword search means 6 searches the index 5 using the keyword in each search item passed from the search condition input means 7 as a key, and the keyword in the search item appears for each search item. A function for obtaining all of the document identifier and the position in the document, a function for obtaining the zone position of the zone indicated by the zone name included in the search item with reference to the zone information table 3 for each search item, For each search item, if there is a search result indicating the position within the zone position in the search results (document identifier, position in the document), the document identifier in the search result and the order number of the search item It has a function of pairing information indicating whether or not the information is a pair and storing the information in the keyword search result storage unit 8.

The logical condition analysis means 9 compares the search result (document identifier) for each search item stored in the keyword search result storage means 8 with each search item in the search condition expression received by the search condition input means 7. It has a function of obtaining a document identifier of a structured document that satisfies the above search condition expression based on a logical operation symbol to be combined.

The search result output means 10 has a function of extracting a structured document having the document identifier obtained by the logical condition analysis means 9 from the document storage means 1 and outputting the structured document to an output device (not shown) such as a printer or a CRT. Have.

FIG. 6 is a flowchart showing a processing example of the in-document position conversion means 2, FIG. 7 is a flowchart showing a processing example of the index creation means 4, FIG. 8 is a flowchart showing a processing example of the search condition input means 7, and FIG. Is a flowchart showing a processing example of the logical condition analysis means 9 when a search condition expression is passed from the search condition input means 7;
FIG. 10 is a flowchart showing a processing example of the logical condition analysis means 9 when an end notification is sent from the keyword search means 6;
FIG. 11 is a flowchart showing a processing example of the keyword search means 6, and the operation of this embodiment will be described below with reference to the drawings.

First, the operation at the time of creating the index 5 will be described.

When the index 5 is created, the in-document position conversion means 2 sets the document storage means 1 as shown in the flowchart of FIG.
, One unprocessed structured document is input (S61). Thereafter, the in-document position conversion means 2 creates a zone position conversion document in which the character string in each zone of the input structured document is moved to the position indicated by the zone position information in the zone information table 3 (S63). Then, the created zone position conversion document and the document identifier of the structured document input in S61 are passed to the index creation means 4 (S64). The above processing is performed until there is no unprocessed structured document (NO in S62).
Repeat.

In the case of this embodiment, the document storage means 1 has
Structured document 2 of document identifiers ID1 and ID2 as shown in
1 and 22 are stored, and the contents of the zone information table 3 are as shown in FIG. 3, so that the in-document position conversion means 2 executes the zone position conversion document 2 as shown in FIG.
1 'and 22' are sequentially created and passed to the index creation means 4.

That is, the in-document position conversion means 2 refers to the contents of the zone information table 3 shown in FIG. 3, and according to the contents, the character string "character name" existing in the zone "name of the invention" of the structured document 21. The document search device is moved to the position starting from the first character, and the character string “Search the document at high speed” existing in the zone “purpose” is moved to the position starting from the 501st character, and the zone “ The character string "index creation means ..." in "" is moved to a position starting from the 1001st character to create a zone position conversion document 21 'as shown in FIG. 12 and create an index together with the document identifier ID1. Pass to means 4. Similarly, the in-document position conversion means 2 moves the character string “document processing device” existing in the zone “name of the invention” of the structured document 22 to a position starting from the first character, Is moved to the position starting from the 501st character, and the character string "search means ..." in the zone "composition" is moved to the position starting from the 1001st character. Is created and transferred to the index creation means 4 together with the document identifier ID2 as shown in FIG.

As can be seen from FIG. 12, each of the zone position conversion documents 21 'and 22' after the conversion processing has the zone names "whole", "name of invention", "summary", "purpose", and "structure".
Will always exist at the position indicated by the zone position information in the zone information table 3.

When the index conversion unit 4 receives the zone position conversion document and the document identifier from the intra-document position conversion unit 2, the index generation unit 4 pays attention to the head position of the zone position conversion document as shown in the flowchart of FIG. 7 (S71). . And at the point of interest,
It is determined whether there is any character to be stored in the key information section 51 of the index 5 (S72). To determine whether or not the character should be stored, for example, a character that does not need to be stored, such as a blank character or a punctuation mark, is determined in advance, and it is determined whether or not the character present at the target position is another character. It is done by doing.

If it is determined in S72 that the character to be stored exists at the position of interest (S72: YE
S) determines whether the character has already been stored (S).
73). If it is determined that the character has not been stored, the character present at the position of interest is stored in the key information part 51 of the index 5, and the document identifier and the position within the document (current position of interest) passed from the in-document position conversion means 2 are stored. Is stored in the position information section 52 (NO in S73, S7
4). On the other hand, if it is determined that the document has been stored, the position information including the document identifier and the position in the document passed from the position conversion unit 2 in the document to the position information unit 52 is stored in the position information unit 52 ( S73 is YES, S75).

When the processes of S74 and S75 are completed, the index creating means 4 moves the target position to the next position (S7).
6) The same processing as described above is performed. Also, when it is determined in S72 that the character to be stored does not exist at the position of interest, the process of S76 is performed.

The above processing is performed until the end of the zone position converted document passed from the in-document position converting means 2 (S77: YE
S) Repeat.

In the case of this embodiment, the index creation means 4
Includes a zone position conversion document 21 ', as shown in FIG.
Since the document identifier ID1, the zone position conversion document 22 ', and the document identifier ID2 are passed, the following processing is performed in the index creating means 4.

When the zone position conversion document 21 'and the document identifier ID1 are passed from the in-document position conversion means 2, the index creation means 4 pays attention to the head position.
The character “sentence” is stored in the key information part 51, and the position information “1-
"1" is stored in the position information section 52 (S71, S74).
Also, when the index creating unit 4 pays attention to the next position (the second character) in the zone position conversion document 21 ′,
The character "" at the position of interest is stored in the key information section 51, and the position information "1-2" is stored in the position information section 52 (S74). Further, for example, when the target position is set to the 504th character of the zone position conversion document 21 ′, the character “sentence” existing at the target position has already been stored. "Is stored in the entry corresponding to". "(S75).
Such processing is performed until the end of the zone position conversion document 21 '. Also when the zone position conversion document 22 'and the document ID are passed, the index creation means 4 performs the same processing as described above. As a result, the contents of index 5 are:
This is shown in FIG.

Here, for simplicity of description, the character length of the character string stored in the key information section 51 is one character, but the present invention is not limited to this.
It may be a character set or a word.

Next, the operation at the time of zone search will be described.

When performing a zone search, the user inputs to the search condition input means 7 a search condition expression including one or more search items consisting of a pair of a zone name and a keyword of a zone to be searched. As described above, in the case of a search condition expression including a plurality of search items, each search item is AND, O
They are connected by logical operation symbols such as R.

Now, for example, when the user sets a search condition expression as
The search condition expression “(name of invention = search) AND shown in FIG.
(Summary = index) "is input to the search condition input means 7. As described above, this search condition expression is a search for a structured document in which the character string “search” appears in the zone name “name of the invention” and the character string “index” appears in the zone name “summary”. Is to indicate.

When the user inputs a search condition expression "(name of invention = search) AND (summary = index)", the search condition input means 7 accepts the search condition and performs a logical condition analysis as shown in the flowchart of FIG. To the means 9 (S81, S8
2).

When the search condition expression “(name of invention = search) AND (summary = index)” is passed, the logical condition analysis means 9 converts the search condition expression into the first condition as shown in the flowchart of FIG. The search item “name of invention = search” and the second search item “summary = index” are divided into two search items, which are returned to the search condition input means 7 (S91,
S92).

Upon receiving the first and second search items “name of invention = search” and “summary = index” from the logical condition analysis means 9, the search condition input means 7 sends them to the keyword search means 6. Hand over (S83, S8 in FIG. 8)
4).

The keyword search means 6 sends the first and second search items “name of invention =
When “search” and “summary = index” are passed, FIG.
As shown in the flow chart of FIG.
11).

Now, for example, if attention is paid to the first search item “name of invention = search”, the keyword search means 6 firstly searches for the first search item “name of invention = search”.
By searching the index 5 using the keyword “search” in the key as a key, the document identifier of the zone position conversion document in which the keyword “search” appears and the position in the document are obtained (S113). In the case of the present embodiment, since the contents of the index 5 are as shown in FIG. 5, by performing S113, the keyword “search” is changed to the document identifier ID1.
Is required to appear in the third to fourth characters of the zone position conversion document 21 ′ and the document identifier in the 1001 to 1002 characters of the zone position conversion document 22 ′ of ID2.

Thereafter, the keyword search means 6 refers to the zone information table 3 and obtains the zone position of the zone indicated by the zone name "name of the invention" in the first search item (S114). In the case of the present embodiment, “first character to 500th character” is obtained as the zone position (see FIG. 3).

Next, the keyword search means 6 proceeds to S11
If there is a search result indicating the position in the zone position obtained in S114 among the search results of S3, the document identifier in the search result and information indicating that the document identifier is for the first search item Are stored as a pair in the keyword search result storage means 8 (S115). In this example, the search result of S113 is “Zone position conversion document 2 of document identifier ID1”.
3rd to 4th characters of 1 '"," document identifier is ID
In the second zone position conversion document 22 ′, the zone position obtained in S114 is “first character to 500th character” in the 1001st to 1002th characters, so that the keyword search unit 6 The identifier ID1 and information indicating that the identifier ID1 is for the first search item are stored in the keyword search result storage unit 8 as a pair.

Thereafter, the keyword search means 6 pays attention to the second search item "summary = index" (S11).
1) Perform the same processing as described above (S113 to S11)
5). In the case of the second search item, the keyword “index” is the zone position conversion document 2 of the document identifier ID1.
1 ′ appears at the 1001st to 1006th characters and the zone position of the zone with the zone name “abstract” is “501st to 2000th characters”. ID1 and information indicating that it is for the second search item are stored in the keyword search result storage means 8 as a pair.

When the above-described processing is performed on all the search items passed from the search condition input means 7 (S11)
2 is NO), the keyword search means 6 sends an end notification to the logical condition analysis means 9 (S116).

When the completion notice is sent from the keyword search means 6, the logical condition analysis means 9 searches for each search item stored in the keyword search result storage means 8 as shown in the flowchart of FIG. Based on the result and the logical operation symbol in the search condition expression passed from the search condition input unit 7, a document identifier of a structured document that satisfies the search condition expression is obtained, and is passed to the search result output unit 10 ( S10
1, S102).

In this example, the keyword search result storage means 8 stores the document identifiers “ID1” and “ID1” as the search results of the first and second search items, respectively. Since the logical operation expression that joins the first search item and the second search item is “AND”, the document identifier “ID” existing in both search results
"1" is passed to the search result output means 10.

The search result output means 10 outputs the document identifier "I
D1 ”, the structured document 21 whose document identifier is“ ID1 ”is read from the document storage unit 1, and the printer, C
Output to an output device (not shown) such as an RT.

FIG. 13 is a block diagram showing an example of the hardware configuration of the structured document search device shown in FIG. 1, and comprises a computer 131, a recording medium 132, and a storage device 133. The recording medium 132 is a magnetic disk, a semiconductor memory, or another recording medium.
A program for causing the computer 131 to function as a structured document search device is recorded.

The program recorded on the recording medium 132 is read by the computer 131, and by controlling the operation of the computer 131, the in-document position conversion means 2 and the index creation means 4 shown in FIG. Keyword search means 6, search condition input means 7, logical condition analysis means 9, search result output means 1
0 is realized. The document storage unit 1, the zone information table 3, the index 5, the keyword search result storage unit 8
Are configured on the storage device 133.

[0056]

As described above, according to the structured document search apparatus of the present invention, the index size can be reduced as compared with the conventional technique of performing a zone search using a conventional full-text index. In addition, the search speed can be increased. The reason is that because the index is created in such a way that the zone location information is common to all structured documents, it is possible to perform a zone search only by referring to a very small zone information table when searching. Because you can.

[Brief description of the drawings]

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

FIG. 2 is a diagram showing an example of contents of a document storage unit 1;

FIG. 3 is a diagram showing a content example of a zone information table 3;

FIG. 4 is a diagram showing an example of the contents of an index 5;

FIG. 5 is a diagram illustrating an example of a search condition expression.

FIG. 6 is a flowchart showing a processing example of the in-document position conversion means 2;

FIG. 7 is a flowchart showing a processing example of the index creation means 4;

FIG. 8 is a flowchart showing a processing example of a search condition input unit 7;

FIG. 9 is a flowchart showing a processing example of the logical condition analysis means 9 when a search condition expression is passed from the search condition input means 7;

FIG. 10 is a flowchart showing a processing example of the logical condition analysis means 9 when an end notification is sent from the keyword search means 6;

FIG. 11 is a flowchart showing a processing example of the keyword search means 6;

FIG. 12 is a diagram showing an example of a zone position conversion document created by the in-document position conversion means 2.

13 is a block diagram showing an example of a hardware configuration of the structured document search device shown in FIG.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... Document storage means 2 ... Document position conversion means 3 ... Zone information table 4 ... Index creation means 5 ... Index 6 ... Keyword search means 7 ... Search condition input means 8 ... Keyword search result storage means 9 ... Logical condition analysis means 10 ... Search result output means 21 and 22 Structured documents 21 ′ and 22 ′ Zone position conversion document 51 Key information part 52 Position information part 131 Computer 132 Recording medium 133 Storage device

Continuation of the front page (56) References JP-A-8-329116 (JP, A) JP-A-8-241332 (JP, A) "Power of high-speed full-text search", Nikkei Byte, October 1996, p. 142-167 (Heisei 8-9-22) Fukushima, Akamine "Flexible character string inversion method for high-speed full-text search (1) Method outline" Information Processing, Vol. 4, 1997, p. 334-335 (Heisei 9-4-15) (58) Fields investigated (Int. Cl. ⁶ , DB name) G06F 17/30 JICST file (JOIS)

Claims (5)

(57) [Claims] 1. A document storage unit in which a plurality of structured documents each including a plurality of zones are stored; a zone information table in which information indicating a position of each zone in a zone position conversion document is stored; An intra-document position conversion unit that creates a zone position conversion document in which each zone in the structured document stored in the storage unit is moved to a position indicated by the contents of the zone information table; Based on the created zone position conversion document, the key character string, the document identifier of the structured document in which the key character string exists, and the position of the key character string in the document in the zone position conversion document correspond to each other. Creation means for creating an index stored therein, and a search condition for receiving a search condition expression including a zone name and a keyword of a zone to be searched Searching means for searching the index by using a keyword in a search condition expression received by the search condition input means as a key, and a document identifier and a position in the document of a structured document in which the keyword obtained as a result is present; A keyword search unit for obtaining a document identifier of a structured document in which the keyword exists in a zone indicated by a zone name in the search condition expression based on the contents of the zone information table. Document retrieval device. 2. The zone information table stores a zone name and zone position information indicating a position of a zone having the zone name in a zone position conversion document. 2. The structured document search device according to claim 1, wherein: 3. The search condition expression has a format in which a plurality of search items composed of a zone name and a keyword of a zone to be searched are linked by a logical operation symbol. For each search item of the search condition expression received by the condition input means, the index is searched using a keyword in the search item as a key, and the search is performed based on the search result and the contents of the zone information table. A structure for obtaining a document identifier of a structured document in which the keyword exists in a zone indicated by a zone name in an item; and a document identifier for each search item obtained by the keyword search means and the search condition input. Based on a logical operation symbol connecting each search item in the search condition expression accepted by the means,
3. The structured document search device according to claim 2, further comprising a logical condition analysis unit for obtaining a document identifier of a structured document satisfying the search condition expression. 4. The structured document search according to claim 3, further comprising: a search result output unit that reads out and outputs a structured document of the document identifier obtained by the logical condition analysis unit from the document storage unit. apparatus. 5. A document storage unit storing a plurality of structured documents each including a plurality of zones, and a zone information table storing zone position information indicating the position of each zone in the zone position conversion document. An intra-document position conversion unit that creates a zone position conversion document in which each computer in the structured document stored in the document storage unit is moved to a position indicated by the contents of the zone information table. Based on the zone position conversion document created by the in-document position conversion means, the key character string, the document identifier of the structured document in which the key character string exists, and the document position of the key character string correspond to each other. Index creation means for creating a stored index, a search condition for receiving a search condition expression including a zone name and a keyword of a zone to be searched Input means, searching the index by using a keyword in a search condition expression received by the search condition input means as a key, and obtaining a document identifier, a position in the document, and a zone information table of a document including the keyword obtained as a result. A machine-readable recording in which a program for functioning as keyword search means for obtaining a document identifier of a structured document in which the keyword is present in a zone indicated by a zone name in the search condition expression based on the contents of Medium.