JP2004258912A - Document retrieval device, method and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a document retrieval device and method capable of quickly and efficiently retrieving document data where structured data and unstructured data are intermingled. <P>SOLUTION: Desired document data are retrieved from a storage means which stores a plurality of document data, including structured data to be retrieved by a first search method which is one of two different retrieval methods, and unstructured data to be retrieved by a second search method which is the other of the two methods. First retrieval requirements for the structured data and second retrieval requirements for the unstructured data are input and the structured data among the document data stored in the storage means are retrieved using the first retrieval method, whereby document data including the structured data satisfying the first retrieval requirements are obtained. The unstructured data among the document data obtained are retrieved using the second retrieval method, whereby the document data including the unstructured data satisfying the second retrieval requirements are obtained. <P>COPYRIGHT: (C)2004,JPO&NCIPI

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a document search device and method.
[0002]
[Prior art]
In recent years, in addition to the conventional relational database (hereinafter referred to as RDB) technology, in recent years, XML (Extensible Markup Language: Media Fusion XML Lab.) A database technology using a markup language, such as “Improvement of Web application using ML database” (published by Softbank Publishing Co., Ltd., March 2001), has become widespread.
[0003]
As a search technique for XML data, XQL (XML Query Language: [URL] http://www.w3.org/TandS/QL/QL98/pp/xql.html) and the like are known. On the other hand, as a search method of document information, a text search, particularly a full text search technique, has been well known.
[0004]
Generally, XML technology is data (semi-structured data) that is positioned between structured data as handled by the conventional RDB technology and unstructured data as handled by the conventional text search technology. It is said to be suitable for processing.
[0005]
By the way, with the spread of XML, there has been a growing demand for converting data conventionally handled by text search into XML data and searching for XML. The cost of complete XML conversion of an existing text document is enormous, and for the time being it is likely that only limited information data in the document will be converted to XML. Further, for management purposes, in many cases, XML conversion is performed by adding XML tags to the document itself to be converted to XML. A document part that is not a target of XML conversion is processed as a comment part by surrounding it with an appropriate tag. In this way, the original data and the XML data can be unitarily managed in file units, and it is obvious which part of the original data has been converted to XML and which part has not yet been converted. Since all the information of the original data is stored in the file, it is convenient to simply add tags even in the future XML conversion work, which is convenient.
[0006]
When an XML search is performed on data in which a part of such information is converted into XML, there is a problem that a data portion that has not yet been converted into XML is not XML data, so that it cannot be searched well as a matter of course.
[0007]
In some cases, a full-text search is used for searching a structured document such as an XML document (for example, see Patent Document 1). In this method, the entire XML document is processed in advance including the comment part as described above to create a full-text search index. For this reason, since a comment part that is not necessary for a normal XML search is also included in the index, the index becomes huge and the search speed is reduced.
[0008]
In addition, there is a method in which the data is converted into XML including the layout information of the document and managed in a unified manner (for example, see Patent Document 2).
[0009]
However, none of the above-described techniques can search for document data in which XML-formatted data and data portions not yet converted to XML are mixed at high speed and efficiently.
[0010]
[Patent Document 1]
JP 2001-167087 A
[0011]
[Patent Document 2]
JP 2000-99543 A
[0012]
[Problems to be solved by the invention]
As described above, conventionally, it is difficult to efficiently and efficiently search document data in which structured data converted to XML data and unstructured data not converted to XML data are mixed. was there.
[0013]
SUMMARY OF THE INVENTION In view of the above problems, an object of the present invention is to provide a document search apparatus and method capable of searching for document data in which structured data and unstructured data are mixed at high speed and efficiently.
[0014]
[Means for Solving the Problems]
The present invention relates to structured data to be searched by a first search method, which is one of two different search methods, and a second search method, which is another one of the two search methods. For searching desired document data from storage means storing a plurality of document data including unstructured data to be searched for, wherein a first search condition for the structured data; By inputting a second search condition for the structured data and searching the structured data in each document data stored in the storage means using the first search method, the first search condition is obtained. Document data including structured data that satisfies the above condition, and searches the unstructured data in each of the obtained document data using the second search method, thereby satisfying the second search condition. Unstructured data Obtains document data including, and outputs the document data obtained as a result. According to the present invention, for document data in which structured data and unstructured data are mixed, a search is performed on the structured data using a search method suitable for searching the structured data. For structured data, a search is performed using a search method suitable for unstructured data search, and the results obtained by first performing a search on the structured data are used as the search target. By performing a search for, document data in which structured data and unstructured data are mixed can be searched at high speed and efficiently.
[0015]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, embodiments of the present invention will be described with reference to the drawings.
[0016]
FIG. 1 shows an example of the configuration of a document search system according to the present embodiment. An input unit 1, a search processing unit 2, a first search unit 3, a second search unit 4, a document storage unit 5, an output unit It is composed of a unit 6.
[0017]
The document storage unit 5 stores a plurality of document data. The plurality of document data stored here are originally unstructured (not marked up with a predetermined tag) unstructured document data, and are surrounded by XML tags for structuring. A mixture of structured (XML) data and unstructured data created by inserting markup data or marking up by surrounding predetermined data in a document with XML tags Document data.
[0018]
As described above, the structured data (structured data) is data including at least one element including a predetermined XML tag for structuring and data surrounded by the tag. Each element can be said to be an XML tag (also referred to as an element name) associated with data corresponding to the XML tag.
[0019]
The unstructured data area in each document is called a comment area here, and the data in the comment area is also called unstructured data. In order to distinguish between the comment area (unstructured data part) and the structured data part in each document, the comment area is enclosed here by the symbols “<! −−” and “−−>”. To A region surrounded by this symbol is processed in a normal XML processing system (for example, in the first search unit 3 in FIG. 1) while ignoring it as a comment.
[0020]
Note that a specific tag for defining a comment area may be defined in advance, and an area surrounded by such a specific tag may be handled as the comment area. In this case, it is necessary to newly add a function of ignoring an area surrounded by the specific tag in the document as a comment area to the processing of the first search unit 3 described later.
[0021]
FIG. 2 shows a specific example of an original document which is unstructured document data that does not include structured data before XML conversion. The content is a notice of the holding of the meeting. FIG. 3 is an XML version of a part of the document data shown in FIG. As described above, the XML document is converted into an XML document by associating data corresponding to a predefined XML tag in the unstructured document data with the XML tag (for example, markup with the XML tag here). . In this example, data representing the name of the meeting, “date and time”, and “location” are marked up with XML tags. That is, the name of the conference is marked up with a <name> tag, “date” is marked up with a <date> tag, and “location” is marked up with a <place> tag. Here, in the element starting with the <date> tag, tags such as <year>, <month>, <day>, <day of the year>, <hour from>, <hour to>, and year, month, It has a structure that includes elements that are formed by marking up data representing the day, the day of the week, the start time, and the end time. An element starting with a <place> tag has a structure including an element formed by marking up address data with an <address> tag.
[0022]
In the document data shown in FIG. 3, the format structure such as line breaks and indents of the original document data (see FIG. 2) is stored as it is. Except for the portion enclosed by the tags as described above, the comment region is enclosed by symbols “<! −−” and “−−>”.
[0023]
Structured data marked up with tags such as <name> tag, <date> tag and <place> tag in each document stored in the document storage unit 5 is used for XML search in the first search unit 3 described later. The area (comment area) of the unstructured data becomes a search target, and becomes a search target of a full-text search (or a full-text search) in the second search unit 4 described later.
[0024]
The input unit 1 has a graphical user interface (hereinafter, GUI) for inputting search conditions for searching for a desired document. Here, the search conditions include a search condition for structured data in the document data and a search condition for unstructured data in the comment area. Here, the former is called a first search condition, and the latter is called a second search condition.
[0025]
FIG. 4 shows an example of a search condition input screen displayed on a predetermined display by the input unit 1. On the search condition input screen shown in FIG. 4, the user inputs a desired character string (keyword) in blank spaces corresponding to the items of “meeting name”, “date and time”, and “location”, thereby performing a search for structured data. A condition (first search condition) can be specified. Items such as “meeting name”, “date and time”, and “location” correspond to tag names included in the document data stored in the document storage unit 5. That is, "meeting name" corresponds to a <name> tag, "date and time" corresponds to a <date> tag, and "location" corresponds to a <place> tag.
[0026]
In the search condition input screen shown in FIG. 4, the user inputs a desired character string (keyword) in a blank portion corresponding to the item of "others", so that the search condition for the unstructured data (second Search conditions) can be specified.
[0027]
The search processing unit 2 uses the search condition input from the input unit 1 as a search condition (first search condition) used in the XML search in the first search unit 3 and a text search in the second search unit 4 And the search condition (second search condition) used in (1). The first search condition is converted into a first search condition sentence in a format that can be interpreted by the first search unit 3 and output to the first search unit 3, and the second search condition is converted to the second search unit 4 Is converted to a second search condition sentence in a format that can be interpreted and output to the second search unit 4. Then, based on the search results output from each of the first search unit 3 and the second search unit 4, display data to be presented to the user is generated and output to the output unit 6.
[0028]
The output unit 6 displays the display data created by the search processing unit 2 on, for example, a predetermined display. A GUI is provided so that the user can browse the search results.
[0029]
The first search unit 3 uses a well-known XML search method to structure each document data stored in the document storage unit 5 based on the first search condition sentence passed from the search processing unit 2. Document data that satisfies the first search condition contained in the first search condition sentence is searched for data as a search target. For example, among the documents stored in the document storage unit 5, an element (component) of the tag name (element name) specified as the first search condition is included, and the first search condition is included in the element. Is searched for a document to which data including a character string specified as "" is associated. The document obtained as a search result is returned to the search processing unit 2.
[0030]
The second search unit 4 is based on the second search condition sentence passed from the search processing unit 2 and based on the comment of each document searched by the first search unit 5 passed from the search processing unit 2 Using a well-known full-text search method as a search target in the area, a document in which the unstructured data in the comment area satisfies the second search condition included in the second search condition sentence is searched. For example, a document having a comment area including a character string specified as a second search condition is searched from the documents searched by the first search unit 5. The document obtained as a search result is returned to the search processing unit 2.
[0031]
Next, the processing operation of the search processing unit 2 will be described with reference to the flowcharts shown in FIGS.
[0032]
For example, on the search condition input screen shown in FIG. 4, the user has specified a character string "December 12" as a search condition in a blank space corresponding to the item "Date" as shown in FIG. And
[0033]
The search condition is transmitted from the input unit 1 to the search processing unit 2 (Step S1). The search processing unit 2 checks the search condition of the user and separates the search condition into the first search condition and the second search condition (step S2). First, a first search condition sentence is generated from the first search condition. In the case of the above example, “date” corresponds to the tag name <date>, and only the first search condition is input. Therefore, from the first search condition, a first search condition sentence for generating a document having an element of <data> tag and including the character string “December 12” is generated. . Then, the first search condition sentence is passed to the first search unit 3, a search is requested, and a search result is obtained (step S3). In the case of the above example, the first search unit 3 has an element of the <date> tag, and the element has a document including the character string “December 12” (that is, an element of the <date>tag;> A document in which the data associated with the tag includes the character string “December 12” is retrieved, and is passed to the retrieval processing unit 2.
[0034]
If the search condition input from the input unit 1 includes the second search condition (step S4), a second search condition sentence is generated from the second search condition, and the first search unit 3 Then, the document obtained by the search in step (1) and the second search condition sentence are transferred to the second search unit 4, and a search is requested (step S6 in FIG. 6).
[0035]
In the case of the above example, since the search condition input by the user is only the first search condition (step S4), display data is obtained from the document obtained as a result of the search by the first search unit 3. It is generated and passed to the output unit 6 (step S5 in FIG. 5). The output unit 6 displays the display data on a predetermined display.
[0036]
FIG. 8 shows an example of a display screen of display data, which is displayed by the output unit 8 and is generated from the document obtained as a result of the search by the first search unit 3. Here, two documents are obtained as a search result, and all or a part of the data associated with the tag name is displayed when corresponding to the meeting name, date and time, and location in the two documents. Is displayed. On this display screen, for example, by clicking the column of the conference name of each document with a mouse, it is possible to shift to a display screen displaying the entire searched document. Note that this display processing can be realized by a conventional HTML document or XML document processing technique.
[0037]
Next, the search processing unit 2 will be described with reference to the flowcharts shown in FIGS. 5 and 6 by taking, as an example, a case where the user inputs a search condition different from the above example (see FIG. 7) on the search condition input screen of FIG. Will be described.
[0038]
For example, on the search condition input screen shown in FIG. 4, the user specifies a character string "December 12" as a search condition in a blank space corresponding to the item "Date" as shown in FIG. At the same time, it is assumed that a character string "Tanaka" is specified as a search condition in a blank space corresponding to the item "others".
[0039]
The search condition is transmitted from the input unit 1 to the search processing unit 2 (Step S1). As described above, the search processing unit 2 checks the search condition of the user and separates the search condition into the first search condition and the second search condition (step S2). First, a first search condition sentence is generated from the first search condition. In the case of the above example, as described above, the first search condition for searching for a document having an element of <data> tag and including the character string “December 12” from the first search condition A sentence is generated, and the first search condition sentence is passed to the first search unit 3, requesting a search, and obtaining a search result (step S3).
[0040]
In the case of the above example, in the search condition input from the input unit 1, a character string of "Tanaka" is specified in a blank corresponding to the item of "other", and the character string corresponding to the item of "other" is specified. The character string entered in the blank is used as a search condition (second search condition) for unstructured data. That is, the character string "Tanaka" is obtained as the second search condition from the search condition input by the user (step S4).
[0041]
From the second search condition, the unstructured data in the comment area is extracted from the documents searched by the first search unit 5 for the document including the character string “Tanaka” given as the second search condition. A second search condition sentence for searching is generated. Then, the second search condition sentence is passed to the second search unit 4, a search is requested, and a search result is obtained (step S6 in FIG. 6). Note that a search key that is not used in markup of an XML document, such as the character string “Tanaka”, may be referred to as a “free word”.
[0042]
In the case of the above example, the second search unit 4 performs a full-text search on the comment area of each document searched by the first search unit 3 and searches the comment area including the character string “Tanaka”. Search for documents that have Then, the searched document is passed to the search processing unit 2.
[0043]
For example, since the comment area of the document shown in FIG. 3 includes the character string “Tanaka”, when this document is searched by the first search unit 3, the document is The part 4 also detects it as a search result.
[0044]
When, for example, the document shown in FIG. 3 is searched by the second search unit 4 (step S7), predetermined language processing is performed on each of the searched documents (step S9). For example, linguistic processing is performed on a portion where a character string (“Tanaka” in the above example) specified as the second search condition in each searched document appears. Here, a fixed number of characters before and after the appearance of the character string are extracted as KWIC (Key Word In Context).
[0045]
Then, display data of the search result is generated using the result of the language processing, and is output to the output unit 6 (step S10). The output unit 6 displays the display data on a predetermined display.
[0046]
FIG. 10 shows an example of a display screen of display data, which is displayed by the output unit 8 and is generated from the document obtained as a result of the search by the second search unit 4. Here, one document (for example, the document shown in FIG. 3) is obtained as a search result, and all or a part of the data associated with the tag name corresponding to the “meeting name” in this document is obtained. , Display for displaying the full text or part of the data associated with the tag name corresponding to the "date and time" specified as the search condition, and a fixed number of characters before and after the character string "Tanaka" specified as the search condition Data is displayed. On this display screen, by clicking with a mouse the place where a fixed number of characters before and after the character string "Tanaka" specified as the search condition is displayed, the location of the character string "Tanaka" in the searched document is displayed. The display can be shifted to. FIG. 11 shows an example of the display screen.
[0047]
In the display screen shown in FIG. 11, the character string "Tanaka" is displayed in a special manner such as inversion and highlighting, so that it is easy for the user to visually recognize it. A known public technique may be used to realize this processing.
[0048]
As described above, it is important to have a function of displaying the context before and after the location of the free word given by the user in the retrieved document and a function of quickly shifting to the display of the corresponding location in the retrieved document. This is because a search based on a free word includes a semantic ambiguity unlike a search for structured data using a tag name or the like (XML search), and a search result is not always desired by a user. It is. The user must confirm whether the retrieved document is the desired document. Information useful in making the determination is context information in the document of the specified free word. The above-described KWIC presentation function and the function of quickly shifting to the display of the corresponding location in the retrieved document provide this context information effectively.
[0049]
In the above-described search example, a case where one character string (for example, “Tanaka” in the above example) is given as a free word is shown. However, when there are a plurality of words, the same processing as described above can be performed. it can. In this case, a plurality of character strings such as "Tanaka", "Yamada", and "Sato" may be listed and described in a blank space corresponding to the item "others" on the search condition input screen. For example, one character string may be divided and entered in one blank with spaces, ",", "/", etc. If there are a plurality of blanks, one character string is entered in each blank. May be entered.
[0050]
Next, referring to the flowcharts shown in FIGS. 5 and 6, taking as an example a case where the user has input a search condition different from the above example (for example, see FIG. 9) on the search condition input screen of FIG. The processing operation 2 will be described.
[0051]
For example, on the search condition input screen shown in FIG. 4, the user specifies a character string "December 12" as a search condition in a blank space corresponding to the item "Date" as shown in FIG. At the same time, it is assumed that a character string of “host” or “EIA” is specified as a search condition in a blank corresponding to the item of “others”.
[0052]
The search condition is transmitted from the input unit 1 to the search processing unit 2 (Step S1). As described above, the search processing unit 2 checks the search condition of the user and separates the search condition into the first search condition and the second search condition (step S2). First, a first search condition sentence is generated from the first search condition. In the case of the above example, as described above, the first search condition for searching for a document having an element of the <data> tag and including the character string “December 12” from the first search condition A sentence is generated, and the first search condition sentence is passed to the first search unit 3, requesting a search, and obtaining a search result (step S3).
[0053]
Further, in the case of the above example, the search condition input from the input unit 1 specifies character strings of “host” and “EIA” in blank spaces corresponding to the item of “others”, and the search condition is referred to as “others”. The character string entered in the blank corresponding to the item is used as a search condition (second search condition) for the unstructured data. That is, “host” and “EIA” are obtained as second search conditions from the search conditions input by the user (step S4).
[0054]
By the way, when a word is specified in two fields of the "others" line on the search screen in FIG. 4, it is assumed that an item name comes in the first field and an item value comes in the next field. The item names are generally nouns and common nouns such as “date and time” and “place”. The item value is often a numerical value or a proper noun such as “December 12,” “Friday”, or “Tokyo”, but may be a general noun. The search word “host” in this case is a general noun, and “EIA” is a proper noun.
[0055]
When a common noun or a common noun is included in a search word serving as a search key as always in a free word search, synonym expansion is executed. In this case, the second search is executed after synonymous search words such as "co-host" and "organizer" are added to the character string "host".
[0056]
On the other hand, with respect to proper nouns such as “EIA”, a process of adding a search word called “expansion notation” is performed. Specifically, presence / absence of a long sound symbol "-", a word in which lowercase katakana "tsu" in the search word is replaced with "tsu", an original word of the abbreviated expression, and the like are added. In this case, a second search is performed after a different notation is added to “EIA” such as “Electric Information Association” or “Electronic Information Association”.
[0057]
Normally, general-purpose and specific word dictionaries are used for synonym expansion and variant notation expansion. Such a synonym expansion and a different expression expansion of the search word can be realized by a known technique of information search.
[0058]
In the following description, explanations of synonyms and notation words developed in this manner are omitted, but it is added that the use of these technologies is not excluded.
[0059]
From the second search condition, the unstructured data in the comment area from the documents searched by the first search unit 5 are converted into the character strings “host” and “EIA” given as the second search condition. A second search condition sentence for searching for a containing document is generated. Then, the second search condition sentence is passed to the second search unit 4, a search is requested, and a search result is obtained (step S6 in FIG. 6).
[0060]
In the case of the above example, the second search unit 4 performs a full-text search on the comment area of each document searched by the first search unit 3, and outputs character strings “host” and “EIA”. Search for documents that have a comment area that includes them. Then, the searched document is passed to the search processing unit 2.
[0061]
In the case of the above example, there are two character strings (“host” and “EIA”) as search keys, but a document in which both character strings appear in one comment area is preferentially searched. This is because a plurality of character strings (here, two character strings) specified as search keys are semantically related, and these plurality of character strings may appear close to each other in the document. Because there are many.
[0062]
For example, the document shown in FIG. 3 is detected as a search result because the comment area includes both character strings of “host” and “EIA” (step S7).
[0063]
Next, linguistic processing is performed on the comment area in each searched document (step S9).
[0064]
The two free words "host" and "EIA" have a semantic structure in which "host" is a key and "EIA" is a value from an informational viewpoint, that is, "host" is an item name and "EIA" is its content. have. When a plurality of free words given in this way have a semantic structure, in the language processing of step S9, a publicly-known / public information extraction technique is applied to the comment area in each document searched. In this case, for example, by applying plain2 (plain2 fan [URL] http://shika.aist-nara.ac.jp/products/plane2/plane2-plane2-j.html) or the like, the item name: “host” Item content: Information indicating a correspondence relationship of "Electronic Information Association (EIA)" can be extracted. Plain2 analyzes the structure of a document by using line feed information and indent information of the document. That is, the structure of a bullet is extracted from the text to structure the text.
[0065]
Since the document shown in FIG. 3 includes line feed information and indent information of the original document (see FIG. 2), plain2 can be applied. Even if there is no format information for such a document, similar information extraction can be performed by using a technology disclosed in MUC (Procedings of the Sixth Message Understanding Conference (MUC-6), Morgan Kaufman) or the like. Is possible.
[0066]
The document stored in the document storage unit 5 may be a specific word processing document or a structured document including document layout information such as itemized list or indentation such as RTF (rich text format) or HTML. The same is true. In such a case, in order to directly obtain the layout information of the document, it is possible to acquire the information indicating the correspondence between “item name” and “item content” using a technique such as the MUC.
[0067]
In the second search unit 4, in each document having a comment area including a plurality of character strings specified as free words, the correspondence between the plurality of character strings specified as free words in the comment area is A document that matches the semantic structure between a plurality of free words as search keys is searched with priority.
[0068]
In step S9, a character string specified as the second search condition in each searched document, that is, a portion where a free word (in the above example, “host” or “EIA”) appears, Perform language processing. Here, a fixed number of characters before and after the appearance of the character string are extracted as KWIC (Key Word In Context).
[0069]
Then, display data of the search result is generated using the result of the language processing, and is output to the output unit 6 (step S10). The output unit 6 displays the display data on a predetermined display.
[0070]
FIG. 13 shows an example of a display screen of display data displayed from the output unit 8 and generated from a document obtained as a result of the search performed by the second search unit 4. Here, one document (for example, the document shown in FIG. 3) is obtained as a search result, and all or a part of the data associated with the tag name corresponding to the “meeting name” in this document is obtained. , The full text or a part of the data associated with the tag name corresponding to the “date and time” specified as the search condition, and a fixed number of characters before and after the character strings “sponsored” and “EIA” specified as the search condition are displayed Display data is displayed. In the second search unit 4, when a corresponding relationship such as “host” is an item name and “EIA” is obtained in the item content, a comment area including “host” and “EIA” in the corresponding relationship is obtained. Is searched. Therefore, based on this correspondence, in the display example shown in FIG. 13, the retrieved document is displayed to express that there is a relation that the item name “host” and the item content include “EIA”. ing. That is, in FIG. 13, the character string “host” and the data including the character string “EIA” are called “host” corresponding to the item name so that it is clear that there is a correspondence between the item name and the item content. The character string is displayed in the same column as the item name corresponding to the tag name such as “name” and “date and time”, and a portion where a fixed number of characters before and after the character string “EIA” is displayed as the item content. The location where the item content is displayed is displayed as KWIC, as in Fig. 10. On the display screen in Fig. 13, the portion displayed as KWIC (the portion with a fixed number of characters before and after the character string "EIA" specified as the search condition) Is clicked with a mouse, it is possible to shift to a display of a place where the character string "EIA" appears in the searched document.An example of the display screen is shown in FIG.
[0071]
In the display screen shown in FIG. 14, the character string "host" and a character string portion corresponding to the item content having the item name (a character string including the character string "EIA") are displayed in a special manner such as inversion and highlighting. , Which makes it easy for the user to see. A known public technique may be used to realize this processing.
[0072]
In the above embodiment, description has been made using “<! −−” and “−−>” as marks for marking up the comment area in each document stored in the document storage unit 5. However, the present invention is not limited to this case. For example, an arbitrary XML tag (markup symbol) defined (having the meaning of a comment) defined to represent a comment area may be used. Absent.
[0073]
In the above embodiment, the XML search in the first search unit 3 specifies a tag name and a character string included in an element of the tag name as search conditions, and has an element of the specified tag name. This is a search for structured data (document data having) in which the element includes a specified character string, but is not limited to this. For example, when only one or a plurality of tag names are specified as a search condition, structured data (document data having) having an element of the specified tag name may be searched. Also, for example, when a document structure such as an element including an element with a certain tag name or an element included in a certain element, or a document structure and a character string included in any element are specified as search conditions Can also be easily realized by using a well-known XML search technique. In the above embodiment, the first search unit 3 merely shows an example of performing a search on the structured data using the known XML search technology, and the first search unit 3 can be realized using the known XML search technology. Any search may be performed as long as the search is appropriate. In this case, an input screen or the like for inputting a search condition may be changed according to the search condition to be input.
[0074]
As described above, according to the embodiment, a predetermined element name to be searched in a first search method (for example, an XML search method), which is one of two different search methods, and the element name (Structured data) composed of at least one element consisting of data associated with a second search method (for example, a full text search method) which is another one of the two search methods. ) Is stored in the document storage unit 5 including the data to be searched (unstructured data in the comment area). Then, when the first search condition for the structured data and the second search condition for the unstructured data are input, the first search unit 3 searches the document data in the document data stored in the document storage unit 5. By searching the structured data using the XML search method, document data including an element satisfying the first search condition is obtained. The second search unit 4 searches the first search unit 3 for a result of the search. Document data having unstructured data that satisfies the second search condition is obtained by searching the obtained unstructured data in each document data using a full-text search method.
[0075]
That is, in the above embodiment, the structured data and the unstructured data are mixed, and the structured data is searched using a search method suitable for the structured data search. For structured data, a search is performed using a search method suitable for searching for unstructured data. Generally, the search time for structured data is shorter than the search time for unstructured data. Therefore, first, a search for unstructured data is performed with a result obtained by performing a search for structured data as a search target.
[0076]
As described above, according to the above-described embodiment, before performing a search on unstructured data having a relatively long processing time, a search is performed on the structured data to narrow down documents to be searched, and furthermore, By performing a search on unstructured data after limiting the search position (to the comment area), it is possible to search for document data in which structured data and unstructured data are mixed at high speed and efficiently.
[0077]
Note that, in the above-described embodiment, as an example of structured data, structured data that has been marked up with an XML tag has been described. However, the present invention is not limited to this case. Even if the data is structured by associating data corresponding to the item with (corresponding to the element name in the embodiment), the same effect as described above can be obtained.
[0078]
The method of the present invention described in the embodiment of the present invention (particularly, the processing operation of the search processing unit 2 shown in the flowcharts of FIGS. 5 and 6 and the processing operation of each unit in FIG. 1 in addition to the search processing operation) In addition, as a program that can be executed by a computer, the program can be stored in a recording medium such as a magnetic disk (such as a flexible disk or a hard disk), an optical disk (such as a CD-ROM or a DVD), or a semiconductor memory and distributed.
[0079]
Note that the present invention is not limited to the above-described embodiment, and can be variously modified in an implementation stage without departing from the gist of the invention. Furthermore, the above embodiments include inventions at various stages, and various inventions can be extracted by appropriately combining a plurality of disclosed constituent elements. For example, even if some components are deleted from all the components shown in the embodiments, at least one of the problems described in the column of the problem to be solved by the invention can be solved, and the effect of the invention can be solved. In the case where the effect described in (1) is obtained (at least one), a configuration from which this component is deleted can be extracted as an invention.
[0080]
【The invention's effect】
As described above, according to the present invention, retrieval of document data in which structured data and unstructured data are mixed can be performed at high speed and efficiently.
[Brief description of the drawings]
FIG. 1 is an exemplary diagram showing a configuration example of a document search system according to an embodiment of the present invention.
FIG. 2 is a diagram showing a specific example of an original document that does not include structured data.
FIG. 3 is a specific example of a document including structured data and unstructured data stored in a document storage unit, and shows a document generated from the original document of FIG. 2;
FIG. 4 is a diagram showing an example of a search condition input screen.
FIG. 5 is a flowchart for explaining a processing operation of a search processing unit in FIG. 1;
FIG. 6 is a flowchart for explaining a processing operation of a search processing unit in FIG. 1;
FIG. 7 is a diagram showing an example of a search condition input on a search condition input screen.
8 is a diagram showing a display example of a search result based on the search condition shown in FIG.
FIG. 9 is a diagram showing another example of a search condition input on a search condition input screen.
FIG. 10 is an exemplary view showing a display example of a search result based on the search condition shown in FIG. 9;
FIG. 11 is a diagram showing another display example of a search result.
FIG. 12 is a diagram showing still another example of the search condition input on the search condition input screen.
FIG. 13 is a view showing a display example of a search result based on the search condition shown in FIG.
FIG. 14 is a diagram showing another display example of a search result.
[Explanation of symbols]
REFERENCE SIGNS LIST 1 input unit 2 search processing unit 3 first search unit 4 second search unit 5 document storage unit 6 output unit

Claims (5)

Structured data to be searched in a first search method that is one of two different search methods, and search data in a second search method that is another one of the two search methods. Storage means for storing a plurality of document data including the following unstructured data;
Input means for inputting a first search condition for the structured data and a second search condition for the unstructured data;
First searching for structured data in each document data stored in the storage unit using the first search method to obtain document data including structured data satisfying the first search condition Search means,
By searching the unstructured data in each document data obtained as a result of the search by the first search means using the second search method, an unstructured data satisfying the second search condition is obtained. Second search means for obtaining document data including data,
Output means for outputting document data obtained as a result of the search by the second search means;
A document search device comprising: The first search condition specifies an element name of an element constituting the structured data and a character string included in data associated with the element name,
The second search condition specifies a character string included in the unstructured data,
The first search means obtains document data including an element having an element name and a character string specified as the first search condition;
2. The document search apparatus according to claim 1, wherein the second search unit obtains document data including unstructured data including a character string specified as the second search condition. Structured data to be searched in a first search method that is one of two different search methods, and search data in a second search method that is another one of the two search methods. A search method for searching for desired document data from storage means storing a plurality of document data including the following unstructured data,
Inputting a first search condition for the structured data and a second search condition for the unstructured data;
First searching for structured data in each document data stored in the storage unit using the first search method to obtain document data including structured data satisfying the first search condition Search step,
By searching the unstructured data in each document data obtained as a result of the search in the first search step using the second search method, an unstructured data satisfying the second search condition is obtained. A second search step for document data containing the data;
An output step of outputting document data obtained as a result of the search in the second search step;
A document search method comprising: The first search condition specifies an element name of an element constituting the structured data and a character string included in data associated with the element name,
The second search condition specifies a character string included in the unstructured data,
The first search step obtains document data including an element having an element name and a character string designated as the first search condition;
4. The document search method according to claim 3, wherein the second search step obtains document data including structured data including a character string specified as the second search condition. Structured data to be searched in a first search method that is one of two different search methods, and search data in a second search method that is another one of the two search methods. A program for retrieving desired document data from storage means storing a plurality of document data including unstructured data,
On the computer,
Inputting a first search condition for the structured data and a second search condition for the unstructured data;
First searching for structured data in each document data stored in the storage unit using the first search method to obtain document data including structured data satisfying the first search condition Search step,
By searching the unstructured data in each document data obtained as a result of the search in the first search step using the second search method, an unstructured data satisfying the second search condition is obtained. A second search step for finding document data with data;
An output step of outputting document data obtained as a result of the search in the second search step;
The program to be executed.

Images