JP2009230296A - Document retrieval system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To acquire a document adapted to a retrieval key and a document relevant to the adapted document in the case of retrieving the document. <P>SOLUTION: A document retrieval server 1 is provided with: a document group generation part 11a for generating a document group by gathering quotation origin documents quoting the same sections of quotation destination documents based on inter-document quotation relation; a document group retrieval index generation part 11b for generating a retrieval index by extracting feature words from the generated document group; a document group retrieval part 11c for retrieving a document group retrieval index; and a document retrieval part 11h for retrieving the document retrieval index for retrieving the document included in the retrieval target document group, and configured to transmit the acquired document group retrieval result 13b and a document retrieval result 13d to a document retrieval client 2. A document retrieval client displays the received document group retrieval result 13b and the document retrieval result 13d in association at a retrieval result display part 11g. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a document search system, and more particularly to a system that generates a document group from a set of documents to be searched and generates an index for search.

  In document retrieval, it is important that a user can find a desired document efficiently without omission. In contrast, Elsevier's bibliographic information search service “SCOPUS” described in Non-Patent Document 1 provides a function of displaying a list of documents that have a citation relationship with a hit document. With this function, the user can easily obtain a document related to the hit document, so that search omission can be reduced. Similarly, the ACM (Association for Computing Machinery) electronic library service provided on the site described in Non-Patent Document 2 has a function of displaying a link to a document having a citation relationship.

  The document search device described in Patent Document 1 has a function of classifying documents included in search results based on citation relationships. With this function, the user can refer to related documents in the search results collectively, so that the search results can be checked efficiently.

Yasushi Adachi. “Academic Information Navigation Service: Scopus”. Information Management. Vol. 47, No. 8, (2004), 558-562. ACM Portal http://portal.acm.org/portal JP 2007-328714 A

  Related documents displayed by the techniques described in Non-Patent Documents 1 and 2 are not always compatible with the search key. Such a document may be noisy for a user searching for a desired document with a search key.

  Documents classified by the technique described in Patent Document 1 are limited to documents that hit the search key. A document that has a citation relationship with a document is not necessarily cited from the same viewpoint. Therefore, there is a possibility that various viewpoints may be mixed in the document group in which the documents are classified based on the citation relationship, and the reference may be complicated.

  The document search system of the present invention includes a document group generation unit that generates a document group based on a citation relationship between documents from a set of documents to be searched, and a document group search by extracting feature words from the document having the citation relationship. Document group search index generation means for generating an index. Here, the document group generation means collects documents that are the same in the cited document, with a document cited from another document as a cited document, a document that cites another document as a cited document, and the like. Generate a group of documents.

  According to the present invention, it is possible to efficiently refer to a document that matches a search key and a related document.

Embodiments of the present invention will be described below with reference to the drawings.
The document search apparatus of the present invention generates a document group based on a citation relationship between documents from a set of documents to be searched, extracts a feature word from a document having a citation relationship, and generates a search index for the document group.

  FIG. 1 is a diagram showing an outline of a procedure for generating a search index in document search, FIG. 1 (a) shows an overview of a procedure for generating a search index in a general document search method, and FIG. An outline of a search index generation procedure in FIG. As shown in FIG. 1A, in a general document search, a search index is generated for each document included in a search target document set. On the other hand, in the present invention, as shown in FIG. 1B, a document group is generated from a set of documents to be searched based on the citation relationship between documents, and a search index is determined by regarding one document group as one document. Generate.

  Here, the citation relationship between documents in the present invention will be described in detail with reference to FIGS.

  FIG. 2 is a diagram showing an example of a document having a citation relationship, and shows a citation relationship between papers. In the figure, the part surrounded by a dotted line in document 1, ie, “long term according to document [1]” refers to another paper. A document that cites another document in this way is hereinafter referred to as a “citation document”. On the other hand, in FIG. 2, document 2 is a document cited as “document [1]” at the citation source location. Such a document cited from another document is hereinafter referred to as a “cited document”. In the figure, the citation source part of the document 1 and the citation destination part of the document 2 will be described later. As in the example of the paper shown in FIG. 2, the relationship between documents defined by the document author is hereinafter referred to as “author citation”.

  On the other hand, there is a case where a document other than the document author defines the relationship between documents. FIG. 3 is a diagram showing an example of a relationship between documents defined by a person other than the document author, and specifically shows an examination reference for a patent application. In the figure, document 3 is a notice of reasons for refusal of a patent application, document 4 is a patent document to be examined as described in the notice of reasons for refusal of document 3, and document 5 is cited as a reason for refusing the patent of document 4. It is a literature that has been. In the notification of reasons for refusal of document 3, the part “claim 1” surrounded by a dotted line is the part surrounded by the dotted line of document 4, and “described in [0027] of cited document 1” is the dotted line of document 5. The document 5 indicates that the document 5 is related to the document 4. In the present invention, even when a document other than the author of such a document is associated with the document, it is treated as a citation relationship, and a certain document (document 4 in the figure, patent to be examined) is referred to as document 5 in the figure. When the document) is defined as related, the former is referred to as “cited document” and the latter is referred to as “cited document”.

  In the document group generation process, a document group is generated by collecting the document groups having the citation relationship described above, and a search index is generated. In the generation of the document group shown in FIG. 1B, when there is a document that does not have a citation relationship with any document in the search target document set, a document group composed of one document is generated. There is a way.

  As described above, the citation relationship is defined by the document author or the document referenceer as the relationship between documents, and has high reliability. In the present invention, a document group is generated based on such a highly reliable relationship, and feature words are extracted. In general, in generating a search index for a document, it is known that the wider the range in which a certain document index is generated, that is, the range of character strings from which feature words are extracted, the smaller the omission of search. From this, it can be expected that if a document group is generated by grouping documents having a citation relationship with respect to a certain document, and a search index is generated by regarding the document group as one document, omission of search for a certain document is reduced. Furthermore, the user can refer to related documents efficiently.

  In the document group generation process, documents having the same cited document are grouped to generate a document group. Here, generation of a document group based on the citation relationship with the citation destination document as the center will be described.

  Some documents are often cited from other documents. Such a document has characteristics such as being highly evaluated in the field to which the document belongs, or having important matters in the field described therein. When collecting documents (quoting documents) that cite a certain document (cited document) in the document group generation process, it is highly evaluated or important in certain fields as described above There is a high possibility that a document group including a document including the description and related documents is generated. The present invention focuses on this point, and generates a document group by collecting documents having the same cited document. As a result, the user can efficiently search for documents that are easily cited in a certain field.

  The document group generation process generates a document group by combining the cited part acquisition process for acquiring a cited part cited from another document in the cited document and the document having the same cited target part. Document summarization processing may be included. Here, in the present invention, a part in which the cited document is cited in the cited document is referred to as a “cited part”, and a part cited from the cited document in the cited document is referred to as a “cited part”. This will be described using an example.

  In FIG. 2 shown above, the part surrounded by the dotted line of the document 1 quotes the document 2 that is the cited document. Such a part is called a “quoting source part”. In the figure, the citation source part of document 1 cites document 2 regarding improvement of symptoms by administration of the drug. On the other hand, the part surrounded by the dotted line in Document 2 describes the improvement of the symptoms by administration of the drug, and corresponds to the previous quotation source part. A portion having a description quoted from another document in this way is referred to as a “quotation place”.

  In FIG. 3 showing an example of the citation relationship of patent documents, in the document 3, that is, the notice of reasons for refusal, [0027] of the cited document is cited as the reason for refusing “claim 1” of the patent to be examined. ing. Even in such a citation relationship based on definitions other than the author of the document, portions corresponding to the citation source document and the citation destination document are referred to as a “citation source location” and a “citation destination location”, respectively. Specifically, in FIG. 3, claim 1 of the patent to be examined, that is, the portion surrounded by the dotted line of the document 4 is surrounded by the dotted line of the cited reference [0027], that is, the document 5 of the cited document. The part is “cited place”.

  In the generation of a document group based on the citation relationship shown in FIG. 1, in the document group generation process considering the citation source location and the citation destination location, the citation destination document is the same and the citation corresponding to the citation destination location in the document is included. A group of documents is generated by collecting citation documents including the original part.

Here, the citation source part of the citation source document and the citation destination part of the citation target document corresponding to the citation source document are manually extracted from the search target document set in advance, or automatically extracted by analyzing the citation-related documents. You can do it. As an example, when a document is registered in a database, a part that is cited by the author in the document is registered at the same time, or the expression for quoting the document as in the notice of reasons for refusal shown in FIG. [Xxxx]
Or the like, a method of extracting a citation source location and a citation destination location using the pattern is conceivable.

  In general, a document includes a plurality of topics. Therefore, a plurality of documents (quoting documents) that cite the same document (cited document) do not always cite the same topic. Each may cite a different topic in the cited document. In the document group generation process considering the citation source part and the citation destination part, the citation source documents having the same citation destination part in the citation target document are collectively generated, so that the generated document group has the same topic. It is more likely to be composed of cited documents. As a result, the user can efficiently find a desired document.

  The document group generation process includes a citation document division process for dividing the citation document and a document allocation process for generating a document group by allocating the citation document to the divided citation document part. Also good.

  As described above, it is conceivable to extract the cited part manually or automatically. However, it may be difficult to extract the quoted part because it is not explicitly described in the cited source document as compared to the cited part. In such a case, in the present invention, the document group is generated by dividing the cited document and assigning the cited document.

  This will be described with reference to FIG. First, citation documents that cite the same citation documents are collected from a search target document set to generate a temporary document group ((i) in the figure). Next, the following processing is repeated for each temporary document group. First, the cited document is divided. The dividing means will be described later ((ii) in the figure). Next, a citation source document included in the same temporary document group is allocated to the divided citation target portion ((iii) in the figure). Specifically, the following processing is repeated for the citation source document included in the same temporary document group as the citation destination document. That is, the similarity with the citation source document is calculated for each portion of the divided citation destination document, and the citation source document is assigned to the portion with the highest similarity. Finally, a document group composed of the cited document and the cited document assigned to the same part of the cited document is generated ((iv) in the figure). Note that, among the divided citation target parts, a part to which no citation document is assigned is not included in any document group. By this method, even when the citation destination part of the citation destination document is unknown, it is possible to generate a document group including the citation destination document portion and the citation source document at low cost.

  An example of the cited document dividing method is a method of dividing a document according to the document structure of the cited document. Specifically, the document is divided into chapters, sections, and paragraphs using a tag indicating the document structure attached to the document or a document layout as a clue. Alternatively, in the case where the cited document is a patent document, a method may be considered in which “claims” are to be divided and divided into each claim to be a part of the cited document. This method makes it possible to easily divide the cited document. In addition, the part of the cited document divided by this method has a part (document structure such as chapter, section, paragraph, etc.) that matches the original cited document. There is also an advantage that the structural relationship is easy to grasp.

  Another example of the cited document dividing method is a cited document term clustering method in which feature words of the cited document are clustered to divide the document. Specifically, feature words are extracted from the cited document, term clustering technology is applied to them, and the generated term cluster is regarded as a part of the cited document. This method makes it possible to easily divide the cited document. In addition, since the portion of the cited document divided by this method represents a topic included in the cited document, there is also an advantage that the user can easily understand the content relationship with the original cited document.

  The document allocation process includes a citation source document clustering process for clustering a plurality of citation documents that cite the same document, and a citation source document cluster is added to a citation source document portion divided by the citation destination document division process. Citation source document cluster assignment processing for generating a document group by assignment may be included.

  In the assignment of the cited document to the cited document part shown in (iii) of FIG. 4, the cited document may be assigned one by one as described above, or the cited document is classified by the document clustering technique. After that, it may be assigned for each cluster. Here, since the citation documents included in the cluster have a semantic relationship with each other, there is a high possibility that the generated document group includes semantically related documents.

  In the document group search index generation process, a feature word is extracted from the cited part of the cited document obtained by the cited part obtaining unit and the part of the citation source document that cites the obtained cited part. The search index may be generated. In addition, the extraction of feature words, in detail, extract words from the quotation destination location and the quotation source location, and then merge the extracted words, or first merge the character strings of the quotation destination location and the quotation source location This can be achieved by generating a temporary document and extracting words from the document. In the former case, the words can be weighted based on the appearance frequency of the words in each document. In the latter case, it is possible to reflect the appearance frequency of the word as viewed in the entire document group in the weight of the word.

  Thus, by extracting feature words from related parts in a document having a citation relationship, a search index that well represents the characteristics of a document group can be generated. As a result, the user can efficiently acquire a document group that matches the search key.

  Here, when extracting feature words from the cited part of the cited document and the cited document part, the feature word is extracted by assigning a higher weight to the word appearing in the cited part, and a search index for the document group is generated. You may make it do.

  All the citation documents included in the document group are related to the citation location of the citation document. For this reason, the citation destination part of the citation target document is positioned as the center of the document included in the document group. At this time, if a high weight is given to the word appearing at the quotation destination position positioned as the center of the document included in the document group, a search index that well represents the characteristics of the document group can be generated.

  As described above, the document group search index generation process extracts feature words from the cited part of the cited document obtained by the cited part obtaining unit and the part of the cited document that cites the obtained cited part. Instead of generating a search index for a group of documents, a feature word is extracted from the cited document part divided by the document dividing process and the document assigned to the cited document part by the document assigning process. A group search index may be generated.

  Also with this method, since feature words are extracted from related portions in a document having a citation relationship, a search index that well represents the characteristics of a document group can be generated. As a result, the user can efficiently acquire a document group that matches the search key.

  When extracting feature words from the cited document part and the document assigned to the cited document part, the feature words are extracted by assigning higher weights to the words appearing in the cited document part. A search index may be generated. In the document group generated by the document group generation process, the cited document part is positioned as the center of the relationship between the cited document, while the cited part is a part of the document, so the word extracted therefrom The number is expected to be small compared to the original document. On the other hand, in this method, a high weight is given to the word appearing at the quotation destination location. Thus, by assigning a high weight to the word appearing at the quotation destination location, a search index that well represents the characteristics of the document group can be generated.

  The document search apparatus according to the present invention includes a document group search unit that searches the document group search index generated by the document group search index generation unit, and a document group search result display that displays the document group search result acquired by the document group search unit. The document group search result display means displays a list of the cited document and the cited source document in the display of the document group, and further displays the cited destination position of the cited document in an easily identifiable state. In the citation source document, the word appearing at the citation location is highlighted.

  As described above, the citation destination of the citation target document is at the central position of the documents included in the document group. From the above document group search result display, the user can easily refer to the citation destination part, which is a part of the citation target document, so that the user can efficiently grasp the characteristics of the document group. Further, since the word appearing at the citation destination location is highlighted in the citation source document, the user can easily grasp the relationship between the citation source document and the citation location. Here, the citation destination part of the citation target document refers to a term cluster having a part of the citation target document or a word appearing in the citation target document as an element.

  The document search apparatus of the present invention also extracts a document group search means for searching the document group search index generated by the document group search index generation means, and extracts feature words from each document included in the document set to be searched. Document search means for searching the generated document search index, document search result display means for displaying the document group search result acquired by the document search means and the document search result acquired by the document search means in comparison. In this case, the user can refer to a plurality of search results for the same set of documents to be searched, so that a desired document can be found efficiently.

  The document search result display unit may display the citation source documents included in the document group search result acquired by the document group search unit by rearranging them according to the rank in the document search result acquired by the document search unit. The document group includes one or more citation documents. By displaying these citation documents in the order of suitability with the input search key, it is possible to easily refer to documents that are suitable for input as a document group and also suitable as documents.

  The document search result display unit may display the rank in the document group search result acquired by the document group search unit in displaying the document search result acquired by the document search unit. In this case, the user can easily compare the suitability rank based on the document group search index and the suitability rank based on the document search index. As a result, a document with a high rank as a document and a document group with a high rank is highly likely to be a desired document, or a document with a low rank but a document group with a high rank is referred to once. It is useful for reference omission and efficiency improvement.

Hereinafter, the present invention will be described in detail by way of examples.
[Example 1]
5 shows the overall configuration of the system according to the first embodiment of the present invention, and FIG. 6 shows the hardware configuration of a computer constituting the system according to the first embodiment of the present invention. . Next, the module configuration and details of each component of the document search system will be described with reference to FIGS. 7 to 12, and the processing of the document search system according to the first embodiment of the present invention will be described with reference to FIGS. The procedure is described. Finally, an example of a search result display screen of the document search system according to the first embodiment of the present invention is shown using FIG. 16 and FIG.

(Overall system configuration)
First, the overall configuration of the system according to the first embodiment of the present invention will be described with reference to FIG. As shown in FIG. 5, the document search system includes a document search server 1, a document search client 2, and a communication line 3. The document search server 1 is a computer that stores a search target document set and a search index, and searches the search index. The document search client 2 is a computer used by a user who searches for a document, and is used to search for a document stored in the document search server 1. The communication line 3 is a transmission path for transferring data between computers, and connects the document search server 1 and the document search client 2, whether wired or wireless. Specifically, it refers to a LAN or the Internet. In FIG. 5, the document search server 1 has a single configuration, but the present invention is not limited to this. The document search server 1 may be composed of a plurality of computers.

(Hardware configuration)
Next, the hardware configuration of the document search server 1 and the document search client 2 according to the first embodiment of the present invention will be described with reference to FIG. Since all of these have the same configuration, description will be made here with reference to the hardware configuration diagram of the document search server 1 shown in FIG.

  As shown in FIG. 6, the document search server 1 includes a CPU 10, a hard disk 20, a memory 30, a display 40, a display control unit 41, a keyboard 50, a keyboard control unit 51, a mouse 60, a mouse control unit 61, and a bus 70. Is done. The CPU 10 executes data input / output, reading, storage, and various processes. The hard disk 20 is a device that stores data, and the memory 30 is a device that loads and stores programs and data. The display 40 is a device that displays data to the user, and is controlled by the display control unit 41. The keyboard 50 and the mouse 60 are devices that receive input from the user, and are controlled by the keyboard control unit 51 and the mouse control unit 61, respectively. The bus 70 passes data between the components.

(Module configuration)
Next, the module configuration of the document search system according to the first embodiment of the present invention will be described with reference to FIGS. FIG. 7 is a diagram showing a module configuration of the document search system according to the first embodiment of the present invention. As shown in FIG. 7, the document search system includes the following three types of modules. That is, a processing unit that executes various processes (indicated by a rectangle in the screen; hereinafter, the same applies to the description of other embodiments), a data storage unit that stores data (drum type; the same applies hereinafter), and a temporary Is a temporary data portion (tape type; the same applies hereinafter) for storing data generated in Hereinafter, each module shown in FIG. 7 will be described.

(Module configuration of document search system: processing section)
As shown in FIG. 7, the processing unit of the document search system includes a document group generation unit 11a, a document group search index generation unit 11b, a document group search unit 11c, a server communication unit 11d, and a document search client 2 in the document search server 1. Client communication unit 11e, search key receiving unit 11f, and search result display unit 11g.

  The document group generation unit 11a in the document search server 1 generates a document group based on the citation relationship between documents from the document stored in the document storage unit 12a, and stores the document group in the document group storage unit 12b. The document group generation unit 11a further includes a citation document division unit 11a1 and a document allocation unit 11a2. The cited document dividing unit 11a includes a cited document structure dividing unit 11a11, and divides a cited document included in a set of documents having a citation relationship based on the document structure. The document assigning unit 11a2 assigns the citation source document to the divided citation destination part of the set of documents having the citation relationship. The document group search index generation unit 11b generates a search index for the document group stored in the document group storage unit 12b and stores it in the document group search index storage unit 12c. The document group search unit 11c searches the document group search index stored in the document group search index storage unit 12c based on the search key stored in the search key data 13a, and stores the result in the document group search result data 13b. . The server communication unit 11 d receives the search key data 13 a transmitted from the document search client 2 and transmits the document group search result data 13 b to the document search client 2.

  The client communication unit 11e in the document search client 2 transmits the search key data 13a to the document search server 1 and receives the document group search result data 13b transmitted from the document search server 1. The search key receiving unit 11f receives a search key from the user and stores it in the search key data 13a. The search result display unit 11g displays the document group search result stored in the document group search result stored in the document group search result data 13b to the user.

  Among the above processing units, the document group search unit 11c, the server communication unit 11d, the client communication unit 11e, and the search key reception unit 11f can be implemented by applying a general document search technology, and thus detailed description thereof is omitted. . Details of the document group generation unit 11a, the document group search index generation unit 11b, and the search result display unit 11g will be described later with reference to flowcharts or screen examples.

(Module configuration of document retrieval system: data storage unit)
Next, a data storage unit indicated by a drum shape in the figure will be described. As shown in FIG. 7, there are three data storage units in the document search server 1 of the document search system: a document storage unit 12a, a document group storage unit 12b, and a document group search index storage unit 12c.

The document storage unit 12a is an area for storing documents to be searched, the document group storage unit 12b is an area for storing document groups generated from the documents stored in the document storage unit 12a, and the document group search index storage unit 12c is a document group. This is an area for storing an index for searching a document group stored in the storage unit 12b.
Here, a detailed configuration of each data storage unit will be described with reference to FIGS.

(Configuration of Document Storage Unit 12a)
FIG. 8 is a diagram showing the configuration of the document storage unit 12a. As shown in FIG. 8, the document storage unit 12a includes a document number 12a01, a text 12a02, and a cited document number 12a03. The document number 12a01 stores a number for identifying a document. The text 12a02 stores a character string of the document. The cited document number 12a03 stores the document number of the document cited by the document. More specifically, in a document such as a paper, a cited document is identified based on bibliographic information of a document described as a reference, and a document number 12a01 corresponding to the document is acquired and stored. The cited document can be identified using a known technique described in “Computer Software, Vol. 13, No. 1, pp. 35-39, (1997)”.

  The first line in FIG. 8 stores a text “<doc> <title> Example of portal pulmonary hypertension </ title> ...” with a tag of a document whose document number is “doc050789”. Further, “doc013024”, “doc027863”, and the like are stored as document numbers cited in the document.

(Configuration of Document Group Storage Unit 12b)
FIG. 9 is a diagram showing a configuration of the document group storage unit 12b. As shown in FIG. 9, the document group storage unit 12b includes a document group number 12b01, a citation destination document number 12b02, a citation destination location 12b03, and a citation source document number 12b04. The document group number 12b01 stores a number for identifying the generated document group. The cited document number 12b02 stores the number of the cited document in the generated document group. Specifically, the cited document number 12b02 stores the value of the document number 12a01 in the document storage unit 12a in which the cited document is stored. The citation location 12b03 stores a character string of a portion quoted from a later-described citation source document in the citation destination document. The citation source document number 12b04 stores the number of the document quoting the citation destination location. Specifically, the citation source document number 12b04 stores the document number 12a01 in the document storage unit 12a in which the document quoting the cited part is stored.

  In the first line in FIG. 9, a document group whose document group number 12b01 is “cl048901” is stored. The citation destination document number 12b02 of the document group is “doc050789”, and the citation destination location 12b03 of the citation destination document is “<p> case of examination ... </ p>”. Further, the citation source document number 12b04 quoting the citation location is “doc070041” and “doc068344”.

  In the figure, the third, fourth, and fifth lines store document groups whose document group numbers 12b01 are “cl048903”, “cl048904”, and “cl048905”, respectively. The cited document number 12b02 of these document groups is “doc050791”, which is the same. On the other hand, the citation destinations 12b03 of these three document groups are different from “<p> Introduction ...”, “<p> Research ...”, and “<p> Summary ...”, respectively. This indicates that one citation destination document includes a plurality of citation destination portions, and each citation source document is assigned separately to generate a document group.

(Configuration of Document Group Search Index Storage Unit 12c)
FIG. 10 is a diagram showing the configuration of the document group search index storage unit 12c. As shown in FIG. 10, the document group search index storage unit 12c includes a document group number 12c01 and a feature word 12c02. The document group number 12c01 stores the number of the document group corresponding to the index. Here, the document group number refers to the document group number stored in the document group storage unit 12b. In the feature word 12c02, the feature word extracted from the citation destination part of the citation destination document constituting the document group and the citation source document quoting the citation destination are stored. Extraction of feature words from a document group is performed by a known technique for extracting feature words from documents described in “Information Processing Society Journal, Vol.38, No.2, pp.299-309, (1997)”. It can be implemented by applying. In FIG. 10, only the character string of the feature word is described in the feature word 12c02, but this is not restrictive. You may store with the value etc. which show the appearance frequency or importance of the feature word in a document group.

  The first row in FIG. 10 stores a search index of a document group whose document group number 12c01 is “cl048901”. Specifically, “high blood pressure”, “diagnosis”, “pulmonary function”, and the like are stored in the feature word 12c02 as feature words of the document group.

(Module structure of document search system: temporary data section)
Next, the temporary data portion shown in the form of a tape in the figure will be described. As shown in FIG. 7, there are three temporary data portions of the document search system: search key data 13a, document group search result data 13b, and document body data 13c.

  The search key data 13 a stores a search key for searching for a document input by the user using the document search client 2. The document group search result data 13b stores the result of the document group search unit 11c searching the document group search index stored in the document group search index storage unit 12c based on the search key data 13a. Specifically, a list of document group numbers that match the search key data 13a is stored together with the matching degree with the search key. The document text data 13c stores the text of the document specified by the user in the document group included in the search result.

  Here, an example of the search key data 13a and the document group search result data 13b among the temporary data portions will be described with reference to FIGS.

(Example of search key data 13a)
FIG. 11 is a diagram illustrating an example of the search key data 13a. As search keys input by the user, “hypertension”, “pulmonary function”, and “blood vessel” are stored. In FIG. 11, only the word is shown as the search key, but a numerical value indicating the weight may be given to each word.

(Example of document group search result data 13b)
FIG. 12 is a diagram illustrating an example of the document group search result data 13b. In the figure, “cl050989”, “cl029845”, and the like are stored as a list of document group numbers that match the search key. In FIG. 12, the numerical value “9.328” indicating the degree of matching of the document group with the search key is stored together with the document group number. However, the present invention is not limited to this. The order suitable for the search key may be used.

(Processing procedure of document retrieval system)
FIG. 13 is a flowchart showing the processing procedure of the document search system in the first embodiment of the present invention.

  As shown in FIG. 13, the document search system first generates a document group from a search target document stored in the document storage unit 12a in the document search server 1 and stores it in the document group storage unit 12b (101). A feature word is extracted from the stored document group, a search index for the document group is generated, and stored in the document group search index storage unit 12c (102). If the user instructs the document search client 2 to end the document search, the process ends (103). Otherwise, the input of the search key is accepted, stored in the search key data 13a, and transmitted to the document search server 1 ( 104). The document search server 1 receives the search key data 13a transmitted from the document search client 2, and searches for the document group stored in the document group search index storage unit 12c based on the search key stored in the search key data 13a. The index is searched and the result is stored in the document group search result data 13b and transmitted to the document search client 2 (105). The document search client 2 receives the document group search result data 13b transmitted from the document search server 1, and displays the stored document group together with the rank of the matching level. Note that the processing of step 101 and step 102 need only be executed once by the document search server 1 and does not need to be executed each time a search is performed.

  Here, the processing of the document group generation step 101 and the document group search index generation step 102 in the above processing will be described with reference to detailed flowcharts shown in FIGS.

(Detailed processing procedure of document group generation step 101)
FIG. 14 is a flowchart showing a detailed processing procedure of the document group generation step 101. As shown in FIG. 14, the document group generation step 101 initializes a number for identifying a document group (10101), sets 0 to a variable i indicating an index of the document group storage unit 12b, and indicates an index of the document storage unit 12a. The variable j is set to 1 (10102), and the jth document in the document storage unit 12a is acquired (10103). Next, the document storage unit 12a is referred to, and the document in which the value of the jth document number 12a01 is stored in the citation destination document number 12a03 is acquired as a citation source document (10104). If the number of acquired citation documents, that is, the document with the jth document as a citation destination is larger than 0, the process proceeds to step 10106, and otherwise, the process proceeds to step 10112 (10105). In step 10106, a document group number is added, and the text stored in the jth text 12a02 of the document storage unit 12a is divided for each document structure using a tag (10107).

  Next, 1 is set to a variable k indicating the number of citation source documents (10108), and the kth document among the citation source documents acquired in step 10104 is assigned to the citation destination document divided in step 10107 (10109). . More specifically, when the cited document is divided into x parts, the degree of similarity between the kth source document and the x parts is calculated, and the degree of similarity is the highest among the x parts. Assign the kth source document to the part. The similarity between the citation source document and the divided document portion can be realized by applying a known technique for obtaining the similarity between documents, such as a vector space method. Next, k is incremented by 1 (10110), and if k becomes larger than the number of the citation source documents acquired in step 10104, the process proceeds to step 10112, and otherwise, the process returns to step 10109 (10111). In step 10112, j is incremented by 1 (10112), and if j becomes larger than the number of documents stored in the document storage unit 12a, the process returns, otherwise returns to step 10103 (10113).

(Detailed processing procedure of document group search index generation step 102)
FIG. 15 is a flowchart showing a detailed processing procedure of the document group search index generation step 102. As shown in FIG. 15, the document group search index generation step 102 sets 0 to a variable m indicating an index of the document group search index storage unit 12c and a variable n indicating an index of the document group storage unit 12b (10201). , N is incremented by 1 (10202). If n is larger than the number of document groups stored in the document group storage unit b, the process returns. Otherwise, the process proceeds to step 10204 (10203). In step 10204, m is incremented by 1, and the document group number stored in the nth document group number 12b01 of the document group storage unit 12b is stored in the mth document group number 12c01 of the document search index storage unit 12c (10205). The feature word is extracted from the character string stored in the nth citation destination location 12b03 of the document group storage unit 12b (10206), and the nth citation source document number 12b04 of the document group storage unit 12b is referred to. The citation source document is acquired from the document storage unit 12a and each feature word is extracted (10207), and the extracted feature word is stored in the mth feature word 12c02 of the document group search index storage unit 12c (10208). Return to 10202.

  The detailed processing procedure of the document group generation step 101 and the document group search index generation step 102 of the document search system according to the first embodiment of the present invention has been described above using the flowchart.

  Next, the process of the search result display step 106 among the above processes will be described using the search result display screen shown in FIGS.

(Example of search result display screen in document search system)
FIG. 16 is a diagram showing an example of a document group search result display screen in the document search client 2 of the document search system according to the first embodiment of the present invention. As shown in the figure, the document search result display screen of the document search client 2 includes a search key input area 401, a search button 402, a document group search result display area 403, and a document number display area 404.

  A search key input area 401 is an area where the user inputs a search key. In the example of FIG. 16, a state in which “hypertension”, “pulmonary function”, and “blood vessel” are input as search keys is displayed. A search button 402 is a button for the user to instruct execution of the search. The document group search result display area 403 is an area where the document search client 2 displays the document group search result data 13b in descending order of the degree of matching with the search key. The document number display area 404 is an area for displaying the number of documents included in the document group displayed in the document group search result display area 403, the number of checked document groups, and the like.

  Here, the document group search result display area 403 and the document number display area 404 will be described in detail.

(Detailed description of the document group search result display area 403)
The document group search result display area 403 includes a document group search result rank 40301, a citation destination document 40302, a citation destination portion 40302, and a citation source document 40304. The document group search result rank 40301 displays the rank of the degree of matching between the document group and the search key. The cited document 40302 displays a cited document in the document group. Specifically, the title, author name, or part of the document is displayed. The citation destination portion 40303 displays a portion allocated from the citation source document in the document group in the citation destination document. The citation source document 40304 displays a citation source document that cites the same portion of the citation destination document in the document group. Specifically, the title or author name of the citation source document or a part of the document is displayed.

  In FIG. 16, the cited document 40302 of the document group having the second highest matching degree with the search key displays the title of the cited document “high blood pressure on lung function ...”. As the citation document 40304 quoting the same citation, the titles of the citation documents “pulmonary ventilation disorder and hypertension ...” and “chronic lung disease patient…” are displayed.

(Detailed description of the document count display area 404)
The document count display area 404 includes a display count 40401, a reference estimate 40402, and a reference 40403.

  The documents included in the document group displayed in the document group search result display area 403 are different for each document group. Therefore, the display number 40401 displays the number of documents included in the document group displayed in the document group search result display area 403. In the example of the figure, the display number 40401 includes “20” document groups included in the search result, and the number of documents (cited document and cited document) included in the 20 document groups is as follows. It is shown that there are “42” cases in total. More specifically, when the number of document groups displayed in the document group search result display area 403 is N, and the number of documents included in the i-th document group is Docnum (i), the number of documents M included in the document group. Is M = Docnum (1) + Docnum (2) + Docnum (3) +... + Docnum (N).

  The reference estimate 40402 displays how many documents in the document text search result display area 403 should be referred to in order to refer to x documents. Specifically, in the reference estimate 40402, the user can input either the number of document groups to be referenced or the number of documents included. In FIG. 16, when the user inputs “20” as the numerical value of “the number of documents included”, the system sets the numerical value that the top “12” documents should be referred to in order to refer to a total of 20 documents. it's shown.

  Referenced 40403 displays the number of document groups referred to by the user in the document group search result and the number of documents included in the document group. In the example of the figure, the reference 40403 indicates that the number of document groups referred to by the user in the document group search result is “3” and the number of documents included in the referenced document group is “10”. Has been.

(Example of document group display screen in the document search system)
FIG. 17 is a diagram showing an example of a document group search result display screen in the document search client 2 of the document search system according to the first embodiment of the present invention. In the document group search result display screen shown in FIG. It is an example of the screen which displays the selected document group.

  As shown in the figure, the document group display screen of the document search client 2 includes a citation location 501, a citation destination document 502, and a citation source document 503. The citation location 501 is a portion of the citation destination document constituting the document group, and displays a portion to which the citation source document included in the document group is assigned. The cited document 502 displays a cited document including a cited part. The citation source document 503 displays the citation source document assigned to the citation destination location.

In FIG. 17, “2. Object and method at this hospital for one year this year” is displayed in the reference place 501. In the cited document 502, the document “high blood pressure affecting pulmonary function ...” including the cited location is displayed, and the portion corresponding to the cited location 501 is highlighted in the document. In the citation document 503, a plurality of documents quoting the citation location 501 are displayed, and the words appearing in the citation location 501 such as “high blood pressure” and “elderly” are highlighted in the document. ing.
The document search system according to the first embodiment of the present invention has been described above.

[Example 2]
Next, a document search system according to a second embodiment of the present invention will be described with reference to the drawings. Since the overall configuration of the system and the hardware configuration of the computer constituting the system are the same as those of the first embodiment described above, description thereof will be omitted.

(Module configuration)
FIG. 18 is a diagram showing the module configuration of the document search system according to the second embodiment of the present invention. Hereinafter, each module shown in FIG. 18 will be described.
(Module configuration of document search system: processing section)
The processing unit of the document search system includes a document group generation unit 11a in the document search server 1, a document group search index generation unit 11b, a document group search unit 11c, a server communication unit 11d, and a client communication unit 11e in the document search client 2. There is a search key receiving part 11f and a search result display part 11g.

  The document group generation unit 11a in the document search server 1 generates a document group based on the citation relationship between documents from the document stored in the document storage unit 12a, and stores the document group in the document group storage unit 12b. The document group generation unit 11a further includes a cited document division unit 11a1 and a document allocation unit 11a2. The cited document dividing unit 11a includes a cited document term clustering unit 11a12, and divides a cited document included in a set of documents having a citation relationship into a plurality of contents by clustering words included in the document. . The document allocation unit 11a2 includes a citation source document clustering unit 11a21 and a citation destination document cluster allocation unit 11a22. The citation source document clustering unit 11a21 clusters the citation source documents among a set of documents having a citation relationship, and generates a document cluster. The citation source document cluster assignment unit 11a22 assigns the generated citation source document cluster to the quoting destination document portion.

  Among the processing units shown in FIG. 18, processing units other than the document group generation unit 11a, that is, a document group search index generation unit 11b, a document group search unit 11c, a server communication unit 11d, a client communication unit 11e, a search key reception unit 11f, and Since the search result display unit 11g is the same as that of the first embodiment described above, the description thereof is omitted.

(Module configuration of document retrieval system: data storage unit)
Next, a data storage unit shown in the form of a drum in the figure is shown. As shown in FIG. 18, the data storage unit includes a document storage unit 12a, a document group storage unit 12b, and a document group search index storage unit 12c. There is one. Since these are all the same as the first embodiment described above, detailed description thereof is omitted.

(Module structure of document search system: temporary data section)
Further, as shown in FIG. 18, there are three temporary data portions shown in a tape form in the drawing: search key data 13a, document group search result data 13b, and document body data 13c. Since these are all the same as the first embodiment described above, detailed description thereof is omitted.

  The module configuration of the document search system according to the second embodiment of the present invention has been described above.

(Processing procedure of document retrieval system)
The processing procedure of the document search system in the second embodiment of the present invention is the same as the processing procedure in the first embodiment shown in the flowchart of FIG. That is, first, a document group is generated from a search target document stored in the document storage unit 12a in the document search server 1 and stored in the document group storage unit 12b (101), and feature words are extracted from the stored document group. The document group search index is generated and stored in the document group search index storage unit 12c (102). If the user instructs the document search client 2 to end the document search, the process ends (103). Otherwise, the input of the search key is accepted, stored in the search key data 13a, and transmitted to the document search server 1 ( 104). The document search server 1 receives the search key data 13a transmitted from the document search client 2 and searches the document group search index storage unit 12c based on the search key stored in the search key data 13a. The index is searched, and the result is stored in the document group search result data 13b and transmitted to the document search client 2 (105). The document search client 2 receives the document group search result data 13b transmitted from the document search server 1, and displays the stored document group together with the suitability rank. Note that the processing of step 101 and step 102 need only be executed once by the document search server 1 and does not need to be executed each time a search is performed.

On the other hand, in the second embodiment of the present invention, the details of the document group generation step 101 in the above processing are different from those in the first embodiment. Specifically, the document group generation step 101 in the second embodiment of the present invention extracts the cited document and the cited document that cites the document from the search target document set, and extracts the word from the cited document. The term clustering technique is applied and related words are grouped to generate a cluster. Next, the document clustering technique is applied to the citation source document to generate a document cluster. A document group is generated by assigning a cluster of citation documents to.
The document search system according to the second embodiment of the present invention has been described above.

[Example 3]
Next, a document search system according to a third embodiment of the present invention will be described with reference to the drawings. Since the overall configuration of the system according to the third embodiment of the present invention and the hardware configuration of the computer constituting the system are the same as those of the first embodiment of the present invention described above, description thereof will be omitted.

(Module configuration)
FIG. 19 is a diagram showing the module configuration of the document search system according to the third embodiment of the present invention. Hereinafter, each module shown in FIG. 19 will be described.
(Module configuration of document search system: processing section)
As shown in FIG. 19, the processing unit of the document search system includes a document group generation unit 11a, a document group search index generation unit 11b, a document group search unit 11c, a server communication unit 11d, and a document search in the document search server 1. The client 2 includes a client communication unit 11e, a search key receiving unit 11f, and a search result display unit 11g.

  The document group generation unit 11a in the document search server 1 generates a document group based on the citation relationship between documents from the document stored in the document storage unit 12a, and stores the document group in the document group storage unit 12b. The document group generation unit 11a further includes a cited part acquisition unit 11a3 and a document collection unit 11a4. The cited part obtaining unit 11a3 obtains a cited part in the cited document. The document summarizing unit 11a4 collects citation documents having the same citation destination part and generates a document group.

  Among the processing units shown in FIG. 19, processing units other than the document group generation unit 11a, that is, the document group search index generation unit 11b, the document group search unit 11c, the server communication unit 11d, the client communication unit 11e, the search key reception unit 11f, and The search result display portion 11g is the same as that of the first embodiment of the present invention described above, and therefore the description thereof is omitted.

(Module configuration of document retrieval system: data storage unit)
As shown in FIG. 19, there are three data storage units shown in a drum shape in the figure: a document storage unit 12a, a document group storage unit 12b, and a document group search index storage unit 12c. Since these are all the same as the first embodiment of the present invention described above, detailed description thereof is omitted.

(Module structure of document search system: temporary data section)
Further, as shown in FIG. 19, there are three temporary data portions shown in a tape form in the figure: search key data 13a, document group search result data 13b, and document body data 13c. Since these are all the same as the first embodiment of the present invention described above, detailed description thereof is omitted.

  The module configuration of the document search system according to the third embodiment of the present invention has been described above.

(Processing procedure of document retrieval system)
The processing procedure of the document search system in the third embodiment of the present invention is the same as the processing procedure in the first embodiment shown in the flowchart of FIG. That is, first, a document group is generated from a search target document stored in the document storage unit 12a in the document search server 1 and stored in the document group storage unit 12b (101), and feature words are extracted from the stored document group. Then, a search index for the document group is generated and stored in the document group search index storage unit 12c (102). If the user instructs the document search client 2 to end the document search, the process ends (103). Otherwise, the input of the search key is accepted, stored in the search key data 13a, and transmitted to the document search server 1 ( 104). The document search server 1 receives the search key data 13a transmitted from the document search client 2 and searches the document group search index storage unit 12c based on the search key stored in the search key data 13a. The index is searched, and the result is stored in the document group search result data 13b and transmitted to the document search client 2 (105). The document search client 2 receives the document group search result data 13b transmitted from the document search server 1, and displays the stored document group together with the suitability rank. Note that the processing of step 101 and step 102 need only be executed once by the document search server 1 and does not need to be executed each time a search is performed.

In the third embodiment of the present invention, the details of the document group generation step 101 in the above processing are different from those in the first embodiment. Specifically, in the third embodiment of the present invention, the document group generation step extracts the cited document and the cited document that cites the document from the set of documents to be searched, and extracts the cited document from the cited document. The destination part is acquired, and a document group is generated by collecting the citation source documents quoting the acquired citation destination part.
The document search system according to the third embodiment of the present invention has been described above.

[Example 4]
Next, a document search system according to a fourth embodiment of the present invention will be described with reference to the drawings. Since the overall configuration of the system according to the fourth embodiment of the invention and the hardware configuration of the computer constituting the system are the same as those of the first embodiment of the present invention described above, description thereof will be omitted.

(Module configuration)
FIG. 20 is a diagram showing the module configuration of the document search system according to the fourth embodiment of the present invention. Hereinafter, each module shown in FIG. 20 will be described.

(Module configuration of document search system: processing section)
As shown in FIG. 20, the processing unit of the document search system includes a document group generation unit 11a, a document group search index generation unit 11b, a document group search unit 11c, a server communication unit 11d, and a document search unit 11h in the document search server 1. And a client communication unit 11e, a search key receiving unit 11f, and a search result display unit 11g in the document search client 2. Among these, the document group generation unit 11a, the document group search index generation unit 11b, the document group search unit 11c, the server communication unit 11d, the client communication unit 11e, the search key reception unit 11f, and the search result display unit 11g will be described first. Since this is the same as the first embodiment of the present invention, the description is omitted.

  The document search unit 11h searches the document search index stored in the document search index storage unit 12d based on the search key stored in the search key data 13a, and stores the result in the document search result data 13d. Since the document search unit 11h can be implemented by applying a general document search technique, detailed description thereof is omitted.

(Module configuration of document retrieval system: data storage unit)
As shown in FIG. 20, there are four data storage units shown in drum form in the figure: a document storage unit 12a, a document group storage unit 12b, a document group search index storage unit 12c, and a document search index storage unit 12d. . Among these, the document storage unit 12a, the document group storage unit 12b, and the document group search index storage unit 12c are the same as those in the first embodiment of the present invention described above, and thus detailed description thereof is omitted.

  The document search index 12d is an area for storing an index for searching for a document stored in the document storage unit 12a. The index stored in the document group search index storage unit 12c is a document group search index for searching a document group composed of a plurality of documents as one unit, whereas it is stored in the document search index 12d. The index differs in that it is a document search index for searching individual documents. The document search index can be generated by applying a general document search technique. Specifically, known means such as extracting feature words from each document and recording them together with their appearance frequencies may be applied. Therefore, detailed description of the generation means is omitted.

(Configuration of Document Search Index Storage Unit 12d)
FIG. 21 is a diagram showing the configuration of the document search index storage unit 12d. As shown in FIG. 21, the document search index storage unit 12d includes a document number 12d01 and a feature word 12d02. The document number 12d01 stores the document number corresponding to the index. Here, the document number refers to the number of the document stored in the document storage unit 12a. A feature word extracted from the document is stored in the feature word 12d02. Extraction of feature words from a document can be implemented by applying a known technique for extracting feature words from a document as well as feature words of a document group. Similarly to FIG. 10, only the character string of the feature word is described in the feature word 12d02 in FIG. 21, but the present invention is not limited to this. You may store with the value etc. which show the appearance frequency or importance of the feature word in a document.

  The first line in FIG. 21 stores a search index of a document whose document number 12d01 is “doc050789”. Specifically, “lung function”, “male”, “case”, and the like are stored in the feature word 12d02 as feature words of the document.

(Module structure of document search system: temporary data section)
As shown in FIG. 20, there are four temporary data portions shown in a tape form in the figure: search key data 13a, document group search result data 13b, document body data 13c, and document search result data 13d. Among these, the search key data 13a, the document group search result data 13b, and the document body data 13c are the same as those of the first embodiment of the present invention described above, and thus detailed description thereof is omitted.

  The document search result data 13d stores the result of the document search unit 11h searching the document search index stored in the document search index storage unit 12d based on the search key data 13a. More specifically, a list of document numbers that match the search key data 13a is stored together with the matching degree with the search key.

(Example of document search result data 13d)
FIG. 22 is a diagram illustrating an example of the document search result data 13d. In the figure, “doc053028”, “doc039402”, and the like are stored as a list of document numbers that match the search key. In the example of FIG. 22, a numerical value “5.798” indicating the degree of matching of the document group with respect to the search key is stored, but the present invention is not limited to this. The order suitable for the search key may be used.

  The module configuration of the document search system according to the fourth embodiment of the present invention has been described above.

(Processing procedure of document retrieval system)
FIG. 23 is a flowchart showing the processing procedure of the document search system in the fourth embodiment of the present invention.

  As shown in FIG. 23, the document search system first generates a document group from a search target document stored in the document storage unit 12a in the document search server 1 and stores it in the document group storage unit 12b (201). A feature word is extracted from the stored document group to generate a search index for the document group, and is stored in the document group search index storage unit 12c (202). Similarly, a feature word is extracted from the document stored in the document storage unit 12a, a document search index is generated, and stored in the document search index storage unit 12d (203). If the user instructs the document search client 2 to end the document search, the process ends (204). Otherwise, the input of the search key is accepted, stored in the search key data 13a, and transmitted to the document search server 1 ( 205).

  The document search server 1 receives the search key data 13a transmitted from the document search client 2 and searches the document group search index storage unit 12c based on the search key stored in the search key data 13a. Search the index. The result is stored in the document group search result data 13b and transmitted to the document search client 2 (206). Similarly, the document search server 1 searches the document search index stored in the document search index storage unit 12d based on the search key, stores the result in the document search result data 13d, and transmits it to the document search client 2 (207). ). The document search client 2 receives the document group search result data 13b and the document search result data 13d transmitted from the document search server 1, and stores the documents included in the document group search result data group in the document search result data 13d. The document search results are displayed in the order of suitability, and the document search results are displayed together with the rank of the suitability in the document group search results stored in the document group search result data 13b (208).

  Of the above steps, the document group generation step 201, the document group search index generation step 202, the end step 204, the search key reception step 205, and the document group search step 206 are the same as in the first embodiment of the present invention shown in FIG. It is. That is, the document group generation step 201 is a document group generation step 101, the document group search index generation step 202 is a document group search index generation step 102, the end step 204 is an end step 103, and the search key reception step 205 is a search key reception. Since step 104 and document group search step 206 are the same as document group search step 105, a detailed description thereof will be omitted. The document search step 207 can be realized by applying a general document search technique, and thus detailed description thereof is omitted.

  Hereinafter, the process of the search result display step 208 among the above processes will be described using the search result display screen shown in FIG.

(Example of search result display screen in document search system)
FIG. 24 is a diagram showing an example of a document search result display screen in the document search client 2 of the document search system according to the fourth embodiment of the present invention. As shown in the figure, the document search result display screen of the document search client 2 includes a search key input area 601, a search button 602, a document group search result display area 603, a document search result display area 604, and a document number display area 605. Composed.

  The search key input area 601 is an area where the user inputs a search key. In FIG. 24, a state in which “hypertension”, “pulmonary function”, and “blood vessel” are input as search keys is displayed. A search button 602 is a button for the user to instruct execution of the search. The document group search result display area 603 is an area where the document search client 2 displays the document group search result data 13b in descending order of the degree of matching with the search key, and the document search result display area 604 searches for the document search result data 13d. This area is displayed in descending order of the degree of matching with the key. The document number display area 605 is an area for displaying the number of documents displayed in the document group search result display area 603 and the document search result display area 604.

  Here, the document group search result display area 603, the document search result display area 604, and the document number display area 605 will be described in detail.

(Details of document group search result display area 603)
The document group search result display area 603 includes a document group search result rank 60301, a document search result rank 60302, a citation destination document 60303, and a citation source document 60304. The document group search result rank 60301 displays the rank of the degree of matching between the document group and the search key. The document search result rank 60302 displays the rank in the document search result of the citation target document or the citation source document included in each document group. When the cited document or the cited document is not included in the document search result, “-” or the like indicating that the rank is unknown is displayed. This is assumed to be a case where the citation destination document or the citation source document has a low degree of matching with the search key as an individual document. The cited document 60303 displays the cited document in the document group. A citation source document 60304 displays a document that cites the same portion of the citation destination document in the document group.

  In the example of FIG. 24, “high blood pressure affecting lung function ...” is displayed in the cited document 60303 of the document group having the second highest matching degree with the search key. This quoted document is displayed as “(12)” that the rank is 12th in the document search result. In addition, “Pulmonary Ventilation Disorder and Hypertension…” and “Chronic Lung Disease Patient…” are displayed as the cited documents that cite the same cited place in the cited document. It is displayed that the ranks are first and fifth in the document search result.

(Details of document search result display area 604)
The document search result display area 604 includes a document search result rank 60401, a document group search result rank 60402, and a document 60403. The document search result rank 60401 displays the rank of the degree of matching between the document and the search key. The document group search result rank 60402 displays the rank in the document group search result of the document group including the document. When the document group including the document is not included in the document group search result, “-” or the like indicating that the rank is unknown is displayed. A document 60403 displays a document that matches the search key.

  In the example of FIG. 24, “(2)” of the document “chronic lung disease patient ...” having the fifth highest matching degree with the search key is ranked second in the document group search result display area 603. It is included in the document group.

(Details of document count display area 605)
The document number display area 605 includes a display number 60501, a reference estimate 60502, and a referenced 60503.

  In the display number 60501, the number of documents displayed in the document group search result display area 603 and the document search result display area 604 is displayed. The documents included in the document group displayed in the document group search result display area 603 are different for each document group. Further, there is an overlap between the document included in the document group and the document displayed in the document search result display area 604. Therefore, the number of documents included in the search result and the number of documents are displayed. In the case of a document group, the number of documents included in the document group is displayed.

  In the example in the figure, the number of documents displayed and the number of documents included in the search result are “20” in the display number 60501. In the case of the document group, the documents included in the 20 document groups (the cited document and It is indicated that the total number of citation documents is “42”. On the other hand, the number of documents is “20” because the number of documents included in the search result matches the number of documents. Further, it is shown that the number of differences excluding duplication among the number of documents included in the document group search result and the documents included in the document search result is “46” in total.

  In the reference estimate 60502, when referring to the displayed document, the result of calculating the total number of documents referred to by referring to the document group search result and the document search result is displayed. Specifically, in the reference estimate 60502, the user can input either the number of document groups to be referred to, the number of documents, or the total number of documents to be referred to. For example, when the user inputs “20” as the value of “total”, the system calculates the document group and the number of documents necessary for referring to the total of 20 documents, and includes “5” document groups (included). The number of documents is 15), and a numerical value indicating that “5” documents should be referred to is displayed. Specifically, the document group included in the document group search result and the document included in the document search result are acquired one by one in order from the top of the rank, and the total number of documents is calculated excluding any duplicate documents. To do. If the total number of documents exceeds the number specified by the user, in the example in the figure, if it exceeds “20”, the number of document groups and documents obtained from the top of the document group search results and the document search results are referred to. It is displayed in the document group number and document number of the estimate 60502.

  As another example, when the user inputs “20” as the “total” value and “5” as the “document” value, the system calculates the number of document groups necessary to refer to the total of 20 documents. To do. Specifically, the document group included in the document group search result is acquired one by one in order from the top of the rank, and the number of documents is excluded unless there is a document that overlaps with the document included in the top five of the document search result. Calculate the total. If the total number of documents exceeds the number specified by the user, the number of document groups acquired from the top of the document group search result is displayed in the document group number of the reference estimate 60502.

  Referenced 60503 indicates the number of document groups referred to by the user in the document group search result, the number of documents included in the document group, the number of documents referred to in the document search result, and the total thereof. In the example of reference 60503 in the figure, the number of document groups referred to by the user in the document group search result is “2”, and the number of documents included in the referenced document group is “5”. The number of documents referred to in the result is “0”, which indicates that a total of “5” documents have been referred to.

  The document search system according to the fourth embodiment of the present invention has been described above.

  The present invention can be applied to a system that searches documents, and in particular, can be applied to a system that searches documents having a citation relationship between documents such as papers and patents.

The figure which shows the outline | summary of the procedure of the search index production | generation in a document search. Explanatory drawing about the citation relationship between documents. Explanatory drawing about the citation relationship between documents. The figure explaining an example of a document group production | generation method. The figure which shows the structural example of a document search system. The figure which shows the hardware structural example of a document search server and a document search client. The figure which shows the module structural example of the document search system by this invention. The figure which shows the structural example of a document storage part. The figure which shows the structural example of a document group storage part. The figure which shows the structural example of a document group search index storage part. The figure which shows an example of search key data. The figure which shows an example of document group search result data. The flowchart which shows the example of the process sequence of the document search system by this invention. 6 is a flowchart illustrating an example of a processing procedure for generating a document group. The flowchart which shows the example of the process sequence of document group search index production | generation. The figure which shows the example of the inspection result display screen of the document search system by this invention. The figure which shows the example of the inspection result display screen of the document search system by this invention. The figure which shows the structural example of the document search system by this invention. The figure which shows the structural example of the document search system by this invention. The figure which shows the structural example of the document search system by this invention. The figure which shows the structural example of a document search index storage part. The figure which shows an example of document search result data. The flowchart which shows the example of the process sequence of the document search system by this invention. The figure which shows the example of the inspection result display screen of the document search system by this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Document search server, 2 ... Document search client, 3 ... Communication line, 10 ... CPU, 20 ... Hard disk, 30 ... Memory, 40 ... Display, 41 ... Display control part, 50 ... Keyboard, 51 ... Keyboard control part, 60 ... Mouse, 61 ... Mouse control unit, 70 ... Bus, 11a ... Document group generation unit, 11a1 ... Destination document division unit, 11a11 ... Destination document structure division unit, 11a2 ... Document allocation unit, 11b ... Document group search index generation , 11c: Document group search unit, 11d: Server communication unit, 11e ... Client communication unit, 11f ... Search key reception unit, 11g ... Search result display unit, 12a ... Document storage unit, 12b ... Document group storage unit, 12c ... Document group search index storage unit, 13a ... search key data, 13b ... document group search result data, 13c ... document text

Claims (15)

A document group generation means for generating a document group based on a citation relationship between documents from a set of documents to be searched;
And a document group search index generating means for generating a search index of the document group by extracting feature words from the document having the citation relationship.   2. The document search system according to claim 1, wherein the document group generation means uses a document cited from another document as a cited document, and a document that cites another document as a cited document. A document retrieval system characterized in that documents having the same document are collected to generate a document group.   3. The document search system according to claim 2, wherein the document group generation means uses a document cited from another document as a cited document, and a document that cites another document as a cited document. A cited part obtaining unit that obtains a cited part cited from another document in the document, and a document collecting unit that collects documents having the same cited part and generates a document group. Feature document retrieval system.   3. The document retrieval system according to claim 2, wherein the document group generation unit uses a document cited from another document as a citation destination document, and a document citation of another document as a citation source document. A document search system comprising: a citation destination document dividing means for dividing a destination document; and a document assignment means for generating a document group by assigning a citation source document to a portion of the divided citation destination document.   5. The document search system according to claim 4, wherein the cited document dividing means divides the document according to the document structure of the cited document.   5. The document search system according to claim 4, wherein the cited document dividing means divides the document by clustering feature words of the cited document.   5. The document search system according to claim 4, wherein the document assigning means includes a citation source document clustering means for clustering a plurality of citation documents that cite the same document, and a citation destination divided by the citation destination document dividing means. A document retrieval system comprising: a citation source document cluster allocating unit for generating a document group by allocating a cluster of the citation source document to a document part.   4. The document search system according to claim 3, wherein the document group search index generation means cites the cited part of the cited document obtained by the cited part obtaining means and the obtained cited part. A document search system that extracts a feature word from a part of an original document and generates a search index of a document group.   9. The document search system according to claim 8, wherein the document group search index generation means extracts a feature word from the cited part of the cited document and the part of the cited document, and is higher in terms of words appearing in the cited part. A document retrieval system characterized by extracting feature words by assigning weights.   5. The document search system according to claim 4, wherein the document group search index generating means is assigned to the cited document part divided by the document dividing means and the cited document part assigned by the document assigning means. A document retrieval system characterized by extracting feature words from a document.   The document search system according to claim 10, wherein the document group search index generation unit extracts a feature word from a quoted document part and a document assigned to the cited document part. A document search system characterized by extracting feature words by giving higher weights to words appearing in a part. The document search system according to claim 2, further comprising: a document group search unit that searches a search index of the document group; and a display unit that displays a search result.
The display means displays the document group search result acquired by the document group search means, displays the cited document and the cited document in parallel in the document group display, and can further identify the cited location of the cited document The document search system is characterized in that the words appearing in the citation source document are displayed in reverse, and the words appearing at the citation destination location are highlighted. 13. The document search system according to claim 12, further comprising a document search means for searching a document search index generated by extracting feature words from documents included in a document set to be searched,
The document search system, wherein the display means displays the document group search result acquired by the document group search means in comparison with the document search result acquired by the document search means.   14. The document search system according to claim 13, wherein the display unit arranges the citation source documents included in the document group search result acquired by the document group search unit in a rank in the document search result acquired by the document search unit. A document search system characterized by being displayed instead.   14. The document search system according to claim 13, wherein the display means adds a rank in the document group search result acquired by the document group search means to the display of the document search result acquired by the document search means. A document retrieval system characterized by that.

Images