JP2004348771A - Technical document retrieval device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a technical document retrieval device capable of clearly specifying a mutual relation among retrieved technical documents. <P>SOLUTION: This technical document retrieval device for retrieving technical documents such as a patent document and a technical article is provided with a retrieval means 12 and a related document extraction means 24 for extracting the information of a reference document quoted in each technical document. The related document extraction means extracts the information of the reference document quoted by the original technical document; the retrieval means retrieves the technical document corresponding to the reference document extracted by the related document extraction means; the related document extraction means extracts the information of the reference document quoted by the technical document retrieved by the retrieval means; and the respective technical documents are displayed so that relations among them can be visually understood by repeating those operations. Since connection among the documents can be retrieved and displayed, a user can easily grasp a technical system and a technical trend. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a technical document search device for searching for technical documents such as patent documents and technical papers, and more particularly, to a device capable of performing an accurate search and displaying a search result in an easily understandable manner. is there

The number of patent documents and technical papers has reached an enormous amount even for existing issuance alone, and it is certain that it will continue to increase in the future. In order to effectively use these technical documents, it is essential to have a search means capable of selecting a document in a target field accurately.
Conventionally, a document search method includes a keyword search and a method of searching for a document similar to a search query using a vector space method.

FIG. 20 shows a configuration of a conventional document search device. This apparatus includes: an input / output unit 11 for inputting a search condition and outputting a search result; a document entity storage unit 14 in which a technical document is stored; an index storage unit 13 in which a search index of the technical document is stored; Search means 12 for searching for a technical document using the search index.
The index storage unit 13 stores keywords, vector data, dictionaries, and the like extracted from each technical document as an index of each technical document stored in the document entity storage unit 14.
When a search condition expression or a search query is input from the input / output unit 11, the search unit 12 uses the index stored in the index storage unit 13 to search for a document that matches the search condition expression or a document similar to the search query. And the search result is displayed through the input / output means 11. The search result document entity is read out from the document entity storage means 14 and displayed according to the selection operation from the input / output means 11.

In addition, in order to show the searched patent documents in an easy-to-understand manner, display using a patent map has been conventionally performed.
FIG. 19 shows an example of the patent map. This example is a patent map relating to a personal computer ink jet printer, which describes the relationship between the components of the printer, the number and keyword of the related patent document for each component, the extraction date of the patent document, the patent classification, and the like. For patent documents, the applicant and the symbol indicating the technical evaluation are displayed.
An apparatus that supports the creation of such a patent map is described in Patent Document 1 below. In this device, patent documents are automatically classified according to a predetermined keyword, patent classification, age, and the like.

Patent Document 2 below describes a system for automatically updating patent documents displayed on a patent map. In this system, when a search key is sent from a client terminal to a patent database server in advance, when the patent database is updated, the patent data in the updated range is searched using the search key, and the search result is sent to the client terminal. It is added to the terminal's patent map database.
As described above, the patent map in which the searched patent documents are displayed is widely used as a material for grasping a technical trend and a right relationship.
It is also widely practiced to display a graph of a set of retrieved patent documents in terms of the number of cases according to the filing date to indicate the trend of filing in the relevant technical field. Japanese Patent Application Laid-Open No. 7-182356 discloses that patent documents and technical information are stored in a database together with data such as fields of use, elemental technologies, and issuance dates. A method of displaying in a technical information space using an axis and a time axis as coordinate axes is disclosed.
JP-A-7-182356 JP-A-8-222435

However, when searching for technical documents using keywords, even if they are related documents, technical documents with completely new concepts or technical documents described using words other than the words specified in the keywords may be omitted from the search. May not be classified correctly.
In technical terms, for example, a `` filter '' has the meaning of a circuit component that limits the frequency range of a signal in the electrical field, and a porous material that separates suspended particulate matter from a liquid in the mechanical field. In the optical field, the same term is often used with different meanings in the technical field to represent an optical component that limits the amount of light. Therefore, in a keyword search of a technical document, a large amount of noise may be included in the search result unless a keyword is appropriately selected.
Further, an apparatus for organizing and displaying searched technical documents from various viewpoints is described in the above-mentioned Japanese Patent Application Laid-Open No. 7-182356 or the like, but focuses on the relevance between the searched technical documents. A device for displaying the relationship between technical documents has not been made so far.

  SUMMARY OF THE INVENTION The present invention solves such a conventional problem. It is possible to efficiently and accurately search and classify technical documents. In addition, the present invention specifies the relationship between the searched technical documents. It is an object of the present invention to provide a technical document search device capable of performing such a search.

  Therefore, according to the present invention, in a technical document search device for searching for a technical document such as a patent document or a technical paper, a search means and a related document extracting means for extracting information of a reference document cited in the technical document are provided. The related document extracting means extracts information of a reference cited in an original technical document, and the search means searches a technical document corresponding to the reference extracted by the related document extracting means. Then, the related document extracting means extracts the information of the reference cited in the technical document searched by the searching means, and repeats this operation so that each of the searched technical documents has a visual relation between them. It is displayed so that it can be easily understood.

Further, a search unit and a related document extracting unit for extracting information of a reference cited in a technical document are provided, and the related document extracting unit extracts information of the reference cited in each technical document. Using the information of the reference document extracted by the related document extracting device to search for a technical document citing the original technical document as a reference document. A search is made for a technical document in which the document is cited as a reference document, and this operation is repeated so that each of the searched technical documents is displayed so that their relationship can be visually recognized.
Therefore, in this device, the search results can be appropriately classified using the information of the technology map.
In addition, it is possible to display the relationship between technical documents included in the bibliographic items and sharing the inventor and the applicant, and to display the relationship between reference-related technical documents.

By using a technical map such as a patent map for a search, the technical document search device of the present invention can easily obtain a desired search result without the user having to make a difficult specification.
Further, based on the search result, the technical document to be registered in the technical map can be accurately and easily identified, and the technical map data can be updated accurately. As a result, technology trends can be more easily and accurately grasped.
Further, the technical document search device of the present invention can search and display a connection between documents, so that a user can easily grasp a technical system and a technical trend, and can search for a related document from the connection. Accordingly, search omissions can be reduced, and a simple and accurate document search can be performed.

(1st Embodiment)
In the first embodiment, a description will be given of a technical document search device that uses information of a patent map, which has been conventionally used to display a search result of a technical document, to classify the searched patent document.
As shown in FIG. 1, the apparatus includes an input / output unit 11 for inputting search conditions and outputting a search result, a document entity storage unit 14 in which patent documents are stored, and a search index for patent documents. Index storage means 13, search means 12 for searching patent documents using a search index, analysis means 15 for analyzing document vectors of patent documents, and a plurality of patent maps created so far are stored. Technology map storage means 17, a technology map storage means 18 for storing the patent map read from the technology map storage means 17, and a document similarity determining the similarity between patent documents using a document vector Determining means 16;
Each patent map stored in the technology map group storage means 17 has a tree-like data structure as shown in FIG. 2, and in this example, a control unit, a computer interface, a printing unit, etc. of an inkjet printer for a personal computer. The search keywords and important patent list data are described for each component. When this patent map is displayed on the screen, it is displayed, for example, as shown in FIG.

The user uses the input / output means 11 to specify a patent map to be used by the user when using the technical document search device. The designated patent map is read from the technology map group storage means 17 and stored in the map storage means 18.
Next, the user inputs a favorite keyword and searches for patent documents. The search means 12 searches for a document matching the keyword using the index stored in the index storage means 13, and the search result is displayed through the input / output means 11. FIG. 3 shows a display screen of a search result when a patent document is searched using “color * high speed” as a keyword. Here, 2000 search results are obtained.

Even if a large number of search results are obtained in this way, by pressing the "classification display" button on the display screen, as shown in FIG. 4, the search results are classified and displayed for each configuration diagram of the patent map. .
When "classification display" is input from the input / output means 11, inside the apparatus, the similarity between each of the search result patent documents and the patent document described for each component of the patent map is calculated, and the search is performed. The resulting patent document is categorized into the most similar component parts. The determination of the similarity is performed using a known technique such as a vector space method.

This process is performed in the following procedure.
(1) The analysis means 15 calculates a document vector of each patent document described for each component of the patent map (this may be calculated in advance).
(2) The analysis means 15 extracts one patent document as a search result and calculates a document vector of this patent document.
(3) The document similarity determination means 16 compares the document vector of the patent document of each component in the patent map with the document vector of the patent document of the search result, and determines the closest similarity to the closest one. Classify. (If they do not resemble anywhere, they are classified as "Other." If they are too different in the overall document space, classify them as "Not relevant" and discard the search results.)
(4) The patent document of the next search result is extracted, and the processing from (2) is repeated. If there is no patent document of the next search result, the process ends.
As described above, in this technical document search device, the searched patent documents can be classified and displayed using the information of the patent map.

In this apparatus, when comparing the similarity between the patent document of the search result and the patent document described in the patent map, the document vectors of the entire document were calculated and compared. It can also be configured to take out and compare.
In this case, as shown in FIG. 5, the apparatus is provided with a partial document extracting means 19 for extracting a part of the place described in the patent document. Other configurations are the same as those of the apparatus shown in FIG.
When calculating the degree of similarity, the partial document extracting unit 19 extracts the description part of “conventional technology” from each of the patent document of the search result and the patent document of the patent map. The analysis unit 15 calculates the document vector of the “conventional technology”, and the document similarity determination unit 16 compares the document vector of “the conventional technology” in the patent document of the search result with “the conventional technology” in the patent document in the patent map. It is compared with the document vector of the "prior art", and classified into the closest place.

Patent documents are very long, and the “embodiment” occupying the largest data amount may include various contents such as various examples and comparative examples. Therefore, when the similarities are compared for the entire patent document, not only the processing load is large, but also the classification accuracy may be reduced.
In this regard, the “conventional technology” emphasizes what has been solved and what has been improved so far, so create a document vector only for this “conventional technology” and compare it with each other. Allows classification.
In the same manner, the "subject to be solved by the invention" of the patent document may be extracted by the partial document extracting means 19, and the document vectors of this portion may be compared and classified.
In the case of a technical paper, a part that refers to a past document is extracted at the center, a document vector is created only from this part, and comparison and classification are performed. In the case of technical papers, unlike patent documents, the format is free and it is difficult to extract specific parts.However, parts surrounding references to past documents are not Since it is conceivable, the description part of the related art can be extracted and classified using the description part of the past document as a guide.
As described above, with this technical document search device, it is possible to increase the accuracy of classification with a small processing load.

Further, as shown in FIG. 6, the technical literature search apparatus is provided with a display synthesizing means 20 for synthesizing display data in which similarity is entangled. It can be displayed in terms of points.
When the "multi-dimensional classification" button on the display screen of FIG. 3 showing the search results is pressed, the bibliographic information (applicant, etc.), time (application date, etc.), and components are displayed as shown in FIG. A point representing the position of a patent document as a search result is displayed in a multidimensional space as an axis.
At this time, the display synthesizing unit 20 displays the size of the point representing the patent document in the search result in correspondence with the magnitude of the similarity calculated by the document similarity determining unit 16.

When the user clicks on the classified display point with the mouse, the corresponding patent document is read from the document entity storage means 14 and the content is displayed. When the "Add to patent map" button is clicked while the content is displayed, the patent document is added to the patent map. Therefore, the user can select a patent document to be added to the patent map based on the size of points in the multidimensional space.
As described above, this technology search device has the function of supporting the creation of the patent map, and can easily update the patent document registered in the patent map.
Here, the case where the size of the display point of the patent document is differentiated by the similarity has been described, but the color or shape of the point may be differentiated according to the similarity.

(Second embodiment)
In the second embodiment, a description will be given of a technical document search device that uses information of a patent map, which has been conventionally used to display search results of technical documents, to narrow down the number of patent document searches.
As shown in FIG. 8, the apparatus includes a search formula combining unit 21 that combines a search formula obtained by adding a keyword of a patent map to a search formula set by a user. Other configurations are the same as those in FIG. 6 of the first embodiment.
The technology map group storage means 17 of this device stores a patent map created in advance. When using the technical document search device, the user specifies a patent map to be used by the user. The designated patent map is read from the technology map group storage means 17 and stored in the technology map storage means 18.

  Next, the user inputs a favorite keyword from the input / output unit 11 and searches for a patent document. When a keyword is input from the input / output unit 11, the search formula synthesizing unit 21 adds a basic keyword specified in the patent map and bibliographic items (such as IPC classification) in addition to the keyword specified by the user. Are automatically assembled, and the search means 12 performs a search using this search formula. The document similarity determination means 16 determines the similarity between the patent document in the search result and the patent document described in the patent map, and excludes those having extremely low similarity (this determination is not narrowed much. To some extent) Display search results.

  For example, suppose that the user wants to search for a patent related to “FAX server device”. Searching for "(FAX + Facsimile + Fax) * Control * Image * Communication * Compression" using a conventional search device reveals that there are numerous NC control device (machine tool numerical control device) patents in addition to the desired FAX server related patents. Will be searched. In addition to narrowing down by IPC classification, the user has to specify words such as NC with the NOT operator and narrow down the search little by little. This requires know-how for each patent field.

In this technical document search device, if the user has a patent map relating to the FAX server device in advance, the patent of the FAX server device can be searched only by designating the patent map.
In addition, when a keyword is specified only for "color", a normal search device searches for everything from a color television to a color film, and the number reaches a huge amount. Can search only the patent document of the FAX server device related to the color FAX.
As described above, this technical document search device can efficiently and accurately search for technical documents by synthesizing search conditions using the patent map.

(Third embodiment)
In the third embodiment, a description will be given of a technical document search device that searches for patent documents having commonality in bibliographic items and displays a relationship between them.
As shown in FIG. 9, the apparatus includes a bibliographic item extracting unit 22 for extracting bibliographic items of a patent document, a search formula synthesizing unit 21 for synthesizing a search formula from the bibliographic items, and a relation between the searched patent documents. And a document relation storage means 23 for storing Other configurations are the same as those in FIG. 1 of the first embodiment.

The user inputs a keyword from the input / output unit 11 and searches for patent documents. As a result, it is assumed that a certain number of patent documents are searched as shown in the screen example of FIG. Suppose one of them was equivalent to the patent document you were looking for. Therefore, the user tries to view a patent document similar to this patent document (called an original patent document). The subsequent procedure is shown in the flowchart of FIG.
Step 1: As shown in FIG. 12, the user selects a document on the screen and presses a "similar bibliographic search" button. Then, the bibliographic item extracting means 22 of the technical document search device extracts the inventor and the applicant from the selected original patent document. It is assumed that the name of “Applicant A” is extracted as the applicant and the names of “Inventor 1”, “Inventor 2”, and “Inventor 3” are extracted as inventors.
Step 2: Search formula combining means 21 composes a search formula of applicant A * inventor 1 + applicant A * inventor 2 + applicant A * inventor 3.
Step 3: The search means 12 performs a search using this search formula. As a result, it is assumed that 50 search results are obtained.
Step 4: The document similarity determination means 16 calculates the similarity with the original patent document for all the obtained 50 cases. If the similarity is equal to or greater than a predetermined threshold value, that is, if the similarities are similar to some extent, the result is retained, but other (dissimilar) search results are discarded.
Step 5: The remaining search results are stored in the document relation storage means 23, and are sorted and output and displayed in chronological order (application order), similarity order, or chronological order.

In this way, as shown in FIG. 13, the similarity between the patent documents common to the applicant and the inventor can be displayed on the time axis.
In this way, this device extracts only some bibliographic items from the original document, synthesizes a search formula from the extracted bibliographic items, conducts a search using this search formula, and searches for similar items from the obtained search results. By searching for and listing only similar ones, it is possible to display the relationship between patent documents having common bibliographic items.

(Fourth embodiment)
In the fourth embodiment, a description will be given of a technical document search device that searches for patent documents having a reference relationship and displays a relationship between the patent documents.
As shown in FIG. 14, the apparatus includes a related document extracting unit 24 for extracting a document referred to in a patent document, a document relation storage unit 23 for storing a relationship between retrieved patent documents, Display synthesizing means 20 for displaying related patent documents in association with each other. Other configurations are the same as those in FIG. 1 of the first embodiment.

The user searches for a patent document by inputting a certain keyword from the input / output unit 11. As a result, as shown in the screen example of FIG. 11, it is assumed that a certain number of patent documents are searched, and one of them is equivalent to the patent document that has been searched. Therefore, the user tries to view a patent document similar to this patent document (referred to as a seed patent document). The subsequent procedure is shown in the flowchart of FIG.
Step 11: The user selects a seed patent document on the screen and presses a “reference document search” button.
Step 13: The related document extracting means 24 of the technical document search device extracts (plural) reference document names cited in the seed patent document from the seed patent document. This is called the parent patent document.
Step 14: The document relation storage means 23 stores the relation between the seed patent document and the parent patent document.
Step 15: Acquire the parent patent document from the document entity storage means 14,
Step 16: Using the parent patent document as a seed, steps 13 to 16 are repeated until N generations (N can be specified) from the first seed patent document.
Step 17: Set the first seed patent document as the seed patent document again,
Step 19: A patent document which refers to this type of patent document is searched by searching the patent document stored in the document entity storage means 14 by full-text search.
Step 20: The document relation storage means 23 stores the relation between the retrieved result (child patent document) and the seed patent document.
Step 21: Steps 19 to 21 are repeated until after M generations (M can be specified) for all child patent documents as seeds.
Step 22: The display synthesizing means 20 displays the relation between the documents stored in the document relation storage means 23 in steps 14 and 20.

16 and 17 show examples of the screen displayed in this manner.
In addition, the extraction of the document name referred to in the literature is described in a formal manner such as “Japanese Patent Publication No. 11-12345” in the case of a patent literature, so that the computer can easily recognize it. In the case of a paper, the reference is described formally at the end of the sentence, such as [8] Yasuki Iizuka, word segmentation by the minimum connection establishment method, 200-NL-139, pp11, 2000, etc. The computer can be easily recognized from the description part.
Also, when searching for a child patent document (child), a full-text search of the database based on the patent publication number of the seed patent document (owner) allows a child patent document (children) referencing the seed patent document (self). It is possible to search for

In addition, whether to search for documents that correspond to siblings' children (nephews), how far back their parents (N generations or X years ago), and how far they should follow their children (M generations or all until now), etc. , So that you can choose.
Further, a document referred to by more documents (a document with many children) may be highlighted by, for example, adding a color when displaying the document. This is to show that having more children is more important, and to make it clear.
Also, in FIGS. 16 and 17, only the patent publication number is displayed for explanation, but in addition to the publication number, the display of the name, the applicant and the like can be selected, or as shown in FIG. Only when the mouse pointer is placed at the position of the public number in the figure, the bibliographic information and the summary content may be displayed in a pop-up display.

Further, by double-clicking the patent publication number box in FIGS. 16 and 17, the patent document can be immediately browsed. If there is a reference patent (parent patent) number in the viewed patent document, the reference patent (parent patent) is displayed in reverse (or colored), and clicking on it displays the reference patent (parent patent). You may comprise so that it may be.
The similarity between related patent documents may be determined by the document similarity determination unit, and the similarity may be displayed so as to be visually recognized.

As described above, this technical document search device can search for a reference relationship between documents and display the relationship between the documents on a screen. By browsing this screen,
You can easily grasp the technical system and technical trends.
Note that, in the embodiment, the search for patent documents has been mainly described, but the technical document search device of the present invention can also be applied to technical papers and the like.

  The technical document search device of the present invention can be used to search for technical documents such as patent documents and technical papers.

FIG. 2 is a block diagram illustrating a configuration of a technical document search device according to the first embodiment; Diagram showing the data structure of the patent map, FIG. 7 is a diagram showing a search result screen of the technical document search device of the first embodiment; FIG. 3 is a diagram showing a patent map classification display screen of the technical document search device of the first embodiment; FIG. 3 is a block diagram illustrating another configuration of the technical document search device according to the first embodiment; FIG. 3 is a block diagram showing another configuration of the technical document search device according to the first embodiment; FIG. 3 is a diagram showing a three-dimensional classification display screen of the technical document search device according to the first embodiment; FIG. 4 is a block diagram illustrating a configuration of a technical document search device according to a second embodiment; FIG. 9 is a block diagram illustrating a configuration of a technical document search device according to a third embodiment; FIG. 9 is a flowchart showing an operation procedure of the technical document search device according to the third embodiment; FIG. 10 is a diagram showing a search result screen of the technical document search device of the third embodiment. FIG. 10 is a diagram showing a similar bibliographic search selection screen of the technical document search device according to the third embodiment; FIG. 10 is a diagram showing a display screen of the technical document search device according to the third embodiment; FIG. 14 is a block diagram illustrating a configuration of a technical document search device according to a fourth embodiment; FIG. 14 is a flowchart showing an operation procedure of the technical document search device according to the fourth embodiment; FIG. 14 is a diagram showing connections between documents displayed by the technical document search device according to the fourth embodiment; The figure which shows the three-dimensional classification display screen in the technical document search device of 4th Embodiment. FIG. 14 is a diagram illustrating a pop-up display on the technical document search device according to the fourth embodiment; Diagram showing patent map, FIG. 11 is a block diagram illustrating a configuration of a conventional technical document search device.

Explanation of reference numerals

11 Input / output means
12 Search method
13 Index storage means
14 Document entity storage means
15 Analytical methods
16 Document similarity judgment means
17 Technology map group storage means
18 Technology map storage
19 Partial document extraction means
20 Display synthesis means
21 Search formula synthesis means
22 Bibliographic item extraction means
23 Document storage means
24 Related Document Extraction Method

Claims (2)

In a technical document search device that searches for technical documents such as patent documents and technical papers,
Search means,
A related document extracting means for extracting information on a reference document cited in the technical document, wherein the related document extracting means extracts information on the reference document cited in the original technical document, The search unit searches for a technical document corresponding to the reference document extracted by the related document extraction unit, and the related document extraction unit searches information of the reference document cited in the technical document searched by the search unit. A technical document search apparatus, which extracts and repeats this operation to display each of the searched technical documents so that their relationship can be visually recognized. In a technical document search device that searches for technical documents such as patent documents and technical papers,
Search means,
Related document extracting means for extracting information of a reference document cited in a technical document, wherein the related document extracting means extracts information of a reference document cited in each technical document, and Searching for a technical document citing the original technical document as a reference document using the information of the reference document extracted by the related document extracting means, and further citing the searched technical document as a reference document A technical document search apparatus that searches for a technical document that has been searched for, and repeats this operation to display the searched technical documents so that their relationship can be visually recognized.

Images