JP4389102B2 - Technical literature search system - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a system for searching technical documents such as a patent information database.
[0002]
[Prior art]
As a conventional patent information search system, for example, a patent information search system such as PATOLIS (registered trademark) -WEB provided by Patrice Co., Ltd. is known. When conducting a prior art search using this conventional system, or a literature search for an intermediate processing objection / invalidation procedure, search criteria such as classification codes for international patent classifications and keywords should be expressed in logical formulas. Text search and so on. A survey using such a search system is generally performed in the following procedure.
Search is performed under appropriate conditions such as procedure 1, classification code and logical expression.
The primary screening is performed by paying attention to the procedure 2, the summary part of the document of the search result, and the like.
Procedure 3 The literature extracted in the primary screening is examined and secondary extraction is performed.
Step 4 From the secondary extracted documents, a portion corresponding to the configuration requirement is extracted and a comparison document is created.
[0003]
Further, in the patent publication search system disclosed in Patent Document 1, the patent publication data batch-transferred from the patent information database is reconstructed by a special processing support program and used, for example, in a company development section. In order to make it easier to understand, it is possible to understand at a glance the development status of the field of specific technology, prior art, etc., so that it is re-edited into a format that can be easily reused for each section in the company etc. There is something.
[0004]
On the other hand, Patent Documents 2 to 4 disclose various search techniques such as general document data although they are not directly related to the patent information database.
Patent Document 2 includes document data structured using a markup language such as SGML (Standard Generated Markup Language), that is, the names and ranges of document components such as a document title, a chapter, and a body. A structured document database is constructed from documents marked in a document using appropriate symbols, and similar documents are searched using keywords constructed by document components.
[0005]
Patent Document 3 discloses a document structure element that is divided into logical structures such as chapters, sections, paragraphs, and frames using a markup language as a document structure element for constructing a database of document data. A data structure is stored in a database, and retrieval is performed using a database retrieval technique or the like.
[0006]
Patent document 4 is an example of constructing and searching with both words extracted from a document as search unit data and document components such as “sentence” and “paragraph” found in other patent documents. Disclosure.
Then, the search relevant document search word related calculation unit for determining the search result calculates the position information of “word” or “document component”, for example, checks how many times the search word appears in the sentence, and determines the appearance position. It is expressed by position information seen from the end of the sentence (or the beginning of the sentence), and the proximity calculation is performed to determine the fitness.
[0007]
[Patent Document 1]
JP 2001-22794 A (paragraph [0020], FIG. 1)
[Patent Document 2]
Japanese Unexamined Patent Publication No. 7-44567 (paragraph [0019], FIG. 1)
[Patent Document 3]
Japanese Patent Laid-Open No. 8-44766 (paragraph [0010], FIG. 3)
[Patent Document 4]
JP 2002-189754 ([Summary], FIG. 1)
[0008]
[Problems to be solved by the invention]
In the above conventional technique, in step 1, the search condition is determined for the entire document data (searching whether each word specified in the search condition is present in the document data), and thus the document having low relevance. There is a problem that many searches are performed (generally called “noise”).
In the case of the procedure 2, since screening is performed by paying attention to only a part of the document data, there is a problem that important documents may be overlooked.
In the case of procedures 3 and 4, there is a problem that the amount of work becomes enormous because the work is performed while creating a configuration requirement check table and the like and grasping the correspondence between each document.
[0009]
If the technique shown in Patent Document 1 is used, the above-mentioned procedures 2 to 3 are made efficient, but it is still necessary to perform a reading inspection of the document data one by one.
If the technique disclosed in Patent Document 2 is used, it is possible to perform inspections in order of similarity, so that there is an increased possibility of finding the target document at an early stage. However, since the degree of similarity is an evaluation of the entire document, no improvement means is provided for the operations of the above procedures 3 to 4.
The technique disclosed in Patent Document 3 is a search technique that focuses on the constituent elements of a document, but is intended for document reuse. Further improvement is necessary to use for the purpose of “searching documents” (including as many as possible).
If the technique disclosed in Patent Document 4 is used, the search condition is not an evaluation of the entire document, but a match determination is performed in units of paragraphs, etc., so that the search noise in the above procedure 1 can be reduced. The effect in the case of using is comparable to the technique of Patent Document 2.
[0010]
Therefore, the present invention has an object to provide a technical document search system capable of quickly and accurately searching for desired patent information / gazette without waste by introducing highly efficient document data search technology, database search technology, and the like. .
[0011]
In order to achieve the above object, the invention described in claim 1 is electronically stored in the storage means. , Consisting of multiple unit data A computer system for searching technical literature data, the search target technology For each of a plurality of components constituting Data representing the component Accept input Component input means, and each Search conditions for each component Accept input Search condition input means, and each Component Every, Based on the search criteria Above Search means for searching technical document data, unit data extracting means for extracting unit data that matches the search condition from the searched technical document data, and for each searched technical document, the technical document is Component array data indicating whether or not the search condition for each component is satisfied; and Matched the search conditions Associating the unit data with the identification data of the technical document Recording means for recording And.
[0012]
According to a second aspect of the present invention, there is provided the technical document data search system according to the first aspect. Accept input It is characterized by comprising an input means and a technical composition decomposition means for analyzing and dividing the technical text data into a plurality of text data and delivering the divided text data to the component input means.
Further, the invention according to claim 3 is the technical document search system according to claim 1, wherein dictionary storage means for storing dictionary data of terms and the constituent elements with reference to the dictionary data. Represents Technical term extraction means is provided for extracting technical terms from the data and delivering the extracted technical terms to the search condition input means.
Further, the invention according to claim 4 is the technical document search system according to claim 1, similar word storage means for storing a plurality of sets of similar words, By search condition input means input Accepted For each keyword included in the search condition, a similar word corresponding to the keyword is acquired from the similar word storage unit, and acquired. Shi Similar search terms Expand A similar word adding means is provided.
Further, the invention according to claim 5 is the technical document search system according to claim 1, in which the fitness condition that defines the condition of the search fitness is defined. Accept input Relevance condition input means and each retrieved technical document Above A search suitability calculating means for calculating a search suitability, wherein the unit data extracting means performs a process of extracting unit data for technical literature data whose search suitability matches the suitability condition. It is said.
The invention according to claim 6 is the technical document search system according to claim 1, wherein the attribute of the technical document data to be searched by the search means Accept input It has a target condition input means, and the search means searches only technical document data that matches the attribute.
The invention according to claim 7 is characterized in that the search means performs a search based on the search condition for each predetermined range in the technical document data.
The invention according to claim 8 is characterized in that the search means does not search for data belonging to a predetermined chapter when the technical document data is chaptered.
[0013]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described with reference to the drawings.
[First Embodiment]
FIG. 1 is a block diagram of a technical literature search system according to the first embodiment of the present invention.
The present embodiment is a software system operable on a personal computer (hereinafter “PC”) or a workstation. Since a commercially available PC or the like can be used, the description of the hardware configuration (not shown) will be briefly given as follows.
[0014]
The search condition input unit 2 in FIG. 1 is a program for displaying a search condition input screen, a display device such as a display, and an input device such as a mouse and a keyboard. The document data 1 is stored in a storage device such as an HDD. The search processing unit 3 is a program that performs search processing based on the search condition input by the search condition input unit 2. The hit table 4 and the extracted data list 5 are output as search results by the search processing unit 3 and stored in a secondary storage device such as a main memory or HDD.
The search result output unit 6 is a display device such as a program for displaying the search result and a display. The search result display screen is operated by an input device such as a mouse or a keyboard.
[0015]
First, input of search conditions will be described with reference to FIG.
FIG. 4 shows an example of a search condition input screen displayed on the computer display.
The column “IPC” in FIG. 4 is used for filtering document data to be searched, and will be described below as “target conditions”. The target condition may be a document classification code different from IPC (International Patent Classification), a publication date of the document, other bibliographic items, or a combination of keywords for searching the document data as a whole. .
[0016]
Text data representing each component of the search target technology is entered in the “component requirement” column.
In the “search condition” column, a search expression corresponding to each component is entered.
The “important” column is used for designating components to be particularly emphasized, and can be switched ON / OFF by clicking with a mouse or the like.
“Search location” designates a location on the storage device where the document data is stored. The directory may be input directly from the keyboard, or by clicking a “reference” button with a mouse, a directory list (not shown) can be selected by GUI operation.
In the “hit condition” column, a threshold value for determining whether to output each retrieved document data as a search result is input.
The column “number of extraction locations” defines how many unit data corresponding to each component are searched or extracted per document.
The “essential check element is required” column represents that the search condition must be satisfied for the configuration requirement in which the “important” column is checked. That is, document data that does not satisfy the search condition of the component requirement B is not output as a search result.
When the operator inputs each of the above using the keyboard or mouse and gives an instruction using the “Search” button, the search process is started.
[0017]
Next, the search process will be described with reference to the flow of FIG.
In starting the search process, first, a component array is generated (S100).
Here, the component array is an array for storing whether or not a search condition corresponding to each component (“component” on the input screen shown in FIG. 4) is satisfied. In this embodiment, it is an integer array having the number of array elements, and the subscript indicating the first array element is 1.
Subsequently, an extracted data list is generated (S101).
As shown in FIG. 3, the extracted data list is an area for storing text data that matches the search condition for each component. Each element of the list corresponds to each component, and each element of the list is configured to be able to store a list of text data (character string data) that matches the search condition.
[0018]
Next, it is determined whether or not the document data to be searched remains (S102). If the determination is affirmative, the hit table is rearranged according to the fitness level and output (S103). If the document data is not over, the document data is continuously read (S104), and it is determined whether the target condition is met (S105). For example, in the case of patent document data, since bibliographic data is included in the document data, the determination can be made by comparing the bibliographic data with the input target condition.
If S105 is affirmative, the process proceeds to S106 to initialize (zero clear) the component array and the extracted data list.
[0019]
Next, parts other than the body part (for example, bibliographic data) are removed from the document data (S107), and text shaping is performed (S108).
The bibliographic item description part in the patent document, the citation / reference document description part in the general technical document, etc. are preferably not to be searched, so the process of removing such a part is performed in S107. In the case of document data such as a patent document in which tags for chapter structure and document structuring are defined, a portion to be removed can be easily determined according to the definition. Except for structured documents, citations and references are generally listed at the end of the document data, so the contents of the document data are inspected, and if there is a line consisting only of a character string such as "cited document" or "reference document" It is recommended to remove the lines after that line.
In the processing in S108, tags in a format such as a markup language are removed to form a normal sentence, and line breaks are rearranged so that one line of text data is obtained for each predetermined range of document data.
Here, the predetermined range may be character string data representing a sentence separated by punctuation marks (. Or.), Or character string data representing a paragraph composed of a plurality of sentences. The paragraph break can be determined by a line feed code such as CR (Carriage Return) or LF (Line Feed).
[0020]
In S109, a variable n for indicating the number of the component being searched is initialized to zero. In S110, the variable n is incremented in order to proceed to the search process for the next component.
In S111, it is determined whether or not the variable n is larger than the number of input components, that is, whether or not the search processing has been performed for all the components. If the determination is affirmative, the process proceeds to S112, and if the determination is negative, the process proceeds to S116.
[0021]
In the procedure of S116 to S119, the text data shaped in S108 is searched for in units of lines. In S116, a variable L for indicating the line number being searched is initialized to zero, and the variable L is incremented in S117. In S118, it is determined whether or not the value of the variable L exceeds the number of lines of the text data. If the determination is affirmative, the process returns to S110, and the search process for the next component is performed. If the determination is negative, in S119, it is determined whether or not the current row L matches the search condition N corresponding to the component n. If the determination in S119 is negative, the process returns to S117, and the search process for the next row is performed. If the determination is positive, the process proceeds to S120.
[0022]
In S120, the value of the component array [n] is incremented. The value of the component array [n] represents the number of rows corresponding to the component n (matching the search condition N).
Next, in S121, the contents of the row L are added to the extracted data list [n].
In S122, it is determined whether or not the value of the component array [n] has reached the value input in “Number of extraction locations” on the input screen shown in FIG. If it is affirmation determination, it will return to S110 and will move to the search process of the following component. If the determination is negative, the process returns to S117, and the search process for the next line is continued.
[0023]
As described above, when it is determined in S111 that the search process for all the constituent elements has been completed, the processes after S112 are performed. In S112, the component satisfaction rate is calculated.
To calculate the component satisfaction rate, refer to the component array,
Number of array elements with value> 0 ÷ total number of array elements × 100
That is,
Number of components with paragraphs (rows) that match the search condition ÷ number of components × 100
Is substantially the same.
Next, in S113, it is determined whether or not the calculated component satisfaction rate is equal to or greater than a predetermined value. Here, the predetermined value is, for example, a value input in the “hit condition” in FIG. If a negative determination is made, the document data as a whole does not match the search conditions, and the process returns to S102 to start the next document data search process.
[0024]
In S <b> 114, it is determined whether or not all the components with the “important” check in FIG. 4 are satisfied. this is,
Value of component array [m]> 0
(M indicates the position from the top of the component that is checked for “important”). If the determination is negative, as in S113, the process returns to S102 and the search process for the next document data is started. Note that if “important check is essential” is not selected in FIG. 4, the determination process of S114 is unnecessary, and the process directly proceeds to S115.
If the determination in S114 is affirmative, the document data matches the search condition, and the information of the document data is added to the hit table in S115.
[0025]
FIG. 5 is an example of a hit table, where “document identification number” is a publication number or the like, and “score” is a number that represents a search suitability, a satisfaction rate, and the like. The “score” may use the component satisfaction rate as it is, or may be calculated according to other conditions. Also, a score may be taken in consideration of the number of paragraphs that match the search condition. Such a “score” may be calculated in S113, and the determination in S113 may be performed based on the score instead of the constituent requirement satisfaction rate.
Also, here, a new memory area for continuously storing the extracted paragraph data (or sentence data) is allocated, the contents of the extracted data list stored in the flow process are copied, and the allocated memory area is copied to the allocated memory area. The pointer is stored in “pointer to extracted data list” of the hit table.
When the search process for all documents is completed by the above procedure, the hit table is rearranged in descending order of the “score” value in S103.
[0026]
FIG. 6 is an example of a display screen on which search results are output based on the hit table.
The “reference number” in FIG. 6 corresponds to the “reference number” in the hit table.
The columns A to F correspond to the “component array” in the hit table.
Component array value> 0
“*” Is displayed to indicate that a paragraph corresponding to the component exists in the document. Instead of displaying “*”, the value of the component array itself may be displayed.
FIG. 6 further shows a state in which the column of the component 1 (in this case B) in the row 1 in the left table is selected by mouse click or the like. That is, on the right side, the sentence data corresponding to the component B (corresponding to “text data of constituent requirement 2” in FIG. 4), the search condition expression corresponding thereto, and the document in line 1 (JP XXXX- Of the text data extracted from (XXXXXX), text data corresponding to the component B is output. In this way, when a predetermined component column is selected in the left table, the corresponding data is displayed on the right with reference to the input data and hit table in FIG. Furthermore, among the extracted sentence data, words appearing in the search condition formula are highlighted.
Of the columns A to F shown in FIG. 6, the column corresponding to the component for which “important” in FIG. 4 is checked may be highlighted by bold or color coding.
[0027]
Further, in FIG. 6, a check box for selecting a document is displayed, and the operator can instruct to create a correspondence table between the selected document and the component (not shown). An example of the correspondence table is as shown in FIG. 7, in which the text data extracted from the selected document has a format corresponding to each component, and may be output to a printer such as a printer. Alternatively, it may be output as a document file in HTML format or the like.
In the screen of FIG. 6, for example, when the user double-clicks on the characters “JPXXXX-XXXXXXX”, the original text data of JPXXXX-XXXXXXX is extracted from the file based on the value of the “file name” in the hit table. It is read and displayed on the display (not shown).
In this way, the operator can easily confirm whether or not the target description exists in each document.
[0028]
[Modification]
In the above, after the literature data stored in the HDD in advance is filtered according to the target condition, the search processing for each component is performed on the literature data that matches the target condition. be able to.
After the input is completed on the input screen shown in FIG. 4 and the operator is instructed to execute a search, the query is connected to an external document database server via a communication line, and the database is searched for the search according to the target condition. May be issued, the search result document data may be downloaded to a designated “search location”, and the search process for each component may be performed on the downloaded document data. In this case, the process of S105 is unnecessary.
[0029]
[Application example]
This embodiment can be further applied as follows.
[Application Example 1]
In FIG. 4, the text data of the constituent requirements is input in a frame corresponding to each constituent requirement, but a series of text data input by a predetermined method is analyzed and automatically assigned to each constituent element. And may be set in each component requirement input field.
Specifically, character string data, which is a series of text data, is separated by punctuation marks (symbols such as “,” and “,”), and each divided part is interpreted as one constituent element.
Further, when the number of characters in each part is equal to or less than a predetermined number, correction may be performed by connecting to the next part to form one component.
[0030]
[Application 2]
It is possible to prepare dictionary data in which technical terms are associated with parts of speech, extract technical terms existing in the dictionary from partial sentence data as constituent elements, and set them in the search condition input field. Further, by preparing general term dictionary data and removing technical terms included in the general term dictionary from the extracted technical term group, phrases that are not directly related to technical features (for example, “means” “step” "Method", "apparatus", etc.) may not be set in the search condition input field.
As for the delivery method to the search condition input field, the extracted technical terms are simply enumerated and set in the search condition input field, and as a result, each term is searched by logical product (AND combination) or logical sum (OR connection). It may be made to do.
As another delivery method, the extracted technical terms may be divided into a noun group and a non-noun group, and the inside of each group may be an OR connection, and each group may be an AND connection.
[0031]
[Application Example 3]
For each search word set in the search condition input field, similar word dictionary data is further prepared. For the terms registered in the similar word dictionary, a search condition in which the search word and the corresponding similar word are OR-coupled. It may be extended to an expression.
As an example, in the search condition input field,
Structure * element * search
Are searched for in the similar word dictionary according to instructions from the operator to expand similar words, and the acquired similar words are used.
(Configuration + Structure) * (Element + Element) * (Search + Detection + Inspection)
The search expression is expanded like this.
In this example, “structure” is stored as a similar word corresponding to “configuration” in the similar word dictionary, and similarly, “element” for “element”, “detection” for “search”, and “Inspection” is stored.
[0032]
In the first to third application examples, after the text data is input, the process may be automatically executed without the operator's instruction, and the search process may be started.
Alternatively, the description style of the text data may be classified into a plurality of types, and text data decomposition / term extraction processing corresponding to the type selected by the operator from the types may be selected and executed.
[0033]
[Second Embodiment]
The second embodiment is operated by a client / server computer system (not shown). The client computer inputs search conditions and transmits them to the server. The server computer uses the search conditions received from the client. Based on the search process, the search result is transmitted to the client.
[0034]
As in the first embodiment, the client inputs a search condition using an input screen shown in FIG. 4 using a general PC. Then, the input content is transmitted to the server via the communication line.
In the first embodiment, search processing is performed for each document data one by one. However, in the second embodiment, a document database having a general configuration is constructed on a server. It differs greatly from the first embodiment in that it is used effectively.
[0035]
FIG. 8 is a block diagram of a search system on a server according to the second embodiment.
In the storage device on the server, the original text database 10 storing the reference data, the bibliographic index 11 in which the reference identification number and the bibliographic data corresponding to the original sentence database 10 are associated, the reference identification number and the index for full text search Is stored in the full-text index 12.
[0036]
In the server, after receiving the search condition from the client, the following processing is performed.
First, a primary search is performed using the target conditions included in the search conditions. If the target condition is a bibliographic item such as IPC, the bibliographic search engine 13 searches the bibliographic index to obtain a list of corresponding document identification numbers (identification data list 1 (15)). When a keyword for full-text search is included in the target condition, the full-text search engine searches the full-text index from 14 and obtains a list of corresponding document identification numbers (identification data list 2 (16)).
[0037]
Next, the arithmetic processing unit 17 calculates the union of the identification data list 1 (15) and the identification data list 2 (16) to obtain the hit list 18.
As described above, the hit list 18 becomes a list of document identification numbers that match the target condition. The document data shown in the hit list 18 is read from the original text database 10 and the search result is obtained by performing a secondary search (search for each component) according to the flowchart of FIG. 2 as in the first embodiment. Obtainable.
[0038]
However, in this method, the number of document data to be processed becomes enormous and the processing load on the server may increase. Therefore, it is preferable to create the final hit list as follows.
After the hit list is obtained, the full text index 12 is searched with the search condition n for each component, and the identification number list 21 ₁ ~ 21 _n (Instead of the identification data list 2 (16)) is output.
The hit list and the identification number list 21 ₁ ~ 21 _n The union of the document identification numbers included in is taken as the final hit list.
In this way, the target documents for the secondary search process can be narrowed down to the minimum necessary.
Then, the document data shown in the final hit list is read from the original text database 10 and the search result is obtained by performing the secondary search (search for each component) according to the flowchart of FIG. 2 as in the first embodiment. Can be obtained. The server sends the hit table and the extracted data list (corresponding to hit table 4 and extracted data list 5 in FIG. 1) obtained as a search result to the client, and the client performs display output shown in FIG. 6 based on this. Can do.
[0039]
At the stage of calculating the final hit list described above, the component satisfaction rate may be calculated, and only those that meet the specified conditions may be output to the final hit list.
In particular,
1. Number of components × specified component satisfaction rate (hit condition) / 100 = required number K (rounded down)
2. Concatenate each hit list, rearrange them in the order of identification numbers, and inspect from the top of the list and output the identification number to the final hit list if there are K or more consecutive identification numbers
That's it. In this way, processing such as “calculation of the component requirement satisfaction rate” (S112) “is the satisfaction rate equal to or greater than a predetermined value” (S113) in the flow of FIG. 2 is not necessary.
[0040]
At the same time, it may be determined whether or not the essential elements are satisfied, and only those that meet the specified condition may be output to the final hit list. In particular,
1. Identification number list 2 corresponding to a constituent element designated as an essential element (checked with “important” in FIG. 4) _n Take the union of (X)
2. Identification number list 2 corresponding to constituent elements other than those designated as essential elements _n (Y)
3. Number of components × specified component satisfaction rate ÷ 100−number of essential elements = required number k (rounded down)
4. Check from the beginning of (Y),
a, there are k or more same numbers
b, the number is in (X)
If both are positive, the final hit list is output.
By doing so, the processing of “all necessary elements are satisfied?” (S114) in the flow of FIG. 2 becomes unnecessary.
[0041]
[Third Embodiment]
The third embodiment is also operated in the form of a client / server.
The difference from the second embodiment is that the full-text search engine has a proximity calculation (neighbor search) function.
Also in the third embodiment, first, a primary search process similar to that in the second embodiment is performed.
Next, using the full-text search engine, a neighborhood search using a search condition formula for each component is performed. As a result of this search processing, data as shown in FIG. 10 is obtained.
[0042]
FIG. 10A is obtained as a result of searching for one component, and a range: position 1 to position m in the document data that matches the search condition formula is obtained.
When the neighborhood search for all the components is completed, the search results in FIG. 10A are obtained by the number of components, and these are integrated as shown in FIG. Specifically, the union of the document identification numbers is taken, and the position information in FIG. 10A is copied to the corresponding area of the component element position information in FIG.
Then, the product of the set of document identification numbers included in FIG. 10B and the set of document identification numbers included in the hit list as a result of the primary search is obtained (AND operation), and the final hit list (FIG. 10 (b)).
[0043]
A process according to the flow of FIG. 9 is performed on each document included in the final hit list.
The flow procedure of FIG. 9 is a modification of the flow procedure of FIG. A significant difference from the procedure of FIG. 2 is that it is not necessary to determine whether or not the search condition is matched for each row. This is because the position of the unit data corresponding to the constituent element has already been extracted from the final hit list (FIG. 10B).
In FIG. 9, the processes in the procedure given the same step numbers as those in FIG. 2 are the same as those described above with reference to FIG. 2, and are therefore described in the first embodiment. See description.
[0044]
In S123, the variable m is initialized, and in S124, the variable m is incremented. The variable m is used as an index value for determining the positions 1 to m in the data of FIG.
In S125, information on the position m is obtained from the plurality of pieces of position information corresponding to the component n.
In S126, text data in a range indicated by the position information acquired in S125 is extracted from the literature data. The subsequent procedure is the same as in FIG.
[0045]
Thus, according to the present invention, an efficient patent information search system that captures literature data as a set of components such as “sentence” and “paragraph” can be constructed, and the invention is not limited to patent information database search technology, By efficiently customizing and generating unit data structures such as technical words and components, it can also be applied to technical literature search systems for other technical literature (eg, ISO, various research institutes, university data databases, etc.) Is possible.
[0046]
【The invention's effect】
As described above, according to the present invention, there is a literature retrieval system for retrieving digitized technical literature data, and the technical literature data is retrieved by inputting a search condition for each constituent element together with the technical constituent element data. The unit data that matches the search condition is extracted, and the component data, the unit data, and the identification data of the technical data corresponding to the unit data are output in association with each other. When searching the patent information database for the purpose or when applied to the search of other technical information, the user's processing is simplified, and the desired document identification number can be searched quickly and accurately.
[Brief description of the drawings]
FIG. 1 is a block diagram of a technical literature search system according to a first embodiment of the present invention.
FIG. 2 is a flowchart of processing of the system shown in FIG.
FIG. 3 is a diagram showing a structure of an extracted data list.
FIG. 4 is a diagram illustrating an example of a search input screen.
FIG. 5 is a diagram illustrating an example of a hit table.
FIG. 6 is a diagram illustrating an example of a search result output screen.
FIG. 7 is a diagram illustrating an output example in which search conditions are associated with search results.
FIG. 8 is a block diagram of a technical literature search system according to a second embodiment of the present invention.
FIG. 9 is a flowchart of processing of a technical literature search system according to a third embodiment of the present invention.
FIGS. 10A and 10B are diagrams showing an example of the output of neighborhood search used in the third embodiment, where FIG. 10A shows a search result for one component, and FIG. 10B is a list in which the components are integrated; Is shown.
[Explanation of symbols]
1 Literature data
2 Search condition input part
3 Search processing section
4 Hit table
5 Extracted data list
6 Search result output section
10 Original database
11 Bibliographic index
12 Full-text index
13 Bibliographic search engine
14 Full-text search engine
15 Identification data list 1
16 Identification data list 2
17 Arithmetic processing part
18 hit list

Claims (8)

A computer system for searching technical literature data composed of a plurality of unit data , which is electronically stored in a storage means,
For each of a plurality of components constituting the search techniques, the components input means for accepting an input of data representing the respective components,
The search condition input means for accepting an input of a search condition for each component,
Wherein each component, a searching means for searching the technical literature data on the basis of the search condition,
Unit data extraction means for extracting unit data that matches the search condition from the searched technical literature data;
For each searched technical document, component element array data indicating whether the technical document satisfies a search condition for each component, the unit data that matches the search condition, and identification data of the technical document A technical literature search system comprising: a recording means for recording in association with each other. The technical literature search system according to claim 1,
Input means for receiving input of technical text data;
A technical literature search system comprising: a technical composition decomposing unit that analyzes and divides the technical text data into a plurality of text data and delivers the divided text data to the component input unit. The technical literature search system according to claim 1,
Dictionary storage means for storing dictionary data of terms;
A technical literature search system comprising technical term extraction means for extracting technical terms from data representing the constituent elements with reference to the dictionary data and delivering the extracted technical terms to the search condition input means. The technical literature search system according to claim 1,
Similar word storage means for storing a plurality of sets of similar words;
Each keyword included in the input reception we were by the retrieval condition input unit search condition, a similar word corresponding to the keyword acquired from the similar word storing means, and expanding the obtained similar words to the search criteria A technical literature search system comprising a similar word addition means for adding. The technical literature search system according to claim 1,
It has a fitness condition input means for receiving input of fitness criteria that define the search fit condition, and a search fitness calculating means for calculating the search adaptability for each retrieved technical literature,
The technical document search system, wherein the unit data extraction means performs a process of extracting unit data on technical document data whose search suitability matches the matching condition. The technical literature search system according to claim 1,
The search means has target condition input means for receiving input of attributes of technical literature data to be searched,
The technical document search system, wherein the search means searches only technical document data that matches the attribute. The technical document search system according to claim 1, wherein the search means performs a search based on the search condition for each predetermined range in the technical document data. 2. The technical document search system according to claim 1, wherein the search means does not search data belonging to a predetermined chapter when the technical document data is chaptered.

Images