JP2002063192A - Patent document system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To extract a significant part related with an invention in patent document. SOLUTION: The 'industrial application field' of an invention is explained as an example of a significant part. The texts of the description of a patent are read sentence by sentence from a database 200, in which the text data of the description of a patent are stored (S210), and whether the sentence has a sentence pattern, from which it is possible to extract the 'industrial application field' or the like is checked by using an extraction table (S250). When the sentence matches with the sentence parttern, from which it is possible to extract the >=ndustrial applicative field' (S240: 'YES'), words indicating the 'industrial application field' or the invention itself are extracted, and outputted and stored in an extracted data storage part 100 as data indicating the 'industrial applicative field' or the like.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a technology for searching documents, and more particularly to a search for patent documents.

[0002]

[Technical Background] In the field of information search, a natural language search method in which the entire text is directly searched without processing documents, and a database that performs synonymous processing of keywords (including unification processing of notation). And a keyword search method. As a keyword search method,
Create a short sentence summarizing the essence of the literature by using commonly used specific terms (for example, in the JAPIO Patent Literature Database, a JAPIO abstract summarizing the invention at the subject level applies), and enter keywords from this short sentence. An extracted and databased keyword search method is often used. In the search using such a keyword, it is necessary to create a short sentence based on a specific term. However, since the specific term is used as a keyword, a search with less noise and omission is possible by unifying the synonym processing levels. It has been. The natural language search method does not require the processing of documents, but uses the expression form that is not controlled by the text creator. For this reason, there are expressions having different concept levels, abstract expressions, and the like, and there are many leaks and noises, and it is impossible to perform a highly accurate search.

[0003] By the way, patent information research in corporations has become more important due to the development of the pro-patent policy. Compared with a database centering on abstracts, which summarizes conventional inventions at the subject level, it is more detailed. Searching for technical information or rights information is required. To meet such needs, a full-text search system using a patent CD-ROM as a database has been provided.

Now, the specification of a patent is described in comparison with the prior art in order to clearly emphasize the invention,
Alternatively, in order to clarify the position of the invention in the field of use, a detailed technical description of the field of use is also given, and so there are many technical descriptions other than the invention. For this reason, there is a problem that a search in which the entire specification is used as a search area increases noise, and an appropriate correct answer cannot be obtained. It is also possible to specify a claim that describes all matters deemed necessary to specify the invention and to search only that area. ,
Usually, in order to secure a broad scope of claims, the claims often deliberately avoid using specific terms, and use high-level concepts, abstractions / functionalities, and creative expressions. Therefore, when searching, it is necessary to formulate a search formula by sufficiently considering such expressions such as high-level concepts, but in practice, it is virtually impossible to identify the intention of the specification creator and create the search formula. Impossible. Therefore, it is said that the retrieving rate (= the number of correct answer documents hit / the number of all correct answer documents in the database) is extremely low in the search of the claims.

[0005]

SUMMARY OF THE INVENTION The first object of the present invention is to provide an important part which is most closely related to the invention described in the patent specification (hereinafter referred to as "an important part of the invention").
To automatically extract only). A second object of the present invention is to provide a patent information search system in which an important part of the invention is set as a search target, and only important parts of the invention in hit patent documents are identified and displayed. The third object of the present invention is to use the general concept, abstract / functional, creative expression and specific technology described in the detailed description of the invention as used in the claims in the patent specification. It is an object of the present invention to provide a technology for automatically creating a related term dictionary in which the relation between terms is clarified. A fourth object of the present invention is to provide a high-level conceptual expression, an abstract / functional,
It is an object of the present invention to provide a system that can obtain information on a creative expression and search for the invention described in the claims.

[0006]

In order to achieve the above object, the present invention provides a patent document system, comprising: a database having at least full text data of a specification;
Means for reading the specification from the database, means for extracting the important part in the read specification using items used in the specification, and means for searching the extracted important part in the database Means for writing data as described above. As the important part,
A means for extracting the "means for solving the problem"; the means for extracting the "means for solving the problem" includes a means for recognizing an invention unit for each related claim; Can be extracted for each invention unit. Means for extracting the "solution of the problem", means for directly creating a query from the claims of the specification, and means for performing a similarity search for the specification for each section using the created query Means for extracting the result of the similarity search, and a similarity search can be performed in the specification to extract a relevant portion for each section. Position information in the document is added to the extracted important part, and the position in the document can be displayed by a search response to the important part. Further, in the patent document system of the present invention, at least from a database in which a general technical term and a compound word within the scope of claims are indexed, a means for searching for a document using the index of the database, and a search result document, Means for displaying and selecting the compound word, and a new search can be performed from the compound word displayed and selected from the document of the search result. In addition, a database having at least full text data of the specification, means for reading the claims of the specification from the database, and association detecting means for recognizing a pattern for associating upper and lower levels from the read claims And means for accumulating the upper-level expression and the lower-level expression in association with each other based on the associated hierarchical relationship, so that a related term dictionary can be created and the dictionary can be used for search.

[0007]

Embodiments of the present invention will be described in detail with reference to the drawings. <Creation of Important Part Data of the Invention> FIGS. 1 to 8 show examples of reference specifications attached as appendices to "How to Write Specifications" prepared by the Patent Office as examples of patent specifications. Mainly using this, data creation of an important part of the invention will be described below. The important part of the invention referred to here means “industrial application field”, “object of the invention”, and “means for solving the problem”. Now, as shown in FIGS. 1 to 8, the patent specification 100 includes [Title of Invention] 110, [Claims] 12
0, [Detailed description of the invention] 140, [Brief description of the drawings] 150, [Description of reference numerals] 152, and headings such as [Drawing] (not shown) are described separately. ing.

When searching for information relating to the invention, it is important to identify the invention from the viewpoints of "industrial application field", "problem of the invention" and "means for solving the problem" and create an index. It is said that there is. By the way, in many cases, [Title of Invention] 110, [Claims] 120 and [Explanation of Signs] 152 of the patent specification contain brief information on the invention and information that is a clue for extracting the invention part. Have been. However, the description in [Detailed Description of the Invention] 140 is described in comparison with the prior art in order to clearly show the invention, or is described in detail in the field of use in order to clarify the position of the invention in the field of use. From the viewpoint of retrieving information on the invention, for example, in conjunction with various technical explanations, it is extremely redundant.

Therefore, referring to the contents of the articles described in [Title of Invention] 110, [Claims] 120 and [Explanation of References] 152, [Detailed Description of the Invention] 1
Headings [Technical field to which the invention belongs] 141, [Problems to be solved by the invention] 143,
[Means for Solving the Problems] 144, [Embodiment of the Invention] 145 and [Example] From the contents of articles delimited by 146, "industrial application fields" and "problems of the invention" that specify the invention. A method for extracting “means for solving problems” by category is described below.

<About "Industrial Application Field">
In the patent specification, a heading "Technical field to which the invention belongs" is provided, and when an article relating to "industrial application field" is described together with the main body of the invention in the part described below this heading There are many. In some cases, the exact expression of the field of use is present in other headings. A case where only “industrial application fields” are extracted from such a case will be described with reference to an example shown in FIG. FIG.
In the case of the description example 210 of FIG.
As shown at 211 to 215, "The present invention relates to ++ used in (*) **** in ****".
++ ".
In this case, the “technical field” is the “sizing device in machine tool” 212, and the “heat influence prevention device” 214 is
FIG. 4 shows an invention main body that specifies the invention in more detail in “industrial application fields”. As can be seen from this example, such a sentence has a feature in the document structure in which “industrial application field” is described in a portion indicated by *** (see 212). With such structural features as clues,
"Machine tool sizing device" which is an "industrial application field" is extracted.

The sentence pattern described above is an example of a clue, and is actually used in industrial applications such as "The present invention can exert an effect of +++++ when applied to a sizing device of a machine tool". There are various expressions.
By using a feature extraction table in which such structural features are patterned, it is possible to extract a portion that simply represents “industrial application field”. The description example 220 is an example of another pattern, and analyzes a sentence structure and the like using parallel conjunctions 226 and 231 and the like. This allows
Recognizing that “related” 225 and 230 are abbreviated and the like, “removing basket that is rotationally driven by the drive motor” 2
Reference numeral 22 denotes "industrial application field" (technical field), and "centrifugal dehydrator" 224, "balance ring" 229, and "manufacturing method" 232 are output as being the main body of the invention.

The extraction process of the "industrial application field" and the main body of the invention will be described with reference to the flowchart in FIG.
In FIG. 10, the sentences of the specification are read out one by one from a database 200 storing text data such as a patent specification (S210), and a feature extraction table in which the structural features of the sentence pattern are patterned is obtained. It is checked whether the sentence pattern can be used to extract “industrial application field” or the like (S240). When the sentence pattern matches the extractable sentence pattern (YES in S240), the words indicating the "industrial application field" or the invention itself are extracted from the matched sentence pattern, and the extracted data is extracted as data indicating the industrial use field or the like. The data is output and stored in the extracted data storage unit 100 in a searchable format (with a key such as a document number) (S250). This operation is repeated until the target patent specification ends (S
220). In the extracted data storage unit 100, extracted data 110 and 120 are stored for each extracted patent specification, for example, using a document number as a key, as shown in the upper right of FIG. You. Position information in the document is also added to the extracted data and stored. In the patent document system of the present invention, this extracted data becomes a search target area for “industrial application field”, and when displaying the document after the search, the corresponding portion of the displayed document is highlighted by the position information. Will be done.

[0013] Incidentally, in the [Title of the Invention], for example, even in the above-mentioned example, there are many cases where terms related to "industrial application fields" such as "machine tool" or "sizing device" exist.
It is also possible to confirm the data by checking it with the information in [Title of Invention] to ensure the reliability of the data. Further, in reality, it is more preferable to incorporate a work part for manually performing exception processing for irregular expression and to improve the accuracy of the extracted data and the feature extraction table. It is desirable that the exception processing is similarly provided in the extraction of the “problem of the invention” and the “means for solving the problem” described below.

<Regarding the "Problem of the Invention"> The "Problem of the Invention" is described in the portion described below under the heading "Problem to be Solved by the Invention" similarly to the above-mentioned "Industrial Application Field". It is shown. However, usually, in order to clarify and highlight the “object of the invention”, the problems of the related art and the solutions thereof are described together, and the sentences are often long. Since the prior art, problems, and the like cause a reduction in the accuracy of the search described later, that is, the precision (the number of correct documents / (the number of noise documents + the number of correct documents)), only the part directly related to the invention is extracted. It is necessary to do. Looking at the description shown in [Problem to be Solved by the Invention] 143 in FIG. 2, paragraph [0003] describes a problem of the prior art, and paragraph [0004] describes the “problem of the invention”. Is described.

[0015] A specific article of the "object of the invention" is, for example, as described above, "The object of the present invention is to obtain ****, and further to provide ****""There is a feature in the document structure described between" the present invention "and" the object ". Also in this case, it is possible to extract only the portion describing the "object of the invention" by using the feature in document expression as in this example as a clue. This is just an example.
Other expressions may be used in the description, such as "The object of the present invention is to provide ****." Therefore, in actual extraction, extraction is performed using a feature extraction table in which features in the document structure are patterned and collected and a headline of the specification as a clue. The extracted data also accumulates position information in the extracted document and accumulates the extracted data.

<Regarding "Means for Solving the Problem"> The result of solving the problem is an invention, and matters deemed necessary for specifying the invention are described in the claims for each claim. However, in order to broaden the scope of the claims, the description of the claims often intentionally avoids the use of specific technical terms, and uses high-level concepts, abstractions, and functional expressions. Therefore, if a claim is directly searched by a keyword using specific technical terms, omissions increase and a sufficient recall cannot be obtained. Therefore, instead of the claims, articles corresponding to the invention described in the claims and described in specific technical terms are extracted from [Embodiments of the invention] or [Examples], and parts thereof are extracted. As a search target, it is possible to perform a search using specific technical terms that anyone can easily come up with, thereby enabling a search with little omission, that is, a high recall rate. There are various methods for extracting the article part most closely related to the claim, but the claim is used as a query, and the parts separated by paragraph numbers in the specification (hereinafter referred to as "paragraph parts") are defined as one document. In the same manner, similar document search is performed, the paragraphs are arranged in the order of similarity, the number of characters, the similarity, etc. are referred to, and the paragraph is used as a search target area. There is a method of using a sentence delimitation information composed of words having a high co-occurrence with a keyword (described later in a method for determining a passage) and a delimitation information by paragraph number to adopt a part most closely related to the claim.

The claims are described for each claim as shown in FIG. 1 and FIG. 2, and the claims include those described in an independent form and those described in a cited form.
When considered in light of the first object of the present invention, an independent claim and all claims that refer to the independent claim (including, of course, the relationship citing the cited matter) form one unit of invention. It is effective to treat as Therefore,
A query is created from an independent claim and all the claims that refer to the independent claim (hereinafter referred to as related claims). The query is created by excluding unnecessary words and terminating the related claims in terms of terms using ordinary morphological analysis techniques, and finally eliminating duplicate terms. Also,
In general, an independent claim is expressed as a general concept, abstract, functional, or creative, and a quotation form claim that refers to an independent claim is a form that more specifically embodies the constituent elements of the independent claim or precedes it. Quoted form claims are often described in such a way as to replace the specifics of the invention. Therefore, by creating a query based on the related claims, the concreteness of the broader conceptual, abstract, and functional expressions can be somewhat improved. However, it is possible to employ a paragraph portion more accurately corresponding to the related claim as a search area.

First, determination of a related claim (unit of invention) will be described with reference to FIG. FIG. 11 shows an example illustrating the dependency of the claims described in [Claims] described in the specification shown in FIGS. 1 and 2. As shown, the invention unit 310 is described in the independent claim 311 of claim 1 and the dependent claims 312 to 316 of claims 2 to 6 are described with reference to claim 1 and the like. It can be recognized by using. This allows
Claims 1 to 6 are treated as one related claim.
Further, the invention unit 320 is the independent claim 321 of claim 7
The dependent claims 322, 323 of claims 8 to 12
Can be recognized by being described in a format citing claim 7 or claim 8.

Regarding the process of recognizing the related claims,
This will be described with reference to the flowchart of FIG. In FIG. 12, shown on the right shoulder is a related claims table showing the result of processing the example of the specification shown in FIGS. Claims are read one by one (one sentence) from the patent document database 200 from [Claims] of the document to be processed (S410). It is checked whether the words of the character strings “claim” and “description” exist in the sentence of this claim (S44).
0). If this word does not exist (No in S440), it is an independent claim, so this independent claim is stored as a key in the related claim table (see 310, 320 in the upper right of FIG. 12). This is the case of claim 1 in the related claim table 310 and claim 7 in the related claim table 320, for example. By the way, in the case of a dependent claim, a number between "claim" and "description" is analyzed to obtain a higher-level claim (S450). The obtained upper-level claim is determined (S470) in the same manner as the processing (S440) of determining whether the claim is a dependent claim or not is a previous independent claim. If the higher-level claim is an independent claim (No in S470),
The claim to be processed is stored in an item of the related claim table, with the higher-level claim as a key. This is, for example,
In the example of FIG. 1, the claims are [Claim 2] and [Claim 8], and since these upper claims are independent claims of [Claim 1] and [Claim 7], the related claim table 310 And 320
[Claim 2] and [Claim 8] are the [Claim 1]
And [Claim 7] are stored in the item having the key as a key. If the higher-level claim is a dependent claim (Yes in S470),
The claim to be processed is stored in the related claim table in an item whose key is a higher-order claim. This is, for example,
In the example of FIG. 1, [Claim 3] and [Claim 9], and these higher claims are [Claim 2] and [Claim 8]. [Claim 3] and [Claim 9] are stored in the items of which [Claim 1] and [Claim 7] are keyed by [Claim 2] and [Claim 8]. In this way, the processing is performed until the claim (sentence) within the scope of the claim of the document is completed (Yes in S420). When the processing is completed, the related claim table is arranged, the duplicated claims are deleted, and the process is terminated.

Next, the extraction of "means for solving the problem" data for each invention unit will be described. For each unit of invention,
The important matter data will be created from the related claims.
For this reason, first, the solution to the problem data is created by excluding unnecessary words in terms of terms using an ordinary morphological analysis technique to eliminate unnecessary words, and finally eliminating duplicate terms. Creating data for each invention unit as an example
This will be described with reference to the related claims 1 to 6, which are the invention units 310 shown in FIGS. First, for each request, a keyword is extracted by removing unnecessary words and duplication by morphological analysis. The results are shown below. [Claim 1] Dehydrating basket / swinging / prevention / centrifugal dehydrator [Claim 2] Swinging / prevention / dehydrating basket / top / mounting / balance ring / centrifugal dehydrator [Claim 3] Dehydrating basket / top / cross section U-shaped / balance ring / mounting part / circumferential / division / mounting / centrifugal dehydrator [Claim 4] dehydrating basket / top / outward / reverse U-shape / bending / balance ring / mounting part / mounting / centrifugal Dewatering device [Claim 5] Dehydrating basket / top / upper / opening / balance ring / mounting part / mounting / centrifugal dewatering device [Claim 6] Dehydrating basket / buffer / centrifugal dewatering device Keywords extracted from each claim Deduplicate and dehydration basket, swing, prevention, centrifugal dehydrator, upper end, mounting, balance ring, U-shaped cross section, mounting part, circumference, split, outward, inverted U-shaped, bent, upward, opening , Buffer means "
A query formula, ie, a query, is created using the query as a keyword. [Example] 1 of the document shown in FIGS.
If a similar document search is performed on the 46 paragraphs [0012]-[0017], each paragraph is answered with a ranking in the order of similarity. Based on this answer and considering the number of characters, the paragraph [0012] [0013] [0014] [0015] is adopted as the part most closely related to the related claim of the patent document, and is stored together with the position information. In the actual search, when the search is performed by designating the solution to the problem, the paragraph [0012] [0013] [0014] [001]
5] is the search target area, and in the document display,
The paragraph will be highlighted. Similarly, paragraphs are extracted from the related claims 7 to 12 which are the invention units 320 and are stored together with the position information.

Here, the “passage determination method” described in the preceding paragraph [0016] will be described. In a full-text search, the longer the sentence to be searched, the more topics, and the exact match search, which matches terms in the text, picks up a combination of terms between different topics and increases noise. Have been done. In order to improve the accuracy rate due to a long sentence, a method has been proposed in which only a topic part in a text corresponding to a question is searched as a search target and the topic part is displayed. This is a passage search. In the passage search, it is important to determine the area of the passage, that is, to divide the text. As a method of dividing the text, a text format such as a chapter, a section, or a paragraph of the text, a fixed number of words, division by an appearance word, a method using the appearance word and syntax information, and the like can be considered. For example, patent publications are divided into chapters,
Sections and paragraphs are separated, and chapters and sections are provided with headings. There is a method of determining a passage using these as document separation information. There is also a method of specifying an area using a period as document delimiter information. The division based on a certain number of words is, for example, a case where an important matter is written at the head (heading) of an article in a newspaper or the like in many cases. For the segmentation based on the appearance words, for example, words related to a bicycle, such as a pedal, a chain, a steering wheel, and a saddle, which have a high degree of relevance (referred to as having high co-occurrence), are often used. Similarly, words having a high degree of relevance, such as drive control, a battery, and a power supply, are often used for an electric motor. When words having high co-occurrence appear in a document, the area of the sentence is considered as a group having one meaning, and the area is regarded as a passage. In addition, in Exact Match search,
In response to the question of electric motors and bicycles, in documents that include electric motors and bicycles, there is a single meaning that includes words such as drive control, battery, and power, which have high co-occurrence with electric motors. There is a single meaningful group containing words such as highly probable pedals, chains, handles, and saddles. The document in which the document part in which it is contiguous or partially overlaps is the answer as the document most suitable for the query formula, and a group of document areas separated by punctuation marks of this document Is the part that can be presented as a passage describing the content of the question. The division based on the appearance words and the syntax information is a method of determining a passage based on syntax information, for example, document delimitation information by paragraphs, and clues such as “where” to change topics. Which of these methods is to be adopted depends on the format of the document to be searched, but in practice, a combination of the above is effective.

<Drawing> Drawings are not searchable,
For understanding the contents of the invention, it is meaningful to indicate portions that are deeply related to the invention using reference numerals in the drawings. The important sign of the drawing can be specified by the frequency of the sign of the drawing used in the passage and the presence or absence in the claims, but
Since drawings are usually image data, even if it is possible to specify important drawing symbols, it is practically impossible to automatically mark and highlight the drawing symbols even if it is theoretically possible. Therefore, it is preferable to display an important code near the drawing or to manually mark the drawing.

<Creation of a related term dictionary> Claims often include expressions such as a general concept, and some of them may be described by embodied general technical terms in cited claims. That has already been mentioned. As described above, in the claims, the claims are described separately for each claim. Independent claims are expressed in terms of the general concept, etc. It is sometimes seen that the claim is described in a more specific form. For example, claim 1 describes "a centrifugal dewatering device in which a dewatering basket (11) is provided with a means (13) for preventing swinging", and in claim 2,
The centrifugal dewatering device according to claim 1, wherein the means for preventing the swing is a balance ring (13) attached to the upper end of the dewatering basket (11). in this way,
Claim 1 states that the "means for preventing swing (13)" expressed in a general concept or the like is specifically "a balance ring (13) attached to the upper end". It has been described. Such a feature in the document structure related to the description of the claim, "**** of claim 1 in which **** is ********" makes it possible to recognize a relationship with an expression such as a conceptual concept. . Some specifications do not have enough clues. For example, [Claim 6] includes "the centrifugal dehydrator according to any one of claims 1 to 5 in which the dehydrating basket (31) is provided with a buffer means". There is no clue that can be specified as a shock absorbing wheel. In such a case, for example, since the word "buffer" is common, an exception processing means shown as a reference and manually determined is provided in the system. In this way, concrete expressions and high-level concepts,
A related term dictionary corresponding to functional and creative expressions can be created.

<Search> In the above, the technique for extracting the important parts of the invention from the specification has been mainly described. Hereinafter, the search for the important parts of the invention using these will be described. The data to be searched according to the present invention is, in addition to conventional “publication full-text data”, extracted as important part data (keywords and the like) in the above-described processing, “industrial application field data”. Location information "," invention subject data location information "," problem solving means data location information ",
“Code list data closely related to the invention” is added to each patent document. The search method according to the present invention is preferably a search method based on an exact match, such as a keyword search method in which keywords are extracted by ordinary morphological analysis to create an index, and a full-text search method using an N-gram method. But it does not rule out similar document searches. By using these, a search by category can be performed by designating "categories of important parts of the invention (for example," industrial application fields "," problems of the invention ", etc.)" as "searches". In addition, when displaying the search result, highlight display can be performed by using position information of an important part of the invention.

(Expansion Search of Claims) Independent claims are often expressed in a high-level concept, abstraction / functionality, and creative, because of the wide scope of rights, avoiding concrete expressions. As described above, it is not suitable for searching. However, in the search for patent information, it is significant if the inventions described in the claims can be directly searched. For this reason, if an expression such as a conceptual concept used in the claims is extracted and registered as an index, it is possible to directly search the claims. Now, as an index, a single word (for example,
Some use only automatic and vending machines, while others have compound words (eg, automatic, vending machines, vending machines). Compound words have the disadvantage of having a large number of indexes and a large memory, but have the advantage of low noise. The present invention goes further and further describes the abstraction used in the claims.
One of the tasks is to extract a functional expression (for example, means for preventing head swing) and use it for an index. The expression form of the concept such as “general concept” in the claims is, for example, “means (member) for preventing the dehydrating basket from swinging.
Provided "," provided with a prediction means (member) for predicting the disturbance before an excessive disturbance acts on the anti-vibration table ", or" provided that the floor is substantially the same as the other floor surface. It is described as "floating means (member) that floats without contacting the floor surface,""drive hand throw (member) that supports and rotates the workpiece," and these are described in "****". ********
** Provide means (members) "or" ********
** means (members) to perform patterning. Compound words, which are abstract / functional expressions, are extracted from these patterns and used for the index. However, even when a compound word that describes a high-level concept or the like is used as an index, the searcher usually uses a compound description that expresses a high-level concept or the like used in the claims described above. It is difficult to imagine words and create queries.

However, by using the extended search of the present invention, the searcher can obtain a compound word (index term) which is a description expressing the indexed general concept of the invention.
Can be easily selected. Figure 1
This will be described with reference to the flowchart of FIG. FIG. 13 shows a case of searching for the technique of preventing the swing shown in FIGS. 1 to 8. As shown in FIG. 13, first, the searcher inputs a logical product 510 of a centrifugal dehydrator and a dehydrating basket, which is a keyword of general technical terms directly related to the present invention, to a search system as a query 1. By inputting (S61
0), a search is performed on the patent information database 200 to obtain a search answer 520 (S620). As the search answer 520, a considerable number of patent documents can be obtained.
In the present invention, the searcher can use the search answer 520 as a general technical term (ordinary keyword) 530 within the scope of the claims of the patent document and the above-mentioned index phrase 540 by using, for example, a RANK function. It is possible to display in descending order (S630). From the displayed items, the searcher selects the item closest to the item to be searched, creates a search expression 550 including the index word 540 again, and uses this search expression 550 to set the search portion to [Claims ], A search is performed on the patent information database 200 (S640). And this search formula 55
A search answer 560 for 0 is obtained (S650). As described above, since the claims can be searched using the compound word itself that expresses a general concept that defines the invention, it is possible to perform a search without omission for the invention. it can.

Further, the use of the related term dictionary makes it possible to easily perform a "claims extended search". That is,
In the related term dictionary, general technical terms, which are specific expressions, and expressions such as broader concepts used in the claims are stored in association with each other. For this reason, for example, a query (search expression) can be created by designating a general technical term and displaying and selecting an expression associated with the general term in this dictionary. In addition, a query using general technical terms may be expanded by using the related term dictionary, and a search may be executed so that a keyword is expanded and searched using a synonym dictionary.

The present invention may be applied not only to a stand-alone computer system, but also to, for example, a client-server system composed of a plurality of systems. Further, the search may be executed by a search system including a large-scale host system and a terminal connected thereto. By reading the program from the storage medium storing the program related to the present invention by the system and executing the program, the configuration of the search system of the present invention can be realized. This recording medium includes a flexible disk, C
There are a D-ROM, a CD-R, a magnetic tape, and the like.

[0029]

According to the present invention, only important parts of the invention described in the claims can be automatically extracted from the patent specification. By using the extracted data, a highly accurate search can be performed.
According to the present invention, information on a high-level concept, an abstract / functional, and a creative expression of a claim is obtained from a search result of an important part of the invention, and the invention described in the claim is also searched. be able to.

[Brief description of the drawings]

FIG. 1 is a diagram showing an example of a specification for explaining the present invention.

FIG. 2 is a diagram showing an example of a specification for explaining the present invention.

FIG. 3 is a diagram showing an example of a specification for explaining the present invention.

FIG. 4 is a diagram showing an example of a specification for explaining the present invention.

FIG. 5 is a diagram showing an example of a specification for explaining the present invention.

FIG. 6 is a diagram showing an example of a specification for explaining the present invention.

FIG. 7 is a diagram showing an example of a specification for explaining the present invention.

FIG. 8 is a diagram showing an example of a specification for explaining the present invention.

FIG. 9 is a diagram for explaining extraction of an industrial application field.

FIG. 10 is a flowchart illustrating an extraction process of an industrial application field.

FIG. 11 is a diagram for explaining the relationship between related claims.

FIG. 12 is a flowchart illustrating a process for determining a related claim.

FIG. 13 is a flowchart illustrating a claim extended search process.

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Shizuo Shibata 4-7-1, Toyo, Koto-ku, Tokyo Sato Daiya Building Within the Japan Patent Information Organization (72) Inventor Masao Tanaka Toyo, Koto-ku, Tokyo 4-7-1-7 Sato Dai-ya Building Japan Patent Information Organization (72) Inventor Seichi Okada 4-7-1 7 Toyo, Koto-ku, Tokyo Sato Dai-ya Building Inside Japan Patent Information Organization ( 72) Inventor Shinichi Asami 4-7-1, Toyo, Koto-ku, Tokyo Sato Daiya Building F-term in the Japan Patent Information Organization (Reference) 5B075 ND20 NS10 PQ02 PQ22

Claims (6)

[Claims] 1. A patent document system, comprising: a database having at least full-text data of a specification; means for reading a specification from the database; and an important part in the read specification used in the specification. And a means for writing the extracted important part in the database in a searchable manner. 2. The patent document according to claim 1, further comprising: means for extracting “means for solving the problem” as the important part, wherein the means for extracting the means for solving the problem is a related claim. A patent document system including means for recognizing an invention unit for each term, and extracting the "means for solving the problem" for each invention unit. 3. The patent document according to claim 1, wherein the means for extracting the “means for solving the problem” includes: means for directly creating a query from the claims of the specification; and Means for performing a similarity search of the specification for each section, and means for extracting a result of the similarity search, performing a similarity search in the specification and extracting a relevant portion for each section. Patent document system featuring. 4. The patent document system according to claim 1, wherein position information in the document is added to the extracted important part, and the position in the document is determined by a search response to the important part. A patent document system characterized by being able to also display a document. 5. A patent document system comprising: a database in which general technical terms and compound words within the scope of claims are indexed; means for searching for documents using the index of the database; Means for displaying and selecting the compound word, wherein a new search can be performed from the compound word displayed and selected from the document of the search result. 6. In a patent document system, a database having at least full text data of the specification, means for reading out the claims of the specification from the database, and associating upper and lower levels from the read out claims An association detecting means for recognizing a pattern; and a means for associating a higher-level expression and a lower-level expression in association with each other based on the associated hierarchical relationship, and creating a related term dictionary and using the dictionary for search. Patent document system featuring.