JP4397264B2 - Technical literature marketability analysis system and marketability analysis program - Google Patents

Description

  The present invention relates to a marketability analysis technique for technical documents, and more particularly, to an analysis system and an analysis program for automatically calculating the market size of each technical document (patent document, academic paper, etc.).

Since patent documents such as application publications are a treasure trove of technical information, patent documents are collected and analyzed in many companies.
In addition, it is also practiced to devise so that the overall tendency in the technical field can be visually grasped by mapping the analysis result on a plane.

Various procedures have been proposed for the so-called patent map creation and expression format. However, the creation of such a map mainly involves human work and requires a lot of labor, and various tools have been proposed to support this. ing.
For example, in Patent Document 1 below, the external dimensions of a patent map, display items, items used for the vertical and horizontal axes, etc. are set in advance on the arrangement condition setting screen, and the two-dimensional object to be patented. A patent map creation system is disclosed that calculates the coordinates of sequence data (CSV file including data such as application date, classification, publication number, and comments) according to these arrangement conditions.
JP 2001-222536 A

However, the conventional patent map, of course, can only grasp the technical trend and distribution of patent applications, and could not display the market scale in the technical field to which each patent application is related with specific numerical values. .
Of course, when examining the direction of technological development, it is meaningful to create a patent map of a specific technical field and discover undeveloped areas, but for example, the marketability of patent applications held by acquired companies It is ineffective for the purpose of grasping as a general numerical value and the potential marketability of invention proposals and research plans.
Even if it is possible to artificially calculate the market size by carefully reading each document, identifying the relevant technical field, and comparing this with the figures of industrial statistics, etc. It is practically impossible to apply this work to a large amount of technical literature because it takes a lot of time and effort.

  The present invention has been made to solve the above-described conventional problems, and automatically identifies related technical fields such as patent documents and academic papers, and applies various statistical data and estimation data thereto. Therefore, the purpose is to realize a technology that makes it possible to present the marketability as a specific numerical value.

In order to achieve the above object, a marketability analysis system for technical literature described in claim 1 includes a dictionary vector storage means for storing dictionary vectors of a plurality of technical terms, and a reference numerical value representing the market size of each product. For each product category, which is a technical term, a means for obtaining a dictionary vector corresponding to the product category from the dictionary vector storage means, and a concept vector for each product category, By multiplying the concept vector by the above reference numerical value, a means for calculating each marketability vector, a means for extracting a keyword as a technical term from electronic technical literature, and a dictionary vector corresponding to the keyword are obtained. obtained from the dictionary vector storage means, by the reference vectors of each keyword vector synthesis, of the technical literature Means for generating a sense vector, to calculate the inner product value of the concept vector concept vector and the product category of the technical literature, Related Products Product Category that this inner product value becomes a predetermined vector similarity over the technical literature A means for certifying as a category; and means for calculating a marketability value of the technical document by integrating the inner product value of the marketability vector of each related product category and the concept vector of the technical document. It is said.

Further, the marketability analysis system for technical documents described in claim 2 is based on the system of claim 1 and further includes means for extracting a plurality of keywords from a plurality of electronic technical documents according to a predetermined standard. And a means for calculating the coordinates of each technical document on a two-dimensional plane by performing principal component analysis on the combination of keywords and the number of occurrences included in each technical document, and the marketability numerical value of each technical document It is characterized by comprising means for generating an image in which a mark provided with a color corresponding to 1 is arranged at the corresponding coordinates on the two-dimensional plane, and means for outputting the image.
The “output” includes printing via a printer and display on a display (the same applies hereinafter).

Market analysis program technical literature as claimed in claim 3, a computer, a dictionary vector storage means for storing the reference vectors of a plurality of technical terms, a reference value representing the market size for each product is the technical term Storage means registered for each product category , dictionary vector corresponding to the product category is acquired from the dictionary vector storage means, and is used as a concept vector for each product category. The concept vector for each product category is multiplied by the reference value. Means for calculating each marketability vector, means for extracting a keyword which is a technical term from electronic technical literature, and obtaining a dictionary vector corresponding to the keyword from the dictionary vector storage means, by vector synthesis dictionary vector of keywords, generating a concept vector of the technical literature That means calculates an inner product value of the concept vector for each product category and concept vector of the technical literature, it means to certify product categories this inner product value becomes a predetermined vector similarity or a related product categories of the technical literature In addition, by integrating the inner product value of the marketability vector of each related product category and the concept vector of the technical document, it functions as a means for calculating the marketability value of the technical document.

  Further, the marketability analysis program for technical documents described in claim 4 is based on the program of claim 3, and further selects a plurality of keywords from a plurality of computerized technical documents according to a predetermined standard. Means for extracting, means for calculating the coordinates of each technical document on a two-dimensional plane by performing principal component analysis on the combination and number of occurrences of keywords included in each technical document, and the marketability numerical value of each technical document It is characterized by functioning as means for generating an image in which marks having colors corresponding to large and small are arranged at corresponding coordinates on the two-dimensional plane and means for outputting the image.

  According to the marketability analysis system and the marketability analysis program of the technical literature according to the present invention, various technologies such as industrial statistics and research results are described based on data describing a numerical value (for example, product shipment value) representing the market size for each product category. It is possible to automatically calculate the marketability value (market size value) of documents without human intervention, macroscopically grasp the market size of a large number of patent applications held by a specific company, invention proposals and research It is effective for the purpose of grasping the market value of plans, etc. in advance.

FIG. 1 is a block diagram showing a configuration example of a marketability analysis system 10 of technical literature according to the present invention, which is composed of a computer 16 such as a PC to which an input device 12 such as a keyboard and a mouse and a display 14 are connected. The computer 16 includes a search processing unit 20, a dictionary vector generation unit 22, a concept vector generation unit 24, a related product category extraction unit 26, a marketability numerical value calculation unit 28, a marketability numerical value display unit 30, and technical literature. A database 32, an extracted document storage unit 34, a technical document database for dictionary 36, a dictionary vector storage unit 38, a reference numerical value database 40, and a marketability numerical value storage unit 42 are provided.
The search processing unit 20, the dictionary vector generation unit 22, the concept vector generation unit 24, the related product category extraction unit 26, the marketability numerical value calculation unit 28, the marketability numerical value display unit 30, the CPU of the computer 16 is an OS and a dedicated application This is realized by executing necessary processing according to a program.
The technical document database 32, the extracted document storage unit 34, the dictionary technical document database 36, the dictionary vector storage unit 38, the reference numerical database 40, and the marketability numerical storage unit 42 are provided in the memory or hard disk of the computer 16. Yes.

In this marketability analysis system 10, a dictionary vector is generated in advance by the dictionary vector generation unit 22 based on registration information in the dictionary technical document database 36 and stored in the dictionary vector storage unit 38.
Hereinafter, a method for generating the dictionary vector will be described.
First, the dictionary technical document database 36 stores text data such as physics and chemistry dictionaries, various glossaries, patent documents, academic papers, etc., and the dictionary vector generation unit 22 converts these data into text mining algorithms. To break down into word levels.
Next, unnecessary words such as conjunctions are deleted and a predetermined number of technical terms are extracted, and the frequency of occurrence of each technical term and the number of appearing documents are tabulated, and these data are substituted into a dedicated algorithm. (Stem vector) is generated.

For example, “Photocatalyst is made of highly active photocatalytic thin film coating material that uses weak ultraviolet light contained in sunlight and room light. In the case where a technical document with the content “development is in progress” is stored in the technical document database 36 for dictionaries, the dictionary vector generation unit 22 will now “photocatalyst”, “sunlight”, “room light”, Technical terms “ultraviolet light”, “photoactive”, “thin film coating material”, “photocatalytic reaction”, “antibacterial”, “self-cleaning”, “air purification material” are extracted. These technical terms are analyzed as one concept of similarity as having high co-occurrence in one document (prone to appear at the same time).
The dictionary vector generation unit 22 repeats such an analysis based on many technical documents, thereby generating the technical terms as described above as vectors each having a direction and forming a dictionary vector.
Since the dictionary vector generation logic itself is already known, further detailed description is omitted.

FIG. 2 shows an image of the dictionary vector 44, which shows a state in which a plurality of technical terms such as PC, computer, ion, and alkali are oriented based on the strength of mutual relation and are unit vectorized. It is.
In this figure, technical terms with close vector directions (narrow angles between vectors) have high similarity, and technical terms with long vector directions (wide angles between vectors) have low similarity. Represents.
For convenience of illustration, the number of technical terms is limited to ten, but actually hundreds of technical terms are extracted and arranged on a vector space of hundreds of dimensions.

Next, according to the flowchart of FIG. 3, the processing procedure of marketability numerical value calculation in the analysis system 10 will be described.
First, when a user inputs a specific company name and year range as search conditions to the computer 16 via the input device 12, the search processing unit 20 extracts technical documents that match the search conditions from the technical document database 32 (S10). ).
The technical document database 32 stores in advance electronic data of a large number of patent documents (for example, application publications), and the search processing unit 20 selects a target document by referring to bibliographic data of each patent document.
Next, the search processing unit 20 extracts the description contents such as the application number, the filing date, the applicant, the inventor, the publication number, and the abstract of each selected document, and converts them into a predetermined data format (CSV or the like). The extracted document storage unit 34 stores it (S12).

Next, the concept vector generation unit 24 is activated, and a plurality of technical keywords are extracted from the substantial description contents (for example, text data of the abstract) of each patent document stored in the extracted document storage unit 34, and these are extracted into the dictionary. By comparing with the dictionary vector 44 stored in the vector storage unit 38, a concept vector of the patent document itself is generated (S14).
When extracting the keywords, the concept vector generation unit 24 decomposes the description of the abstract into word levels according to a text mining algorithm, removes unnecessary words registered in advance, and extracts only technical terms together with the appearance frequency. .
For example, when a description of “a coating material that contains a new photocatalyst that can also use visible light and removes nitrogen oxides in the air” is included in a certain patent document, the concept vector generation unit 24 includes the following: Thus, technical terms and their appearance frequencies are extracted.
"Visible light" ... once "photocatalyst" ... once "in air" ... once "nitrogen oxide" ... once "remove" ... once "coating material"・ ・ 1 time And, the concept vector generation unit 24 determines whether or not there is a corresponding technical term in the dictionary vector 44 in the dictionary vector storage unit 38 for each extracted technical term word. Get dictionary vector.
The acquired dictionary vector is finally combined with the appearance frequency, thereby generating a concept vector of the patent document.
Since the concept vector generation logic is also known, further description is omitted.

Next, the concept vector generation unit 24 refers to the reference numerical value database 40 and generates a product category concept vector according to the same logic as described above (S16).
That is, the reference numerical value database 40 stores statistical numerical values (for example, product shipment values) for each product category such as a personal computer, a camera, an automobile, an engine, and a storage device for each year. A concept vector of each product category is generated by comparing the category (technical term) or the technical term in the commentary with the dictionary vector 44 in the dictionary vector storage unit 38.

After the concept vector of each patent document and the concept vector of each product category are calculated by the concept vector generation unit 24, the related product category extraction unit 26 is activated to specify the product category related to each patent document (S18). .
FIG. 4 shows an image when selecting a related product category. The related product category extraction unit 26 calculates the inner product value of the concept vector T of the patent document and the concept vector Ci of each product category. A product category whose inner product value is equal to or greater than a predetermined vector similarity α (0 <α <1) is recognized as a related product category of the patent document.
In the figure, the concept vectors C2 and C3 of the product category are selected as related product categories, and the concept vectors C1 and C4 are excluded as unrelated product categories.
Here, the degree of similarity between the concept vector of the patent document and the concept vector of the product category is a problem as to whether it should be recognized as a related product category. It is appropriately set according to the level of detail of the product category in the numerical database 40.

Next, the marketability numerical value calculation unit 28 is activated and calculates the marketability numerical value (market scale) of the patent document by applying the reference numerical value to the related product category of each patent document (S20).
In other words, in order to simplify the vector operation when extracting the related product category, it was based on a unit vector having no size (concept vector of the product category), but the product category originally indicates the market size. Since specific numerical values (reference numerical values) are associated, the marketability of the patent document can be derived as a specific amount by reflecting this.

FIG. 5 shows an image of the marketability vector Mi in which the reference numerical value is reflected in the concept vector of each product category.
In this case, the marketability vector of the product category is obtained by multiplying the concept vector of the product category having a certain degree of similarity with the concept vector T of a certain patent document by multiplying a reference numerical value (product shipment amount, etc.) representing the market size. It shows that Mi is required.

Specifically, the marketability value calculation unit 28 calculates the marketability value of each patent document according to the following mathematical formula.

That is, the inner product value of the concept vector of each product category and the concept vector of the patent document is multiplied by the reference value of the product category, and the sum total of these values is obtained by the marketability value calculation unit 28 as the marketability value ( Certified as market size).
In this way, since the inner product values of the marketability vector of each product category and the concept vector of the patent document are accumulated, the marketability vector of the product category closer to the concept vector T of the patent document (the angle formed) is smaller. The marketability vector of the product category which is evaluated as enlarged and is farther than the concept vector of the patent document (the angle formed is larger) is reduced and evaluated.
The marketability numerical value as the calculation result is stored in the marketability numerical value storage unit 42 in association with the attribute information of each patent document by the marketability numerical value calculation unit 28 (S22).

Finally, the marketability numerical value display unit 30 is activated to extract the marketability numerical values (market scale amounts) and bibliographic items of all patent documents stored in the marketability numerical value storage unit 42, and use them as predetermined templates. The calculation result display screen is generated by filling and displayed on the display 14 (S24).
FIG. 6 shows an example of this. A list 46 including items of marketability values is displayed on the screen in addition to the application number, publication number, title of the invention, and general display items of the applicant. ing.
The market value of each patent document can be recognized based on the specific amount of money described in the item of the market value, and the market value between patent applications can be compared.

This marketability analysis system 10 has a function of more effectively displaying marketability of a large number of patent documents.
FIG. 7 shows a system configuration example in that case. The computer 16 includes a keyword extraction unit 48, a coordinate calculation unit 50, a drawing processing unit 52, a language dictionary group 54, an extracted document storage unit 34, and the like. , A keyword storage unit 56, a coordinate data storage unit 58, and a marketability numerical value storage unit 42.
The keyword extraction unit 48, the coordinate calculation unit 50, and the drawing processing unit 52 are realized by the CPU of the computer 16 executing necessary processes according to the OS and a dedicated application program.
The language dictionary group 54, the keyword storage unit 56, and the coordinate data storage unit 58 are provided in the memory or hard disk of the computer 16.

Hereinafter, a processing procedure for realizing this display method will be described with reference to the flowchart of FIG.
First, the keyword extraction unit 48 is activated, and a shaping process is performed on the actual part (for example, description of the abstract) of the patent document stored in the extracted document storage unit 34 (S30).
That is, the keyword extraction unit 48 decomposes the text data constituting the substantial part of each patent document into word levels according to a text mining algorithm.
Next, the keyword extraction unit 48 refers to the required replacement / deletion word dictionary 60 of the language dictionary group 54, deletes unnecessary words included in the text data, and executes necessary replacement processing.
Unnecessary words here are typical paragraph titles such as “task”, “solution”, “effect”, etc. included in the abstract, or conjunctions such as “by the way”, “however”, “where”, etc. , “Is”, “in this case”, and other words that have no technical meaning. In the required replacement / deletion word dictionary 60, a large number of required deletion words that should be excluded from analysis are registered in advance.
The necessary replacement processing corresponds to, for example, converting the Japanese calendar display of “Japanese Patent Application No. 55-” included in the application number into the Western calendar display of “Japanese Patent Application 1980-”. For this reason, correspondence data between the Japanese calendar and the Western calendar is stored in the replacement / deletion word dictionary 60 in advance.

Next, the keyword extraction unit 48 extracts keyword candidates from the formatted text data (S32), and totals each keyword candidate in the order of the number of appearance documents.
At this time, the keyword extraction unit 48 refers to the synonym / necessary word dictionary 62 to unify a plurality of technical terms that are substantially synonymous into one term (representative word). For example, if there is “carbon fiber” in one patent document and “carbon fiber” is described in another patent document, the keyword extraction unit recognizes that “carbon fiber” is described for each, It counts as an appearance literature of the keyword candidate concerned.

Here, the number of keyword candidates to be extracted from an infinite number of words can be set in the system 10 in advance, or can be set on the user side when an analysis request is made.
For example, when the upper limit of the number of extraction cases is set to 100, the keyword extraction unit 48 selects words in the top 100 of the number of appearance documents as keyword candidates.

However, if a specific technical term is set as a necessary word in the synonym / necessary word dictionary 62, the keyword extracting unit 48 ranks the word corresponding to the necessary word based on the number of appearance documents, for example, 100th or less. Even if there are, select them as keyword candidates.
In addition to preparing the necessary words in advance by the system operator and storing them in the synonym / necessary word dictionary 62, the user can also specify the required words when processing is requested.

FIG. 9 exemplifies the keyword candidate extraction result list 64, and the keyword candidate “image” having the highest appearance number appears 212 times in total in 105 patent documents. It shows that the ratio (appearance frequency) is 2.16%.
The keyword candidate extraction result is displayed on the display 14 (S34).
On the other hand, the user inputs a check mark to a check box of a necessary keyword candidate from the extraction result list 64 and clicks a “select” button (not shown).

In response to this, the keyword extraction unit 48 (S36) recognizes the keyword candidate designated by the user as an official keyword and stores it in the keyword storage unit 56 (S38).
At this time, the correspondence between the application number of each patent document, the extracted keyword, and the number of appearances of each keyword is also registered.

In the above, an example is shown in which the keyword extraction unit 48 extracts noun keyword candidates, but the present invention is not limited to this, and the noun and noun combination, the adjective and noun combination, Alternatively, a combination of nouns and verbs can be extracted as keyword candidates.
For example, “dirt of water (noun + noun)”, “high hydrophilicity (adjective + noun)”, “irradiate ultraviolet rays (noun + verb)”, and the like are applicable.
In addition, as described above, the keyword extraction unit 48 first lists keyword candidates, and instead of certifying a keyword selected by the user as a keyword, the keyword extraction unit 48 automatically determines the keyword. It can also be set.

Next, the coordinate calculation unit 50 is activated to calculate the position coordinates on the two-dimensional plane of each patent document (S40).
That is, the coordinate calculation unit 50 inputs each keyword combination associated with each patent document and the data of the number of appearances to the principal component analysis algorithm of multivariate analysis, as shown in FIG. X-axis coordinates (first principal component) and Y-axis coordinates (second principal component) of the document are calculated.
The calculation result is stored in the coordinate data storage unit 58.

Next, the coordinate calculation unit 50 calculates the distribution density of the patent documents on the two-dimensional plane based on the coordinate data of each patent document (S42).
That is, the coordinate calculation unit 50 converts the number of documents included in a certain area on the coordinate axis into the distribution density of the region.
FIG. 11 shows an image of the conversion process. After plotting each patent document on the two-dimensional plane 66 according to coordinates (black dots in the figure indicate the positions of the patent documents), the X axis and The Y-axis is divided into a plurality of regions by partitioning at a predetermined interval, and the number of documents included in each region is tabulated as the distribution density of the region.
For example, since four patent documents are included in the region α, the distribution frequency is counted as “4”. On the other hand, since one patent document is not included in the region of β, the distribution frequency is “0”.
The distribution density data is stored in the coordinate data storage unit 58.

Next, the coordinate calculation unit 50 inputs data on the number of patent documents including each keyword and the number of appearances to the principal component analysis algorithm, as shown in FIG. Principal component) and Y-axis coordinates (second principal component) are calculated (S44).
The calculation result is stored in the coordinate data storage unit 58.

Next, the drawing processing unit 52 is activated, and based on the calculation result by the coordinate calculation unit 50, a patent map reflecting the distribution density data and the keyword coordinate data is generated on a two-dimensional plane and displayed on the display 14 ( S46).
FIG. 13A shows an example of the patent map 68. Points indicating the presence of each keyword are plotted on the corresponding coordinates, and a corresponding keyword (character string) is displayed in the vicinity of each point. Has been.

Further, on the patent map 68, a figure (pattern) 70 corresponding to the distribution density of the patent document is displayed.
That is, as described above, the distribution density is associated in advance with the subdivided areas constituting the patent map 68, and the drawing processing unit 52 assigns different colors (display patterns) to the areas according to the distribution density. By assigning, a graphic (pattern) 70 reflecting the distribution density is drawn on the patent map 68.
There is no limitation on how colors are assigned to each region, but an example is as follows.
(1) Area where distribution density is 50 or more → Red
(2) Distribution density range from 40 to 49 → orange
(3) Distribution density is 30 to 39 → yellow
(4) Distribution density range of 20-29 → Yellowish green
(5) Area where distribution density is 10-19 → light blue
(6) Area with distribution density 1-9 → Blue
(7) Area with zero distribution density → Indigo

Since the patent map 68 in FIG. 13A is expressed in black and white, it is difficult to intuitively grasp the distribution density. However, since the distribution density is actually drawn vividly by the above colors, the user can By simply looking at the patent map 68 displayed on the display 14, it is possible to recognize the density of patent documents.
Instead of expressing the change in distribution density by color, it can also be expressed by other display patterns. For example, it corresponds to expressing a change in distribution density by giving a difference between shades in the same color. Alternatively, a change in distribution density may be expressed by assigning different patterns (hatched lines, mesh lines, etc.) to each region according to the distribution density.

Hereinafter, what information the user can read by observing the patent map of FIG. 13A will be described.
(1) First, the user can recognize the number of patent applications related to the keyword (technical theme) by referring to the distribution density spreading around the specific keyword. For example, when the periphery of the keyword “charger” is surrounded in red, it means that there are many patent applications related to “charger”.
(2) Next, the user can check the uniqueness of the keyword based on the position of each keyword on the map. That is, the keyword arranged at a position close to the center of the patent map 68 means that it is relatively orthodox, and it can be read that it is a unique technical element as it deviates from the center.
(3) The user can also infer the relationship and positioning between the technical elements based on the distances and combinations between the keywords.

Next, as shown in FIG. 13B, the drawing processing unit 52 generates a distribution map 74 in which dots 72 indicating the existence of each patent document are arranged in a two-dimensional plane according to the respective coordinates, and is displayed on the display 14. It is displayed (S48).
At this time, the drawing processing unit 52 refers to the marketability numerical value (market scale amount) of each patent document stored in the marketability numerical value storage unit 42 and assigns a color corresponding to the size to each dot.
There is no limitation on how colors are assigned to each market value, but an example is as follows.
(1) Market size is over 100 billion yen → Red
(2) Market size is 70 billion yen or more and less than 100 billion yen → Orange
(3) Market size is 50 billion yen or more and less than 70 billion yen → Yellow
(4) Market size is 30 billion yen or more and less than 50 billion yen → Yellowish green
(5) Market size is 10 billion yen or more and less than 30 billion yen → Light blue
(6) Market size is 5 billion yen or more and less than 10 billion yen → Blue
(7) Market size is less than 5 billion yen → Indigo

Since the distribution map 74 in FIG. 13B is expressed in black and white, it is difficult to intuitively grasp the market scale represented by each dot 72, but in reality, the dot 72 is vividly colored by the above colors. Therefore, the user can visually recognize the distribution status of the market scale of patent documents.
In addition, each patent document in the distribution diagram 74 in (b) is arranged at the same position as each patent document that is the basis of the distribution density calculation in the patent map 68 in (a), so that both of them should be compared and observed. Thus, it is possible to grasp the relationship between the density of patent applications and the market size.
Furthermore, the market value of each patent document is recalculated by changing the reference value of the related product category for each year, and this calculation result is reflected in the above distribution map 74. It can also be expressed as a change in color.

By using the marketability analysis system 10 of this technical document, a so-called PPM (Patent Portfolio Management) analysis can be performed on a group of patent applications related to a specific technical field possessed by a specific company.
As shown in FIG. 14, this PPM analysis is based on four quadrants (potential area, stable business area, business review area, and niche area). ), And the patent application group (Patent Portfolio) related to a specific technical field possessed by a specific company is positioned in the corresponding quadrant to analyze its characteristics.

For example, patent applications related to semiconductor manufacturing equipment owned by a general electronics manufacturer account for 30% of all patent applications in this field (occupation ratio: high), and each patent application calculated by this marketability analysis system 10 If the total market size is over 100 billion yen (market size: high), it falls under the “stable business domain” and it can be judged that there is no need to rush into patent application. Become.
In contrast, the noise filter-related patent applications owned by the manufacturer account for less than 5% of all patent applications in the same field (occupation rate: low), and the total market size of each patent application is less than 3 billion yen. If it is (market size: low), it falls under the “business review area”, and it can be judged that no more patent applications are needed.

In the above, a so-called stand-alone system configuration in which the system 10 is configured by setting up a dedicated application program in the computer 16 such as a PC is exemplified, but the present invention is not limited to this.
That is, the present invention can be realized as a client-server type system in which a dedicated application program is set up in a server and a service is provided to a large number of client terminals connected via a network such as the Internet.
In the above description, the technical document database 32 is stored in the hard disk of the computer 16, but the necessary document information is obtained from the document database stored in the DB server connected via the network. It can also be configured to extract.

In the above, an example of calculating the marketability of Japanese patent documents was shown, but this system 10 is a market for technical documents written in languages other than Japanese such as English, Chinese, German, and French. Needless to say, it can also be applied to sex calculations.
In addition, although patent documents are cited above as examples of technical documents, they are also effective for the purpose of calculating marketability of academic papers, invention proposals, research plans, and the like in advance.

It is a block diagram which shows the structure of the marketability analysis system of the technical literature based on this invention. It is a conceptual diagram which shows the image of a dictionary vector. It is a flowchart which shows the calculation procedure of a marketability numerical value. It is a conceptual diagram which shows the image at the time of selecting a related product category. It is a conceptual diagram which shows the image of the marketability vector which reflected the reference | standard numerical value in the concept vector of the product category. It is a layout figure which shows the display screen of a marketability numerical value. It is a block diagram which shows the system configuration | structure in the case of producing | generating the distribution map of the patent document reflecting the marketability numerical value. It is a flowchart which shows the process sequence in the case of producing | generating the distribution map of the patent document which reflected the marketability numerical value. It is a layout figure which shows the extraction result list | wrist of a keyword candidate. It is explanatory drawing which shows the calculation result of the X-axis coordinate of each patent document, and a Y-axis coordinate. It is a conceptual diagram which shows the image at the time of calculating distribution density from the coordinate of a patent document. It is explanatory drawing which shows the calculation result of the X-axis coordinate of each keyword, and a Y-axis coordinate. It is a layout figure which shows the distribution map of the patent document which reflected the patent map and the marketability numerical value. It is a conceptual diagram in the case of performing PPM analysis based on a marketability figure.

Explanation of symbols

10 Marketability analysis system for technical literature
12 Input device
14 display
16 computers
20 Search processing section
22 Dictionary vector generator
24 Concept vector generator
26 Related product category extractor
28 Marketability calculation part
30 Marketability value display
32 Technical literature database
34 Extracted document storage
36 Technical literature database for dictionaries
38 Dictionary vector storage
40 standard numerical database
42 Marketability numerical memory
44 dictionary vector
46 List of marketability figures
48 Keyword extractor
50 Coordinate calculator
52 Drawing processor
54 language dictionaries
56 Keyword storage
58 Coordinate data storage
60 Required replacement / deletion word dictionary
62 Synonyms and Necessary Words Dictionary
64 Keyword candidate extraction result list
66 2D plane
68 Patent Map
70 Figures corresponding to the distribution density of patent documents
72 Dots indicating presence of patent documents
74 Patent literature distribution map

Claims (4)

Dictionary vector storage means for storing dictionary vectors of a plurality of technical terms;
A storage means for registering a reference numerical value representing the market size of each product for each product category, which is a technical term ,
Means for obtaining a dictionary vector corresponding to the product category from the dictionary vector storage means, and as a concept vector for each product category ;
Means for calculating each marketability vector by multiplying the concept vector of each product category by the above-mentioned reference numerical value;
Means for extracting keywords, which are technical terms, from electronic technical literature;
Means for obtaining a dictionary vector corresponding to the keyword from the dictionary vector storage means and generating a concept vector of the technical document by vector synthesis of the dictionary vector of each keyword ;
Means for calculating the inner product value of the concept vector of this technical document and the concept vector of each product category, and certifying a product category whose inner product value is equal to or greater than a predetermined vector similarity as a related product category of the technical document;
Means for calculating the marketability value of the technical document by integrating the inner product value of the marketability vector of each related product category and the concept vector of the technical document;
Technical literature marketability analysis system characterized by comprising: Means for extracting a plurality of keywords according to a predetermined standard from a plurality of electronic technical documents;
Means for calculating the coordinates of each technical document on a two-dimensional plane by performing a principal component analysis on the combination of keywords and the number of occurrences included in each technical document;
Means for generating an image in which a mark having a color corresponding to the magnitude of the marketability value of each technical document is arranged at corresponding coordinates on the two-dimensional plane;
Means for outputting the image;
The marketability analysis system for technical literature according to claim 1, comprising: Computer
Dictionary vector storage means for storing dictionary vectors of a plurality of technical terms;
Storage means for registering the reference numerical value representing the market size of each product by product category, which is a technical term ,
Means for obtaining a dictionary vector corresponding to the product category from the dictionary vector storage means, and as a concept vector for each product category ;
Means for calculating each marketability vector by multiplying the concept vector of each product category by the above reference numerical value,
Means for extracting keywords, which are technical terms, from electronic technical literature,
Means for acquiring a dictionary vector corresponding to the keyword from the dictionary vector storage means and generating a concept vector of the technical document by vector synthesis of the dictionary vector of each keyword ;
Means for calculating the inner product value of the concept vector of the technical document and the concept vector of each product category, and certifying a product category in which the inner product value is equal to or higher than a predetermined vector similarity as a related product category of the technical document;
Means for calculating the marketability value of the technical document by integrating the inner product value of the marketability vector of each related product category and the concept vector of the technical document,
A marketability analysis program for technical literature, characterized by functioning as Computer
Means for extracting a plurality of keywords according to a predetermined standard from a plurality of electronic technical documents;
Means for calculating the coordinates of each technical document on a two-dimensional plane by performing principal component analysis on the combination of keywords and the number of occurrences included in each technical document;
Means for generating an image in which a mark having a color corresponding to the marketability numerical value of each technical document is arranged at corresponding coordinates on the two-dimensional plane;
Means for outputting the image;
The marketability analysis program for technical literature according to claim 3, wherein

Images