Patent evaluation device, patent evaluation method, and program

Abstract

The present invention provides patent evaluation more in line with interests of users. A patent evaluation device 1 according to an aspect of the present disclosure receives attribute information (attribute information α) regarding a plurality of patents from a user, and evaluates the patents using at least a weight (weight W) that is dependent on the attribute information α.

Description

Patent evaluation device, patent evaluation method, and program

[0001] The present disclosure relates to techniques for assessing the value of patent applications or patent rights (hereinafter collectively referred to as "patents").

[0002] When considering a company's intellectual property strategy, evaluating the value of patents is not only useful for identifying patents of competitors that should be guarded against, but is also extremely effective in determining whether to allocate resources to maintaining a company's own patents or licensing activities. Conventionally, the value of a patent has been calculated using a model with a predetermined evaluation method based on publicly available information about the patent, such as bibliographic information such as the filing date, content information such as the number of claims, and historical information such as whether the patent was divided (see, for example, Patent Document 1).

[0003] Patent No. 4344813

[0004] However, there are limitations to evaluation based solely on publicly available information. The relationship between business activities and patents is generally not publicly available, and each company's concerns and the weighting of those concerns are likely to differ depending on its industry, capital relationships, business strategy, intellectual property strategy, etc. In other words, the problem with conventional methods is that they do not necessarily provide patent evaluations that are in line with the user's concerns.

[0005] The present disclosure has been made to solve the above-mentioned problems, and aims to provide patent evaluation that is more in line with users' interests.

[0006] In order to solve the above problem, a patent evaluation device according to one aspect of the present disclosure receives attribute information about a plurality of patents from a user and evaluates the patents using at least weights that depend on the attribute information.

[0007] The patent evaluation device of the present disclosure can provide patent evaluations that are more in line with the user's interests.

[0008] FIG. 1 is a diagram showing an example of the functional configuration of a patent evaluation device according to this embodiment. FIG. 2 is a diagram showing an example of the processing flow of a patent evaluation method according to this embodiment. FIG. 3 is a diagram to assist in the explanation of the processing of the coefficient calculation unit 21 and the coefficient correction unit 23 in FIG. 1. FIG. 4 is a diagram showing an example of the functional configuration of a patent evaluation device according to Modification 1 of this embodiment. FIG. 5 is a diagram to assist in the explanation of the processing of the correlation coefficient calculation unit 24 in FIG. 4. FIG. 7 is a diagram showing an example of the functional configuration of a patent evaluation device according to Modification 2 of this embodiment. FIG. 8 is a diagram to assist in the explanation of the processing of the patent evaluation method according to Modification 2 of this embodiment. FIG. 9 is a diagram to assist in the explanation of the processing of the reference index generation unit 25 in FIG. 7. FIG. 10 is a diagram to assist in the explanation of the processing of the reference index generation unit 25 in FIG. 7. FIG. 11 is a diagram showing an example of the functional configuration of a computer.

[0009] <Character notation> The symbol "￣" (overline) used in text should normally be written directly above the character immediately following it, but due to limitations in text notation, it is written immediately before the character in question. In mathematical formulas, these symbols are written in their proper position, i.e., directly above the character. For example, "￣x" is expressed in a mathematical formula as follows:

[0010] A patent evaluation device according to an embodiment of the present disclosure receives attribute information (attribute information α) for multiple patents from a user and evaluates a patent (evaluation target patent T) using at least a weight (weight W) that depends on the attribute information α. ​​By configuring the patent evaluation device as described above, users can provide attribute information α for patents of interest, and this tendency can be reflected in the calculation of evaluation points for patent value evaluations, thereby providing patent evaluations that are more in line with the user's interests. Below, a patent evaluation device 1, an example of an embodiment of the present disclosure, is described in detail using figures. Also, components having the same functions are designated by the same numbers, and duplicate explanations will be omitted.

[0011] When evaluating the value of a patent, for example, some companies may have their own criteria for determining whether to file a divisional application, amend the specification, request examination, etc. In such cases, the importance of each item based on the unique criteria may differ compared to conventional evaluations using predetermined methods. The patent evaluation device 1 is a device that addresses such cases. As shown in FIG. 1, the patent evaluation device 1 according to an embodiment of the present disclosure includes an attribute quantification unit 10, a model construction unit 20, and an evaluation unit 30. The model construction unit 20 includes a coefficient calculation unit 21, a polarity assignment unit 22, and a coefficient correction unit 23. The patent evaluation device 1 performs the patent evaluation method of this embodiment by executing the processing flow shown in FIG. 2. The patent evaluation device 1 receives attribute information α for multiple patents previously specified by the user and a patent T to be evaluated. An input receiving unit (not shown) sends the attribute information α to the attribute quantification unit 10, and sends the patent T to the evaluation unit 30. However, the patent T to be evaluated may be configured to be transmitted to the evaluation unit 30 via the attribute quantification unit 10, or may be configured to be transmitted to the evaluation unit 30 via the attribute quantification unit 10 and the model construction unit 20.

[0012] The attribute information α includes at least attribute items indicating the attribute information items of the patent (for example, types of information such as "application number" and "registration number" described below) and attribute data that is data corresponding to the attribute items (for example, data such as "20230315", "2023 / 03 / 15", and "2023-03-15" if the application date is March 15, 2023). The attribute information α may include public information and non-public information. Public information may include information made public by the Patent Office and information that can be objectively learned from that information. Specific examples of public information include, but are not limited to, (i) to (iii) below. (i) bibliographic information such as application number, registration number, technical field information, filing date information, priority date information, and whether or not an exception to lack of novelty applies; (ii) content information correlating the number of claims, the number of independent claims, the average number of characters per claim, the number of specification pages, and the number of drawings; and (iii) historical information such as whether or not a divisional application has been filed and the number of times it has been filed, whether or not a request for accelerated examination has been made, whether or not a patent decision has been made in an appeal against a decision of rejection, whether or not a decision to maintain a patent in an opposition, whether or not a decision to maintain a patent in an invalidation trial has been made, whether or not a priority claim has been made, whether or not a PCT application has been filed, the number of countries in which the patent has been filed, whether or not the file wrapper has been viewed, and the number of times the patent has been cited. If a database of public information is accessible, the information may be obtained from the database to supplement the user's input. For example, if only a registration number is entered, attribute information α, such as the number of claims for the patent corresponding to that registration number, may be obtained from a database (not shown) and used for subsequent processing.

[0013] Non-public information refers to information about patents other than that published by the Patent Office. Examples of non-public information include, but are not limited to, the following (iv) through (vii): (iv) revenue information, such as licensing revenues, damages claimed in patent infringement lawsuits, and settlement payments; (v) rights utilization information, such as the presence and number of licenses, the presence or absence of licenses to group companies, the presence or absence of licenses to other companies, and the presence and frequency of license negotiations; (vi) background information, such as whether or not a patent has been implemented in-house, the presence and priority of development related to the patented invention, and the investment costs leading up to the invention; and (vii) subjective information, such as manual assessments of the patentability of claims and the likelihood of proving implementation, performed by personnel. For example, an organization may wish to assess the value of a patent by balancing the priority of in-house development with quality, such as the breadth of the scope of the patent. In such cases, combining non-public information with public information can provide a patent valuation that meets the above concerns.

[0014] (Attribute Quantification Unit 10) The attribute quantification unit 10 quantifies the attribute data contained in the attribute information α of the received multiple patents (Step S10). The attribute data in the attribute information α is converted into a numerical value based on, for example, the following ideas (1) to (5), but is not limited to these. (1) For dates such as the filing date, the length of the period until the patent expiration, a function value that exponentially decays with respect to that period, or a numerical value that uniquely identifies the date is assigned and used. (2) For information such as presence / absence, "presence" and "absence" are expressed using any two different values, such as 1 and 0 or 1 and -1. (3) For information with patterns, such as transition countries, a 1 or 0 (Japan transition presence / absence, China transition presence / absence) is assigned to each pattern, corresponding to presence / absence. Alternatively, a uniquely identifiable numerical value is assigned and used. (4) The number of times may be used as is or may be a logarithmic value. (5) For ratings such as A, B, and C, numerical values ​​that correspond to the ranking of the ratings, such as 1, 2, and 3, are used, from lowest to highest.

[0015] The attribute digitizing unit 10 transmits the converted value corresponding to each digitized attribute information α to the model building unit 20 .

[0016] (Model Construction Unit 20) The model construction unit 20 calculates a weight (weight W) for each piece of attribute information α from the converted value of the received quantified attribute information α, and constructs a model (model M) for evaluating patents using the weight W. The process of constructing the model M is performed by the coefficient calculation unit 21, polarity assignment unit 22, and coefficient correction unit 23.

[0017] (Coefficient Calculation Unit 21) The coefficient calculation unit 21 calculates a predetermined coefficient for each of the converted values ​​of the received quantified attribute information α (step S21). As an example of a method for calculating the coefficient, for each attribute information α, the standard deviation among the input multiple patents (hereinafter also referred to as "patent group") is calculated, and the reciprocal of the standard deviation is used as the coefficient. For simplicity's sake, as shown in FIG. 3, it is assumed that there are two patents for which the attribute information α items include the number of characters in the claim (claim character count C), the number of characters in the specification (specification character count D), and the resources required to arrive at the invention (resources R). Symbols indicating the attribute data of the attribute information α are written in lowercase. Furthermore, to distinguish between the two patents (patent 1, patent 2), subscripts are written in the attribute data of the attribute information α. ​​That is, when the attribute data of the attribute information α of the two patents input by the user are written in the order of (C, D, R), the following is obtained: (c １ , d １ , r １ ), (c ２ , d ２ , r ２ In this case, the coefficient a for the number of claim characters C is Ｃ , coefficient a for the number of characters in the specification D Ｄ , coefficient a for resource R Ｒ is calculated as follows: In the above formulas (1) to (3), a is a coefficient that can be corrected by the coefficient correction unit 23, which will be described later. The standard deviation may be calculated by taking the square root of the mean square of each data, as in the formulas (1) and (2), or by taking the square root of the mean absolute value, as in the formula (3). The calculated coefficient a Ｃ , a Ｄ , a Ｒis transmitted to the polarity assigning unit 22.

[0018] (Polarity Assignment Unit 22) The polarity assignment unit 22 assigns a predetermined polarity to each of the received coefficients (step S22). Assigning a polarity means assigning a positive or negative sign to the coefficient of each attribute information α calculated by the coefficient calculation unit 21. As an example of how polarity is assigned, when considering the value evaluation of a patent, a positive sign is assigned if the value is considered to be improved, and a negative sign is assigned if the value is considered to be reduced. To explain using a specific example, if the attribute information α item is the number of claim characters C, a shorter number of claim characters generally tends to broaden the scope of rights. Therefore, if it is believed that the greater the number of claim characters, the lower the evaluation score of the patent value. In the case of the number of specification characters D (or the number of specification pages), generally, the greater the number, the more likely it is that there will be more grounds for amendment and that it will be easier to assert patentability. Therefore, if it is believed that the greater the number of specification characters D (or the number of specification pages), the higher the evaluation score of the patent value, a positive sign is assigned.

[0019] Therefore, after polarity is imparted, the formulas (1) to (3) become the following formulas (1)' to (3)'. ' Equations (2)' and (3)' are assigned positive signs and are the same as equations (2) and (3), respectively. However, these are merely examples and are not limited to these rules. In other words, it is sufficient to assign a polarity that is qualitatively determined in advance to each piece of attribute information α. ​​Coefficient information including the polarity is transmitted to the coefficient correction unit 23.

[0020] (Coefficient Correction Unit 23) The coefficient correction unit 23 corrects the coefficients taking into account the received coefficients and polarity (step S23). The coefficient correction method, for example, involves multiplying the obtained coefficient (here, a coefficient including polarity) by the digitized converted value of each attribute information α for each patent in the input patent group, and then summing the respective values ​​to calculate the raw score. Then, similar to calculating a deviation score, the coefficient including polarity is corrected by calculating the constant a described above and the constant b described below so that the average value, variance value, or maximum and minimum values ​​become predetermined values. However, the correction method described above is merely an example and is not limited to this. Furthermore, whether or not to perform this correction is optional, and if correction is not performed, the processing of step S23 is not necessarily required.

[0021] For ease of understanding, the above-mentioned correction method will be explained in the case where the group of patents received by the attribute digitization unit 10 is two patents, Patent 1 and Patent 2, as shown in Fig. 3. In this case, the constructed model M is defined as in the following equation (4). Using equation (4), the evaluation points of the patent value for the two patents given in advance above are given by the following equation (5) for patent 1 and by the following equation (6) for patent 2. The coefficient correction unit 23 performs a process of correcting a and b so that the average value of the above equations (5) and (6) becomes 50 and the variance value becomes 1. In the case of the above (4), after the values ​​of a and b are determined, the coefficient a Ｃ , a Ｄ , a Ｒ , and b correspond to the weight W described above, and equation (4) itself after the values ​​of a and b have been determined corresponds to the model M constructed by the model construction unit 20. However, equation (4) is merely an example. The definition equation is not limited to this. In other words, the weight W may be defined by other expressions other than a linear model. The constructed model M is transmitted to the evaluation unit 30.

[0022] As can be seen from equation (4), for each term in model M, the weight W for each attribute information α is determined by a coefficient obtained from a predetermined group of patents. In other words, the weight W depends on the attribute information α. ​​When the coefficient changes, the degree of change in the patent value evaluation score changes when the corresponding attribute information conversion value changes. In other words, the weight W indicates the degree of change in the patent value evaluation score when the quantified conversion value of the attribute information α changes. Building a model involves determining the weight W for each attribute information α. ​​For example, if the change in the patent value evaluation score when the quantified conversion value of the attribute information α changes from 0 to 1 differs between Patent 1 and Patent 2, the weight W changes taking this difference into account. Furthermore, when determining the weight W, the polarity of each coefficient is qualitatively determined. In other words, in the example used to explain the processing of the polarity assigning unit 22, the more positive the deviation from the average, the higher the patent's value, and the more negative the deviation from the average, the lower the patent's value. Note that when the numerically converted conversion value of attribute information α is evaluated as 1 or 0, the model places more weight on the lower frequency. In other words, in this case, a method is adopted in which higher points are given to events that are less likely to occur in reality.

[0023] (Evaluation Unit 30) The evaluation unit 30 uses the received model M to generate information about the evaluation of the patent (evaluation information V) for the evaluation target patent T, which is the designated patent to be evaluated (step S30).

[0024] The patent evaluation device 1 does not necessarily need to be configured so that only one evaluation target patent T, which is the designated patent to be evaluated, is input; multiple patents may be designated. Furthermore, the method of inputting the evaluation target patent T may involve designating it from a group of patents for which attribute information α has been input, or designating a new patent separate from this group of patents. When a new evaluation target patent T is designated, the user may be prompted to input the attribute information α of the patent when designating the new evaluation target patent T, so that the attribute information α of the patent is also input into the patent evaluation device 1. Alternatively, the device may be configured so that when information that can identify the evaluation target patent T is obtained, the information can be obtained by accessing a specific database (not shown).

[0025] Specific examples of information regarding patent valuation (valuation information V) include, but are not limited to, the following (1) to (3): (1) displaying the numerical value of the valuation score of the patent, or outputting numerical data of the valuation score of the patent; (2) displaying a rank such as A, B, or C according to the numerical value of the valuation score of the patent, or outputting rank data; (3) if there are multiple patents T to be evaluated, displaying or outputting the ranking of the patents according to their valuation scores. In the case of (3) above, a method may be adopted in which the data is processed to the extent that the relative highs and lows of the valuation scores of the patents for the multiple patents T to be evaluated are not reversed, and information indicating the highs and lows of the valuation scores of the patents is presented as visual information or data.

[0026] According to the embodiment described above, the user can provide attribute information α of patents of interest, and this tendency can be reflected in the calculation of the evaluation score of the patent value, thereby providing patent evaluations that are more in line with the user's interests.

[0027] <Variation 1> As shown in Fig. 4, the patent evaluation device 1 may be configured as a patent evaluation device 1a having a model construction unit 20a instead of the model construction unit 20. For example, when evaluating the value of a patent to identify patents that are likely to generate a large amount of license revenue, it may be possible to provide a patent evaluation that meets the user's interests by appropriately setting the weight for each attribute information α from the perspective of the amount of license revenue, with reference to the amount of license revenue from past patents. The patent evaluation device 1a is a device that can handle such cases.

[0028] The model construction unit 20a does not have the coefficient calculation unit 21 and polarity assignment unit 22 in the model construction unit 20, but instead has a correlation coefficient calculation unit 24. Also, the coefficient correction unit 23 has been replaced with a coefficient correction unit 23a. In the patent evaluation device 1a, the model construction unit 20a calculates the weight W by using one piece of attribute information α designated from the received (hereinafter also referred to as "received") attribute information α as a reference index (reference index β) and taking into consideration the degree of correlation of each of the other attribute information α with the reference index β.

[0029] The patent evaluation device 1a performs the patent evaluation method of this modified example 1 by implementing the processing flow shown in Fig. 5. In Fig. 5, step S24 is added instead of steps S21 and S22 in Fig. 2. Also, step S23 has been changed to step S23a. Therefore, the following explanation will focus on the processing of steps S24 and S23a, and other explanations will be omitted.

[0030] As shown in FIG. 4 , the patent evaluation device 1a receives attribute information α for multiple patents previously determined by the user, a patent T to be evaluated, and one type of reference indicator β, which is information on the attribute items of the reference attribute information α. ​​An input receiving unit (not shown) is configured to send the attribute information α to the attribute quantification unit 10, the reference indicator β to the model construction unit 20a, and the patent T to be evaluated to the evaluation unit 30. However, the patent T to be evaluated may be transmitted to the evaluation unit 30 via the attribute quantification unit 10, or may be transmitted to the evaluation unit 30 via the attribute quantification unit 10 and the model construction unit 20a. Furthermore, the reference indicator β may be transmitted to the model construction unit 20a via the attribute quantification unit 10. In this first modification, as described above, one type of reference indicator β is provided by the user. In the following description of this first modification, the license revenue amount (license revenue amount L) is assumed to be specified as the reference indicator β. That is, the model construction in the model construction unit 20a of the present modified example 1 constructs a model in which the linear sum of each attribute information α is used as an evaluation point of the value of the patent based on the license income amount L for the group of patents provided by the user.

[0031] (Correlation Coefficient Calculation Unit 24) The correlation coefficient calculation unit 24 uses one attribute item designated from the received attribute information α as a reference index β, and calculates a coefficient by taking into consideration the degree of correlation between each attribute data of the reference index β and the attribute data of each other attribute information α (step S24). As an example of a coefficient calculation method, for each attribute information α, the standard deviation other than the reference index β is calculated among the input patent group, and its reciprocal is used as the coefficient. Furthermore, a new coefficient is calculated by multiplying the coefficient by the correlation coefficient (correlation coefficient F) between each attribute information α and the license revenue amount L, which is the reference index β. Here, as shown in FIG. 6, it is assumed that the number of claim characters C, the number of specification characters D, and the license revenue amount L are input for each of two patents (Patent 1 and Patent 2) as items of attribute information α, and the license revenue amount L is specified as the reference index β. Expressed in the same format as in the above-described embodiment, the user inputs attribute data of the attribute information α as (c １ , d １ , l １ ), (c ２ , d ２ , l ２ ) is input, and the license income amount L is specified as the reference index β. In this case, the correlation coefficient between the number of claim characters C and the license income amount L is calculated as the correlation coefficient F Ｃ , the correlation coefficient between the number of characters in the specification D and the license income L is the correlation coefficient F Ｄ Then, the correlation coefficient F Ｃ Coefficient a taking into account Ｃ , correlation coefficient F Ｄ Coefficient a taking into account Ｄ is calculated as in the following equations (7) and (8). In general, if the correlation coefficient between x and y is r, the correlation coefficient r can be calculated by the following formula (9): where n is the number of data (x, y), x ｉ is the i-th value of x, y ｉ represents the i-th value of y, ￣x represents the average value of x, and ￣y represents the average value of y. Correlation coefficient F Ｃ , F Ｄ may be calculated with reference to equation (9). In the above formulas (7) to (8), a is a coefficient that can be corrected by the coefficient correction unit 23a. The standard deviation may be calculated by the square root of the mean square of each data, as calculated in the above formulas (7) and (8), or by the square root of the mean absolute value, as in the above formula (3). The calculated coefficient a Ｃ , a Ｄ is transmitted to the coefficient correction unit 23a.

[0032] (Coefficient Correction Unit 23a) The coefficient correction unit 23a corrects the coefficients taking into account the received coefficients (step S23a). The correction method, for example, is to multiply the obtained coefficient by the digitized converted value of each attribute information α for each patent in the input patent group, and then calculate the raw score by summing the respective values. Then, like a deviation value, the coefficients are corrected by calculating the above-mentioned constants a and b so that the mean, variance, or maximum and minimum values ​​become predetermined values. However, the above-mentioned correction method is merely an example and is not limiting. Furthermore, whether or not to perform this correction is a matter of choice, and if correction is not performed, the processing of step S23a is not necessarily required.

[0033] The correction method will be explained using the examples of Patent 1 and Patent 2 shown in Fig. 6. In this case, the constructed model M is defined as in the following equation (10). Using equation (10), the evaluation points for the patent value assessment of the two predetermined patents described above are given by the following equation (11) for patent 1 and the following equation (12) for patent 2. The coefficient correction unit 23a performs a process of correcting a and b so that the average value of the above equations (11) and (12) becomes 50 and the variance value becomes 1. In the case of the above equation (10), after the values ​​of a and b are determined, the coefficient a Ｃ , a Ｄ , and b correspond to the weight W described above, and the formula (10) itself after the values ​​of a and b have been determined corresponds to the model M constructed by the model construction unit 20a. However, formula (10) is merely an example. The definition formula is not limited to this. In other words, the weight W may be defined by other expressions other than a linear model. The constructed model M is transmitted to the evaluation unit 30.

[0034] When determining the weight W, the slope of the linear regression line of the number of claim characters versus the license revenue L may be used instead of the ratio of the correlation coefficient F to the standard deviation. The coefficient may be calculated using multivariate regression analysis with the license revenue L as the reference indicator β. Model M may be a single-layer or multi-layer neural network instead of a linear sum model. When using a neural network, the neural network parameters are updated an appropriate number of times using backpropagation or other methods to construct a model so that the output of the neural network using attribute information α as input and the actual license revenue L are close on a predetermined scale. If the patent group given in advance changes, the resulting neural network parameters will change, changing the weight W of each attribute information used in calculating the evaluation score for the patent valuation. This will change the degree of change in the evaluation score for the patent value when the value of attribute information α changes during actual evaluation. Regarding the license revenue L, a flag indicating whether a license agreement has been concluded may be used as an indicator. In this case, a numerical value such as 1 / 0 may be used.

[0035] <Variation 2> As shown in FIG. 7, the patent evaluation device 1a may be configured as a patent evaluation device 1b having a model construction unit 20b instead of the model construction unit 20a. The value of a patent is not necessarily measured by a single, clear indicator, such as the amount of license revenue L. There are various possible benchmarks (Merkmar K) that are not the ultimate goal for a company, but can be considered intermediate targets for the performance expected from a patent, such as obtaining a contract regardless of revenue, implementing a patented invention in-house, or using it as advertising in sales activities. In such cases, the relative merits of these benchmarks may vary depending on the company and the timing of the patent value evaluation. In this variation 2, a portion of the attribute information is pre-defined as a benchmark through user interviews, and the relative merits of the defined benchmarks are also pre-defined. This generates a reference index (equivalent to the reference index β in Variation 1) used to construct a model for evaluating the value of a patent. By quantifying the benchmarks while taking into account the user's internal circumstances and building a model of evaluation points for patent value assessment, we will be able to provide patent evaluations that are more in line with the user's interests.

[0036] 7, the model construction unit 20b has a reference index generation unit 25 as a process before the correlation coefficient calculation unit 24. In the patent evaluation device 1b, the calculation of weights in the model construction unit 20b generates a reference index (reference index Q) taking into consideration merkmal K, which is a plurality of attribute information items designated from the received attribute information α, and the order (order P) between the merkmal K, and this reference index Q is used as the reference index β in the above-mentioned modified example 1.

[0037] The patent evaluation device 1b performs the patent evaluation method of this modified example 2 by executing the processing flow shown in Fig. 8. In Fig. 8, step S25 is added before step S24 in Fig. 5. Therefore, the explanation will focus on step S25, and other explanations will be omitted.

[0038] As shown in FIG. 7 , the patent evaluation device 1b receives input of attribute information α for multiple patents previously determined by the user, the patent T to be evaluated, and information on a merit mark K, which is an attribute item of the multiple attribute information α that serves as a merit mark from the attribute information α, and its ranking (rank P). An input receiving unit (not shown) sends the attribute information α to the attribute quantification unit 10, the merit mark K and ranking P to the model construction unit 20b, and the patent T to be evaluated to the evaluation unit 30. However, the patent T to be evaluated may be transmitted to the evaluation unit 30 via the attribute quantification unit 10, or may be transmitted to the evaluation unit 30 via the attribute quantification unit 10 and the model construction unit 20b. The merit mark K and ranking P may be transmitted to the model construction unit 20b via the attribute quantification unit 10. Here, the merit mark K is input as attribute data that can be expressed as presence / absence or 1 / 0, such as contract history.

[0039] (Reference index generation unit 25) The reference index generation unit 25 quantifies each achieved milestone K based on the received milestone K and ranking P to generate (calculate) a reference index Q (step S25). The reference index Q is a quantification of the degree of influence that the milestone K achieved by the patent T to be evaluated has on the value of the patent for the company evaluating the patent T to be evaluated.

[0040] When quantifying the achieved Merkmal K, the following rules (1) to (4) are taken into consideration, for example, but are not limited to these.

[0041] (1) For each milestone K, count how many patents in the patent group provided by the user have achieved milestone K of rank P or higher than that milestone K, and use this as the number of patents that have achieved that milestone K.

[0042] (2) Set arbitrary numerical values, a ceiling standard value (ceiling standard value CR) and a point width standard value (point width standard value PR), multiply the rate of the number of patents that achieve each milestone K to the total number of patents in the patent group by the point width standard value PR, and subtract the value obtained by multiplication from the ceiling standard value CR to obtain the quantified conversion value for that milestone K. (3) For patents that do not achieve any milestone K, the conversion value is obtained by subtracting the resource R from the ceiling standard value CR.

[0043] (4) When evaluating each patent based on the degree of achievement of the benchmark, the patent is evaluated based on the converted value of the benchmark K of the highest rank P that the patent has achieved.

[0044] A specific example of a method for calculating the reference index Q based on the above rules (1) to (4) is as follows. For example, as shown in FIG. 9, suppose that the attribute information α designated as the milestone mark K has two attribute items, "licensing history" and "in-house use," and that the ranking P between these milestone marks K is input as "licensing history" being first and "in-house use" being second. In other words, suppose that information is given indicating that licensing history is ranked higher than in-house use. Based on the above rule (1), if there are 500 patents given in advance, it is calculated that of these 500, 100 have a licensing history, 200 have an in-house use, 20 have both a licensing history and an in-house use (hereinafter also referred to as "those that meet both criteria"), and 220 do not meet either criteria.

[0045] Assume that the ceiling reference value CR is set to 120 and the point width reference value PR is set to 100. Considering the above rules (2) and (3), the result is as follows: The converted value for licensed patents is 120 - 100 / 500 x 100 = 100. The converted value for in-house implementation is 120 - (100 + 200) / 500 x 100 = 40. The converted value for patents that have achieved neither licensed patents nor in-house implementation (those that do not qualify) is 120 - 100 = 20. Furthermore, taking into account the above rule (4), the final generated reference index Q is calculated as shown in Figure 10: 100 for patents that have achieved the achieved milestone K but have only achieved licensed patents, and patents that have achieved both licensed patents and in-house implementation; 40 for patents that have achieved the achieved milestone K but have only achieved in-house implementation; and 20 for patents that have achieved neither milestone K. The reference index Q calculated for all 500 input patents is sent to the correlation coefficient calculation unit 24 together with the attribute information α.

[0046] The calculation method for the standard index Q is based on the idea that the lower the frequency of achievement of a benchmark K, the higher the hurdle for achieving it, and the higher the hurdle for achieving it, the greater the impact upon achievement. While it is possible to simply quantify the achievement rate of each benchmark K, it is not necessarily true that important benchmarks K will have a relatively low achievement rate. Therefore, as in rule (1) above, a more intuitive index can be derived by using the cumulative number of achievements based on the ranking P for the quantification calculation. The number of patents and the number of patents achieved may also be logarithmic. Furthermore, the ceiling reference value CR and the point width reference value PR may be determined so that the average, variance, maximum, and minimum values ​​when scoring a given group of patents are predetermined.

[0047] The correlation coefficient calculation unit 24 performs the process of step S24 described above using the received reference index Q and attribute information α. ​​Specifically, it is assumed that the number of claim characters C, the number of specification characters D, and the reference index Q are input for two patents (Patent 1 and Patent 2). In other words, using the notation method already described, the reference index generation unit 25 generates (c １ , d １ , q １ ), (c２ , d ２ , q ２ ) is received, the reference indicator Q is set as the new reference indicator β in the first modification, and the processes of steps S24, S23a, and S30 described above are similarly performed.

[0048] In providing information in step S30 in this modified example 2, when presenting the evaluation score of the value of the designated patent, if it is known whether or not the patent has achieved Merkmal K, the scoring by Merkmal K used in model construction (the scoring result in which the value of the standard index Q is regarded as the evaluation score) may be presented in parallel. The evaluation score of the value of the patent output from the model and the score by Merkmal K may be processed into the average, maximum, minimum, etc., and provided as an overall score.

[0049] <Variation 3> When receiving private information from a user, there is a growing need to ensure security to prevent information leaks. On the other hand, if a program for creating an evaluation model is given to a user, there is a risk that it may be copied. Therefore, the calculations in the above example may be performed in encrypted form using a secure computation process that allows calculations to be performed while the data is encrypted, so that even if the information received from the user is leaked, a third party cannot view the information. The above embodiment, or Variation 1 or Variation 2, can be realized more safely.

[0050] In this case, a cryptosystem for which the user holds the key is used. The user inputs encrypted attribute information α (as well as the patent T to be evaluated, the reference index β, the merit K, and the ranking P) into a patent evaluation device 2 with a secure computation function (herein, the patent evaluation device 1, the patent evaluation device 1a, or the patent evaluation device 1b with a secure computation function will be collectively referred to as the "patent evaluation device 2"). Based on the received attribute information α, the patent evaluation device 2 performs a calculation equivalent to that of the above embodiment using secure computation to construct an encrypted model M. Once the model M is constructed, the encrypted attribute information related to the patent designated by the user is input into the encrypted model M, and encrypted evaluation information V is obtained through a secure computation process. The user then decrypts the encrypted evaluation information V using their own key to obtain the information.

[0051] By configuring the patent evaluation device 2 as described above, the patent evaluation device 2 and third parties cannot decipher the attribute information α (other information, the patent to be evaluated T, the reference index β, the merit K, the ranking P) input by the user, the constructed model M, and the evaluation information V output from the model, so non-public information can be handled with peace of mind.

[0052] Although the embodiments and modifications of this disclosure have been described above, the specific configurations are not limited to these embodiments and modifications, and it goes without saying that appropriate design changes, etc., are included in this disclosure as long as they do not deviate from the spirit of this disclosure. The various processes described in the embodiments and modifications may not only be executed chronologically in the order described, but may also be executed in parallel or individually depending on the processing capacity of the device executing the processes or as needed.

[0053] [Program, Recording Medium] The various processes described above can be implemented by loading a program that executes each step of the above method into the recording unit 2020 of the computer 2000 shown in FIG. 11 and operating the control unit 2010, input unit 2030, output unit 2040, display unit 2050, etc.

[0054] The program describing the processing contents can be recorded on a computer-readable recording medium, which may be, for example, a magnetic recording device, an optical disk, a magneto-optical recording medium, a semiconductor memory, or any other suitable recording medium.

[0055] The program may be distributed by, for example, selling, transferring, lending, etc. portable recording media such as DVDs and CD-ROMs on which the program is recorded. Furthermore, the program may be stored in a storage device of a server computer, and then transferred from the server computer to other computers via a network, thereby distributing the program.

[0056] A computer that executes such a program may first temporarily store the program recorded on a portable recording medium or transferred from a server computer in its own storage device. Then, when executing a process, the computer reads the program stored on its own recording medium and executes the process in accordance with the read program. Alternatively, the computer may read the program directly from a portable recording medium and execute the process in accordance with the program. Furthermore, the computer may execute the process in accordance with each program transferred from the server computer. Alternatively, the server computer may not transfer the program to the computer, but may instead execute the process through a so-called ASP (Application Service Provider) service, which realizes the processing function by issuing an execution instruction and obtaining the results. In this embodiment, the program includes information used for processing by a computer that is equivalent to a program (e.g., data that is not a direct instruction to the computer but has properties that define computer processing).

[0057] Furthermore, in this embodiment, the device is configured by executing a predetermined program on a computer, but at least a part of the processing contents may be realized by hardware.

[0058] 1, 1a, 1b Patent evaluation device 10 Attribute quantification unit 20, 20a, 20b Model construction unit 21 Coefficient calculation unit 22 Polarity assignment unit 23, 23a Coefficient correction unit 24 Correlation coefficient calculation unit 25 Reference index generation unit 30 Evaluation unit C Number of claim characters CR Ceiling reference value D Number of specification characters F Correlation coefficient K Merkmar L License income amount M Model P Ranking PR Point width reference value R Resources T Patent to be evaluated V Evaluation information W Weight α Attribute information β, Q Reference index

Claims

1. Each user receives attribute information about multiple patents, Using at least weights that depend on the received attribute information, information regarding the evaluation of the patent to be evaluated is generated for each user. Patent evaluation device.

2. The patent evaluation apparatus according to claim 1, wherein the attribute information includes confidential information.

3. An attribute quantification unit that quantifies attribute information of multiple patents defined for each user, A model building unit calculates weights for each attribute using the quantified attribute information and constructs a model for evaluating patents using the weights, An evaluation unit that uses the aforementioned model to generate patent evaluation information for each user regarding the specified patent to be evaluated, A patented evaluation device having the following features.

4. The calculation of weights by the aforementioned model building unit is as follows: A coefficient calculation unit that calculates a predetermined coefficient for each of the aforementioned attribute pieces of information, A polarity assigning unit assigns a predetermined polarity to each of the aforementioned coefficients, A correction unit that corrects the coefficient by taking into account the coefficient and polarity. The patent evaluation apparatus according to claim 3, which is implemented by further implementation.

5. The calculation of weights in the aforementioned model building unit is as follows: The patent evaluation apparatus according to claim 3, wherein one attribute information specified from among the multiple patent attribute information defined by the user is used as a reference index, and the degree of correlation of each attribute information with respect to the reference index is taken into consideration.

6. The calculation of weights in the aforementioned model building unit is as follows: The patent evaluation apparatus according to claim 5, which generates a new standard index that takes into account multiple attribute information items designated from the received attribute information as benchmarks and the order between the benchmarks, and uses the new standard index as the standard index.

7. Each user receives attribute information about multiple patents, Using at least weights that depend on the received attribute information, information regarding the evaluation of the patent to be evaluated is generated for each user. Patent evaluation method.

8. A program for causing a computer to function with the patent evaluation apparatus described in claims 1 to 6.