JP6799235B2 - Patent value evaluation device and patent value evaluation method - Google Patents

Description

The present invention is an intellectual property right obtained by filing an application with the Patent Office, such as a patent application and a patent right, a utility model registration application and a utility model right, a design registration application and a design right, and a trademark registration application and a trademark right. Regarding the evaluation device and evaluation method for evaluating the value.

Since these intellectual property rights have similar legal systems and procedures, the present invention can be applied to any of them, but the following description describes patent applications and patents including patent rights. ..

Conventionally, various methods have been proposed as methods for calculating the economic value of a patent. For example, the DCF method is a method of estimating the cash flow that the target patent will generate in the future and calculating the present value (current price calculation) at a discounted interest rate. This future cash flow estimate relies strongly on subjective judgment as well as future changes in the technical and economic environment of the subject patent. Therefore, there is a drawback that the objectivity is not sufficient.

In addition, the market approach method is a method of inferring the value of a target patent from an example of the transaction price of a patent similar to the target patent, but it is rare that a reference similar patent is traded independently. It is difficult to correctly evaluate the value of each patent. Furthermore, since this market approach method is basically a bilateral sale, the patent value is determined by referring to the transaction established by the agreement between the seller and the buyer for similar patents, so this method is also not sufficiently objective. ..

In addition to this, there are cost approach methods, modeling methods, etc., but all have the same problems, and when trying to evaluate a large number of target patents, the cost becomes high and processing takes time. is there.

On the other hand, a data-centered statistical evaluation method utilizing internal information such as examination status and rights status, applications subject to accelerated examination, number of cited applications, number of days elapsed from filing date, and progress information has also been proposed. (Patent Document 1 and Patent Document 2). These prior arts include prescribed actions by the applicant and the right holder of the target patent, such as applications subject to accelerated examination and international applications, and prescribed actions by other companies for the target patent, such as trial for invalidation and provision of information. The examiner's prescribed actions, such as the number of notices of reasons for refusal and the number of cited documents, and the passage of time, for example, the number of days elapsed from the filing date and the number of days elapsed from the registration date, are incorporated into a certain formula and parameter allocation. Therefore, the value of the patent is evaluated simply and mechanically in an attempt to solve the above-mentioned problems of lack of objectivity and high cost.

Since the prior art described in Patent Documents 1 and 2 is a data-centered evaluation method, it is difficult for arbitrary judgment and operation of the target patent to intervene, so that a certain improvement in objectivity is observed. Arbitrary judgment is included in the evaluation method itself regarding the selection of the calculation formula and the parameter allocation method.

Patent Document 1 can be said to be a direct evaluation method because the target patent is evaluated based on the predetermined act of the target patent, for example, the failure of the trial for invalidation is evaluated through the degree of attention. If there is no patent, it will not be evaluated. Therefore, there is a problem that most patents cannot be evaluated because only a few percent of all patents have a prescribed act from another person. Therefore, for example, in the case of a small and medium-sized enterprise, since the number of applications filed annually is small, the value of the patents it holds cannot be evaluated, and it is not possible to properly evaluate the value of the patents. Further, in Patent Document 1, there is also a problem that subjective judgment intervenes in the setting of the cost determined for each predetermined act and the obsolescence method itself over time.

Based on the idea that only patents having economic value commensurate with the maintenance cost are maintained, Patent Document 2 adopts information that brings about a significant maintenance rate difference among the attribute information related to the target patent as an evaluation item. The evaluation items adopted are calculation formulas according to the type of items, for example, binary items such as the presence or absence of divisional applications are assigned points according to the presence or absence, that is, 1 or 0 / (total number of populations) ^0. The evaluation score is calculated according to ^.5 . Then, Patent Document 2 discloses a calculation method in which the summation method is the square root of the sum of squares and the summation value empirically approximates a lognormal distribution. However, there is no rational basis for the total value that is the result of these series of calculation formulas and total methods to be a lognormal distribution approximation, and it is simply empirically that the total value approximates a lognormal distribution. Only by itself, subjective judgment intervenes in the evaluation result. In addition, depending on the technical field, if exceptional evaluation items are included, there is a risk that an extremely biased and inappropriate evaluation result will be obtained. For example, in the field of chemistry, there are relatively many cases where experimental data is newly added to the prior application and a domestic priority application is filed, whereas in the field of machinery and electricity, there are few domestic priority applications. In the prior art of Patent Document 2, since there is no concept of selecting an appropriate evaluation item for each technical field, the evaluation result may be biased.

Japanese Patent No. 5273840 Japanese Patent No. 4344813

As described above, in the prior art described in Patent Document 1 and Patent Document 2, subjective judgment intervenes in the selection of evaluation items, the selection of calculation formulas, and the parameter allocation among the attribute information related to the target patents. Therefore, there is a possibility that the result of the patent value evaluation includes an arbitrary judgment.

The present invention has been made in view of such problems, and is a patent value evaluation device and a patent value that enable evaluation with enhanced objectivity and rationality by eliminating arbitrary patent value evaluation as much as possible. The purpose is to provide an evaluation method.

The patent value evaluation device according to the present invention is
A population storage unit that stores multiple patents subject to patent evaluation as a population and stores this population,
The evaluation item storage unit that stores the evaluation items and the public information published in the gazettes, etc. regarding each patent belonging to this population are used as evaluation items.
Random variable X _ab (for the _ath patent) determined according to the content of the evaluation item for each patent belonging to the population by reading out the information stored in the population storage unit and the evaluation item storage unit. (Bth evaluation item), this random variable X _ab is multiplied by each patent, and the patent reference value for the ath patent is Π (all evaluation items), where h is the total number of evaluation items. When X _a = X _a1 × X _a2 × ・ ・ ・ × X _ab × ・ ・ ・ × X _ah , and m is a disparity parameter set according to the characteristics of the technical field of the population, Y _ab = Calculate log (X _ab ^m ), convert all random variables X _ab to Y _ab , and calculate the mean E (Y _b ) and variance V (Y _b ) of Y _ab for each evaluation item. Distributed arithmetic unit and
The random variable _Yab is set as a reference value so that the average E (Y _b ) is 0 and the variance V (Y _b ) is 1, and the obtained evaluation item reference values are added up for each patent to obtain an evaluation item reference. value sum sigma (total evaluation _{item) Y a = Y a1 + Y} a2 + ··· + Y ab + ··· + Y ah, the reference value sum calculator for calculating all patents,
A random variable storage unit that stores the average E (Y _b ) and variance V (Y _b ) of the random variables _Yab before being converted to the reference value, and the total Σ (all evaluation items) Y of the evaluation item reference values.
As will be described later, this Σ (all evaluation items) Y is approximated to a normal distribution by the central limit theorem, but the degree to which this Σ (all evaluation items) Y deviates from the normal distribution is calculated, and among the above evaluation items An optimization calculation unit that removes one or more evaluation items and calculates Σ (all evaluation items Y) again to minimize the degree to which Σ (all evaluation items) Y deviates from the normal distribution.
Calculate the patent reference value for each patent by Π (all evaluation items) X = {exp (Σ (all evaluation items) Y)} ^{1 / m} based on the total evaluation item reference values approximated to the normal distribution. Patent standard value calculation unit and
The patent standard value storage unit that stores the patent standard value and
It is characterized by being equipped with.

In this patent value evaluation device,
Based on the total evaluation item reference values approximated to the normal distribution, the patent power deviation value for each patent is calculated by the patent power deviation value = 50 + (Σ (all evaluation items) Y) × 10 / h ^0.5. Patent power deviation value calculation unit and
A patent power deviation value storage unit that stores the patent power deviation value,
May have.

In addition, other patent value evaluation devices according to the present invention are
A population storage unit that stores multiple patents subject to patent evaluation as a population and stores this population,
The evaluation item storage unit that stores the evaluation items and the public information published in the gazettes, etc. regarding each patent belonging to this population are used as evaluation items.
The information stored in the population storage unit and the evaluation item storage unit is read out, and the content of the evaluation item is set to the random variable X _ab (b-th evaluation item for the a-th patent) for each patent belonging to the population. ), This random variable X _ab is multiplied by each patent, and the patent reference value for the ath patent is Π (all evaluation items) X _a = X _a1 with the total number of evaluation items as h. × X _a2 × ・ ・ ・ × X _ab × ・ ・ ・ × X _ah , and when m is a disparity parameter set according to the characteristics of the technical field of the population, Y _ab = log (X _ab ^m) ) calculates and converts the random variable _{X ab} to _{Y ab,} and the average E _{(Y b)} and the average and variance calculating section for calculating a variance V _{(Y b)} of the _{Y ab} of each evaluation item,
The random variable _Yab is set as a reference value so that the average E (Y _b ) is 0 and the variance V (Y _b ) is 1, and the obtained evaluation item reference values are added up for each patent to obtain an evaluation item reference. value sum sigma (total evaluation _{item) Y a = Y a1 + Y} a2 + ··· + Y ab + ··· + Y ah, the reference value sum calculator for calculating all patents,
A random variable storage unit that stores the average E (Y _b ) and variance V (Y _b ) of the random variables _Yab before being converted to the reference value, and the total Σ (all evaluation items) Y of the evaluation item reference values.
The degree to which this Σ (all evaluation items) Y deviates from the normal distribution is calculated, one or more of the evaluation items are removed, and Σ (all evaluation items Y) is calculated again, and Σ (all evaluation items) is calculated. Item) An optimization calculation unit that minimizes the degree to which Y deviates from the normal distribution,
Based on the total evaluation item reference values approximated to the normal distribution, the patent power deviation value for each patent is calculated by the patent power deviation value = 50 + (Σ (all evaluation items) Y) × 10 / h ^0.5. Patent power deviation value calculation unit and
A patent power deviation value storage unit that stores the patent power deviation value,
It is characterized by being equipped with.

In these patent value evaluation devices, for example,
It may have a patent data input processing unit for inputting information on the patent of the population and information on evaluation items.

In addition
It may be possible to have a disparity input processing unit capable of inputting the disparity parameter m and changing the disparity parameter m used in the calculation.

Furthermore, it is also possible to provide a population extraction means for extracting the population of the patents to be evaluated from the patent database based on a predetermined standard and storing the published information in the gazette or the like in the population storage unit. it can.

Furthermore, in order to analyze the patent value based on the patent standard value or the patent power deviation value, it is preferable to have a graph drawing unit for graphing the patent standard value or the patent power deviation value. In this case, the graph drawing unit can also graph the patent reference value or the patent power deviation value over time.

The patent value evaluation method according to the present invention is
Multiple patents to be evaluated for patent value are used as the population, and public information published in publications, etc. regarding each patent belonging to this population is used as the evaluation item.
When the random variable X _ab (the bth evaluation item for the ath patent) is assigned according to the content of the evaluation item for each patent belonging to the population, this random variable X _{ab is used} for each patent. Multiply calculation is performed, and the patent standard value for the ath patent is calculated with h as the total number of evaluation items, and Π (all evaluation items) X _a = X _a1 × X _a2 × ・ ・ ・ × X _ab × ・ ・ ・When it is calculated as × X _ah and m is a disparity parameter set according to the characteristics of the technical field of the population, _Yab = log (X _ab ^m ) is calculated and the random variable X _{ab is set} to _Yab . Convert and calculate the average E (Y _b ) and variance V (Y _b ) of _Yab for each evaluation item.
The random variable _Yab is set as a reference value so that the average E (Y _b ) is 0 and the variance V (Y _b ) is 1, and the obtained evaluation item reference values are added up for each patent to obtain an evaluation item reference. Total value Σ (all evaluation items) Y _a = Y _a1 + Y _a2 + ... + Y _ab + ... + Y _ah is calculated for all patents.
The degree to which this Σ (all evaluation items) Y deviates from the normal distribution is calculated, one or more of the evaluation items are removed, and Σ (all evaluation items Y) is calculated again, and Σ (all evaluation items) is calculated. Item) Perform optimization so that the degree to which Y deviates from the normal distribution is minimized.
It is characterized in that the patent reference value and / or the patent power deviation value is calculated based on the total evaluation item reference values approximated to a normal distribution.

In this patent valuation method, for example
It is preferable to be able to change the disparity parameter m.

Further, the patent reference value and / or the patent power deviation value can be configured to be graphed over time.

Furthermore, information analysis can be performed by combining the patent reference value and / or the patent power deviation value with bibliographic items and keywords of patent information.

According to the present invention, a plurality of evaluation items are assigned to a plurality of patents as a population, and a plurality of evaluation items are added up (multiplied) for each patent with the score X as the patent reference value. seeking _PaiX, by calculating the ^{_{Y ab = log (X ab m}} ), converts the random variable _{X ab} to _{Y ab,} seek evaluation item reference value sum ΣY each patent, this evaluation item reference value total ΣY Since it is approximated to a normal distribution by the central pole limitation theory, the degree of the approximation is subjected to a chi-square test to optimize the approximation by combining evaluation items, and the patent reference value and / or the patent power deviation value is obtained. , The obtained patent standard value and / or patent power deviation value basically follows statistical theory, and arbitrary operations are excluded as much as possible, enabling the value evaluation of patents with extremely high objectiveness and rationality. ..

Further, by having a disparity input processing unit capable of inputting the disparity parameter m and changing the disparity parameter m used for calculation, the disparity parameter m can be adjusted, and the patent reference value can be set as a population. It is possible to accurately approximate the disparity in actual economic value in the technical field of.

Further, it is necessary to have a population extraction means for extracting the population of the patents to be evaluated from the patent database based on a predetermined standard and storing the published information in the publication or the like in the population storage unit. For example, it is convenient for selecting a population.

Furthermore, in order to analyze the patent value based on the patent standard value and / or the patent power deviation value, by having a graph drawing unit that graphs the patent standard value and / or the patent power deviation value, the patent is patented. Evaluation of patent value based on standard value and / patent power deviation value becomes easy and can be understood at a glance. In this case, by graphing the patent standard value and / or the patent power deviation value in relation to the filing year, etc., it is possible to easily understand the change over time in the patent standard value and / or the patent power deviation value. become.

It is a block diagram which shows the patent value evaluation apparatus which concerns on embodiment of this invention. The horizontal axis is the maximum value of the patent standard value, and the vertical axis is the average value of the patent standard value, and it is a graph showing the distribution of the patent standard value of companies A to F. The horizontal axis is the patent power deviation value, and the vertical axis is the number of patents, and the patent power deviation values of companies A to F are ranked by rank (15 or less, 15 to 25, 25 to 35, 35 to 45, 45 to 5). It is a graph which shows the number of patents (55, 55-65, 65-75, 75-85, more than 85). It is a pie chart which shows the total value of the patent standard value of company A to company F. The horizontal axis is the year of filing, and the vertical axis is the number of patents of company A (the length of the bar graph) and the range of the patent power deviation value. The number of patents acquired in each year and the number of patents acquired It is a graph which shows the high and low of the patent power deviation value. The horizontal axis is the year of filing, and the vertical axis is the total value of the patent standard value of company A (the length of the bar graph) and the maximum and average values of the patent standard value (line). It is a graph which shows the transition of the maximum value and the average. It is a flowchart which shows the operation of the patent value evaluation apparatus of this embodiment. It is a figure which shows the "initial setting screen (top screen)". It is a figure which shows the "evaluation item acceptance / rejection code confirmation / change" screen.

Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. FIG. 1 is a block diagram showing a patent value evaluation device according to an embodiment of the present invention. Various data are input from the input device 1, the data is input to the information processing device 2, and the calculation unit 23 obtains the patent reference value. In the process of this calculation, various data are input to and stored in the storage device 3, and the data is read out from the storage device 3 to the calculation unit 23 and used for the calculation. In the information processing apparatus 2, the patent data input processing unit 21 inputs and processes the published information of the publications and the like relating to each patent belonging to this population, with a plurality of patents to be evaluated for patent value as a population. Further, in the disparity input processing unit 22, the disparity parameter m is input processed. The information of this population is stored in the population storage unit 31 of the storage unit 3. Further, the evaluation items are stored in the evaluation item storage unit 32 of the storage unit 3. In addition to the above evaluation items, the data input from the input device 1 also includes bibliographic information such as the applicant, the patentee, and the year of application.

The average and variance calculation unit 24 of the calculation unit 23 reads out the information stored in the population storage unit 31 and the evaluation item storage unit 32, and sets the content of the evaluation item as a random variable X for each patent belonging to the population. _{When ab} (bth evaluation item for the ath patent) is set, this random variable X _ab is multiplied by a factor for each patent, and the patent reference value for the ath patent is calculated as h for the total number of evaluation items. As Π (all evaluation items) X _a = X _a1 × X _a2 × ・ ・ ・ × X _ab × ・ ・ ・ × X _ah . Further, the average and variance calculation unit 24 inputs the disparity parameter m set according to the characteristics of the technical field of the population from the disparity input processing unit 22. The mean and variance calculating unit ₂₄ calculates the ^{_{Y ab = log (X ab m}} ), converts the random variable _{X ab} to _{Y ab,} the average E _{(Y b)} and _{Y ab} of each evaluation item Calculate the variance V (Y _b ). The reference value total calculation unit 25 sets the random variable Y _ab as a reference value so that the average E (Y _b ) is 0 and the variance V (Y _b ) is 1, and the obtained evaluation item reference values are set as reference values. summing the patent, the evaluation item reference value sum sigma (total evaluation _{item) Y a = Y a1 + Y} a2 + ··· + Y ab + ··· + Y ah, operate on all patents. The average E (Y _b ) and variance V (Y _b ) of the random variable Y _ab before being converted into the reference value and the total evaluation item reference value Σ (all evaluation items) Y are the random variable storage units of the storage device 3. It is stored in 33.

The optimization calculation unit 26 calculates the degree to which this Σ (all evaluation items) Y deviates from the normal distribution, removes one or a plurality of evaluation items from the evaluation items, and again performs Σ (all evaluation items Y). The calculation is performed, and the evaluation items are optimized so that the degree to which Σ (all evaluation items) Y deviates from the normal distribution is minimized. The patent power deviation value calculation unit 27a sets each patent by patent power deviation value = 50 + (Σ (all evaluation items) Y) × 10 / h ^0.5 based on the total evaluation item reference value approximated to the normal distribution. Calculate the patent power deviation value for. This patent power deviation value is stored in the patent power deviation value storage unit 34a of the storage device 3. Further, the patent reference value calculation unit 27b sets Π (all evaluation items) X = {exp (Σ (all evaluation items) Y)} ^{1 / m} based on the total evaluation item reference values approximated to the normal distribution. Calculate the patent reference value for each patent. This patent reference value is stored in the patent reference value storage unit 34b of the storage device 3.

The patent power deviation value and the patent reference value obtained by the patent power deviation value calculation unit 27a and the patent reference value calculation unit 27b of the calculation unit 23 are the patent power deviation value storage unit 34a and the patent reference value storage unit 34b of the storage device 3. At the same time, it is stored in the graph by the graph drawing unit 28 of the information processing apparatus 2, and is output processed by the output unit 29. The obtained graph is printed by the output device 4 or displayed on the display device.

Next, the arithmetic processing in the arithmetic unit 23 will be described. First, the average and variance calculation unit 24 calculates the average and variance of the random variable Y. A plurality of patents (population) subject to patent evaluation are designated as Patent 1 to Patent n. In addition, the evaluation items adopted in the evaluation of patent value are, for example, information on the bibliographic items of the target patent, prescribed actions by the applicant or patentee (for example, application examination request, submission of written opinion and procedure amendment, payment of pension). Etc.), prescribed actions from others (information provision, request for trial for invalidation, etc.), prescribed actions of the examiner or judge (notification of reasons for refusal, decision of refusal, decision of patent, decision of patent trial, etc.), and filing date, etc. The time has passed from the examination request date and registration date to the present. Among these evaluation items, those adopted in the patent value evaluation are designated as evaluation items 1 to evaluation item h.

In addition, the population may select a group of patents for which patent value evaluation is to be performed based on some criteria and input it as data to the input device 1, or a patent database in which all available patents are stored. Then, data (search formula) that serves as a reference for sorting the patents to be used as the population is input, and the population extraction unit uses the patent data stored in the patent database to become the population according to the search formula. The patent may be extracted.

For example, as shown in Table 1 below, when there are h evaluation items for each of the n patents in the population to be analyzed, the evaluation contents of the evaluation items are "Yes" / "No". It is assumed that it is evaluated.

Then, a random variable X _ab corresponding to “yes” and “no” can be assigned to each evaluation item for each patent. In this case, the subscript "a" indicates the a-th patent (patent a), and "b" indicates the b-th evaluation item (item b). This is shown in Table 2 below.

Then, the evaluation contents of these evaluation items, as shown in the following equation 1, by summing (Renjo) for each patent, all patents: for (a 1 to n), obtaining the patent reference value ΠX _a.

In this case, the patent reference value ΠX is defined so that it can be added within the population. That is, for example, when the patent reference value ΠXa of patent _a becomes (patent reference value ΠX1 of patent ₁ ) + (patent reference value ΠX2 of patent ₂ ), the value of patent a and patent 1 The sum of the values of Patent 2 is equivalent.

Next, the random variable X _ab is converted into the random variable Y _ab according to the following mathematical formula 2.

However, log is a natural logarithm. Further, m is a disparity parameter. This disparity parameter m is set according to the characteristics of the technical field of the population.

The random variable Y _ab is obtained by the mathematical formula 2, and the X _ab can be obtained by the following mathematical formula 3 by the inverse transformation thereof.

The relationship between the random variable Y _{ab for} each evaluation item and its average E (Y) and variance V (Y) is shown in Table 3 below. In this case, if the table 1, "present", in the Y _ab shown in Table 3 in the case of "Y _{b (high)",} and "no" in Table 1, the Y _ab shown in Table 3. " Corresponds as "Y _b (low)".

Then, as shown in Table 3, the average E (Y _b ) and the variance V (Y _b ) of the random variables Y for each evaluation item are calculated.

Then, _Y mean E _{(Y b)} of _ab and variance V _{(Y b)} is such that each 0 and _1, to the reference value of the _{Y ab} for each evaluation item. That is, the average Y _b (high) and Y _b (low) values for each evaluation item are calculated so that the average E (Y _b ) for each evaluation item becomes 0 and the variance V (Y _b ) becomes 1. .. Specifically, if the probability of occurrence of Y _b (high) for each evaluation item is P, the probability of occurrence of Y _b (low) is 1-P. As evaluation items, for example, whether the right has been vested or pending at the JPO in the examination / right situation can only be obtained in two values, and whether or not an accelerated examination request has been made can also be obtained in only two values. Since there are no items, the patent value evaluation item can often have only two values. Therefore, for example, the score at the time of vesting the right in the examination / right situation is X = X (high), the score at the time of pending patent office is X = X (low), and the corresponding Y is Y (high), respectively. Let Y (low). Then, if the probability of occurrence of X (high) is P, the probability of occurrence of X (low) is 1-P. Therefore, the mean E (Y) and the variance V (Y) are expressed by the following mathematical formula 4. .. However, as will be described later, since the average E (Y) is 0, the term (E (Y)) ² in V (Y) is 0.

Then, Y (high) and Y (low) are calculated so that the average E (Y) = 0 and the variance V (Y) = 1. That is, in the formula 4, by setting E (Y) = 0 and V (Y) = 1, the following formula 5 can be obtained.

The standardized Y thus obtained is shown in Table 4 below. However, P ₁ , P ₂ , ..., P _b , ... _Ph are the evaluation items 1, the evaluation item 2, ... the evaluation item b, ... the Y (high) in the evaluation item h, respectively. The probability of occurrence.

In this way, the standardized random variable Y is obtained for each evaluation item and each patent. If h evaluation items 1 to h, which have the same mean and variance, are mutually independent random variables, then mean Y = ΣY / h = (Y ₁ + Y ₂ + ... + Y _b + ... The distribution of + Y _h ) / h converges to the normal distribution: N (0.1 / h) by the central limit theorem, which is the basic theorem of statistical theory. That is, when h is sufficiently large, the average Y follows a normal distribution: N (0, 1 / h), and the sum of Y for all evaluation items is ΣY = Y ₁ + Y ₂ + ... + Y _b + ... + Y. _h follows a normal distribution N (0, h). The central limit theorem discusses the error between the sample mean and the true mean, and in general, the error is approximated when the sample size is increased, regardless of the distribution of the population. It is a probabilistic theorem that approximates a normal distribution.

In the above embodiment, only the evaluation items that can take only two values are targeted, but this is the evaluation item that is the farthest from the convergence to the normal distribution indicated by the central limit theorem, and is three or more values to continuous values. Similarly, the central limit theorem is naturally applied to the evaluation items that take the above by standardizing with Y such that the mean E (Y) = 0 and the variance V (Y) = 1.

Then, the reference value total calculation unit 25 calculates the total evaluation item reference value. Table 5 below is a table showing the standardized _Yabs for each evaluation item and each patent.

The sum _{ΣY a = Y a1 + Y a2} + ··· + Y ab + ··· + Y ah for all evaluation items, the normal distribution: N since according to (0, h), for all patents in the population The total value ΣY is calculated, and this is named the total evaluation item reference value. Then, the patent power deviation value calculation unit 27a calculates the patent power deviation value based on the following mathematical formula 6. This patent power deviation normal distribution: approximating the N (50,10 ^2). This patent power deviation value is stored in the patent power deviation value storage unit 34a of the storage device 3.

Next, the optimization process of the evaluation item combination in the optimization calculation unit 26 will be described. In this optimization process, the degree to which Σ (all evaluation items) Y deviates from the normal distribution is calculated, one or more of the evaluation items are removed, and Σ (all evaluation items) Y is calculated again. Then, the degree to which Σ (all evaluation items) Y deviates from the normal distribution is minimized (step S7). In the technical field of the population including the target patent, if exceptional evaluation items (items with strong correlation, etc.) are included in all the items adopted as evaluation items, an extremely biased patent power deviation value is calculated. Resulting in. Therefore, the evaluation items are optimized by removing such inappropriate evaluation items and finding a combination of evaluation items of the population that best realizes the central limit theorem. The optimization step will be described below.

First, from all the items adopted as the evaluation items, the combination of the evaluation items excluding one or a plurality of evaluation items is automatically set in sequence, and the chi-square value for each combination is calculated.

Then, the patent power deviation values of all the patents in the population are sorted by rank to obtain the number of patents by rank. This is the number of achievements by rank. This rank is, for example, 9 ranks in which the patent power deviation value is "15 or less", "15 or more to 25 or less", "25 or more to 35 or less", ... "85 or more".

Normal distribution: N (50,10 ²⁾ by rank of number composition ratio can be calculated theoretical value from the statistical theory. That is,
"Theoretical number by rank" = "Theoretical value by rank / number composition ratio" x "Total number of patents in the population"
Can be obtained.

The chi-square value is the number of patents by rank as the actual number by rank.
"Chi-square value" = Σ (all ranks) {(rank / actual number-rank / theoretical number) ² / (rank / theoretical number)}
As you can see. However, ranks with less than 5 theoretical cases by rank will merge with adjacent ranks. Then, the goodness of fit to the normal distribution is determined based on this chi-square value. That is, as the chi-square value is small, patent power deviation of all patents in the population is normal distribution with theoretical distribution: conforms to N (50,10 ^2).

The combination of evaluation items that minimizes the chi-square value is defined as the optimization combination of evaluation items. In order to eliminate inappropriate evaluation items that would calculate an extremely biased patent power deviation value, the combination of evaluation items is automatically set in sequence to calculate the chi-square value, and the chi-square value is minimized. The combination of evaluation items is adopted as the optimum evaluation item for the population. As a result, the evaluation item that causes an extremely biased patent power deviation value can be eliminated, and the central limit theorem can be realized best.

In the patent standard value calculation unit 27b, when the evaluation items are optimized (evaluation items 1 to h are optimal) as described above, the chi-square value becomes the minimum for the evaluation items 1 to h. ) when the, for example, the patent a, obtaining patent reference value PaiX _a by the following mathematical formula 7.

As described above, the total ΣY _a (= _Ya1 + _Ya2 + ... + _Yab + ... + _Yah ) of all the evaluation items of the evaluation item reference value is normally distributed over the entire population: N (0, h). Follow. At this time, from formula 3,
X _ab = {exp ( _Yab )} ^{1 / m}
Since it is, Patent reference value PaiX _a formula 7, as follows, is transformed into evaluation item reference value total ShigumaY.
ΠX _a = X _a1 × X _a2 × ・ ・ ・ × X _ab × ・ ・ ・ × X _ah
= {Exp (Y _a1 )} ^{1 / m} × {exp (Y _a2 )} ^{1 / m} × ・ ・ ・ × {exp (Y _ah )} ^{1 / m}
= {Exp (Y _a1 + Y _a2 + ... + Y _ab + ... + Y _ah ) ^{1 / m}
= {Exp (ΣY _a )} ^{1 / m}
Then, by substituting ΣY of the mathematical formula 6 into this section to obtain the relationship between the patent reference value and the patent power deviation value, the following mathematical formula 8 is obtained.

Next, since the total evaluation item reference value: ΣY follows the normal distribution: N (0, h), the probability distribution that the patent reference value ΠX follows is the lognormal distribution: LN (0, h / m ² ). To explain. In this lognormal distribution LN (0, h / m ² ), "0" in the previous term indicates that the mean of the logarithms is 0, and "h / m ² " in the latter term has a logarithmic variance of h / m ² . Indicates that there is. When the total evaluation item reference value Z = ΣY of the patents in the population follows a normal distribution: N (0, h), the probability density function φ (z) is expressed by the following mathematical formula 9.

Here, when the probability density function of patent reference value: ΠX = {exp (ΣY)} ^{1 / m} = (exp (Z)) ^{1 / m} is f (x), the cumulative distribution function is obtained. By differentiating it, the probability density function f (x) is obtained. Assuming that the probability that X ≦ x is F (x), this cumulative distribution function F (x) is transformed as shown in the following mathematical formula 10. However, Pr indicates a probability. In addition, ∫ [−∞, log (x ^m )] indicates that the lower limit of integration is −∞ and the upper limit is log (x ^m ).

By differentiating this cumulative distribution function F (x) with respect to x, the probability density function f (x) is obtained by the following mathematical formula 11.

This equation 11 shows that the probability density function of ΠX = (exp (Z)) ^{1 / m} : f (x) has a lognormal distribution: LN (0, h / m ² ).

The distribution of the patent reference value is determined by the disparity parameter m, which depends on the technical field of the population to be analyzed. In addition, m is a continuous value. Table 6 below is a list showing the multiples of the disparity parameter m and the patent standard value of the 99% value. Here, the 99% value is, for example, the 99th numerical value when 100 numerical values are arranged in ascending order from the smallest. Similarly, the 50% value is the 50th numerical value, the so-called intermediate value. The disparity is (99% value patent standard value) / (50% value patent standard value), but since the 50% value patent standard value is exp (0) = 1, the disparity is 99% value. It becomes the patent standard value.

As described above, since the disparity is 1.00 or more and can be increased as much as possible depending on the disparity parameter m, and since m is a continuous value, the appropriate disparity parameter m is based on the known "intermediate value theorem". By selecting, it is possible to accurately approximate the disparity in actual economic value in the technical field of the population. As a result, the addability of the above-mentioned patent standard value ΠX is guaranteed. The economic value is premised on addability, and the calculation of the patent standard value is premised on being read as the economic value of the patent, and is proportional to the economic value of the patent.

From the left-right symmetry of the normal distribution, the inverse function of the cumulative distribution function of the normal distribution: N (0, h) is F ^-1 (0.5) = 0. Therefore, the inverse function of the cumulative distribution function of the lognormal distribution: LN (0, h / m ² ) is G ^-1 (0.5) = exp (0) = 1. On the other hand, the inverse of the cumulative distribution function of the log-normal distribution when the m = 1, G ^-1 | putting a (0.99 m = 1) = x , the inverse of the cumulative distribution function of the normal distribution: F ^{- 1} (0.99 | m = 1) = log (x). It is proved by Equation 11 that the variance of the lognormal distribution of m = k is 1 / k ² , that is, the standard deviation is 1 / k, and the standard deviation of the lognormal distribution obtained by lognormal transformation is also 1 / k. Therefore, the following mathematical formula 12 is established.

Therefore, the inverse function of the cumulative distribution function of the lognormal distribution when m = k is G ^-1 (0.99 | m = k) = x ^{1 / k} , and Table 6 is proved.

As described above, the patent reference value ΠX of each patent is obtained in the probability distribution in which the total evaluation item reference value ΣY is approximated to the normal distribution and the patent reference value ΠX is approximated to the lognormal distribution. The patent reference value ΠX is stored in the patent reference value storage unit 34b of the storage device 3.

Further, the graph drawing unit 28 graphs the patent reference value and the patent power deviation value for analysis of the patent reference value and the patent power deviation value. 2 to 6 are graphs showing an example thereof. FIG. 2 is a graph showing the distribution of patent reference values of companies A to F, with the maximum value of the patent reference value on the horizontal axis and the average value of the patent reference value on the vertical axis. In FIG. 2, the area of the circle is proportional to the total value of the patent reference value. From FIG. 2, it can be seen that the average values of the patent reference values of companies A and B show an approximate distribution, but the maximum value is higher in company B.

Further, in FIG. 3, the horizontal axis represents the patent power deviation value, and the vertical axis represents the number of patents, and the ranks of the patent power deviation values of companies A to F (15 or less, 15 to 25, 25 to 35, 35). ~ 45, 45-55, 55-65, 65-75, 75-85, over 85) The number of patents is shown. In each rank, the bar graphs are company A, company B ... company F in order from the left. From FIG. 3, it can be seen that each applicant has a characteristic that the patent power deviation value is biased toward a higher value or a lower value.

Further, FIG. 4 is a pie chart showing the total value of the patent reference values of companies A to F. FIG. 4 shows the composition ratio of the patent standard values of companies A to F to the patent standard values of the entire population. It can be seen that Company A has the highest total value of the patent standard value. A high total value of the patent standard value means that the value of the patents owned by Company A is high overall.

In Fig. 5, the horizontal axis is the year of filing, the vertical axis is the number of patents of company A (the length of the bar graph), and the range of the patent power deviation value is taken, and the number of patents acquired in each year and the acquisition thereof. It is a graph which shows the level of the patent power deviation value of the patented patent. The upper end of the line segment is the maximum value and the lower end is the minimum value, and the upper end of the rectangle is the 80% value and the lower end is the 20% value. From this figure, it can be seen that the number of patents for each application year of Company A, the height of the patent power deviation value for each application year, and the time-series changes in the deviation value. In FIG. 5, the filing year is shown on the horizontal axis, but the filing date or filing month may be used, or the registration year or publication year may be used. It can be graphed over time.

In FIG. 6, the horizontal axis is the year of filing, and the vertical axis is the total value of the patent standard value of company A (the length of the bar graph) and the maximum value and the average (line) of the patent standard value. It is a graph which shows the transition of the total value, the maximum value, and the average value of. From this figure, it can be seen that the applicant's patent reference value by application year changes over time.

The graphs obtained by these graph drawing units 28 are output to the outside from the output unit 29, printed by the output device 4, or displayed on the display device.

Next, the operation of the patent value evaluation device and the method configured as described above will be described. FIG. 7 is a flowchart showing this operation. First, when logging in to the software (step S1), for example, the "initial setting screen (top screen)" shown in FIG. 8 is displayed on the display device as the output device 4 (step S2). On this screen, select "Applicant" and so on. Also, specify the "evaluation graph". A "Data Import" button is arranged at the top of the screen, and by pressing this "Data Import" button, select the population data given as CSV data (step S4), and use this as the calculation unit. It is taken into 23 (step S5). Next, when the "calculation execution" button is selected (step S6), the calculation unit 23 executes the calculation. Then, various graphs are created using the obtained patent reference value and patent power deviation value (steps S7 and S8). This graph is output to the output device 4 and displayed on a print or display screen (step S9). In this case, in order to change the combination of evaluation item optimization according to the internal information of the company using this software, press the "evaluation item acceptance / rejection code" button of the master management on the initial setting screen (step). S10) and the "evaluation item acceptance / rejection code confirmation / change" screen shown in FIG. 9 are displayed. In the "evaluation item" displayed on this screen, select "adopt" of "1" in the user correction column, or select "not adopted" of "0". As a result, the acceptance / rejection result is displayed on the screen. When the acceptance / rejection of this evaluation item is changed, the calculation is executed again (step S3). Population data will also be captured again if necessary.

In this way, according to the present embodiment, the patent reference value and the patent power deviation value are calculated based on the data processed so as to be approximated to the normal distribution, and the patent value with high accuracy can be obtained. In addition, since there is little room for the operation of the result by the operator's arbitrary selection in this calculation, it is possible to analyze the patent value with high accuracy.

In the above embodiment, the patent power deviation value calculation unit 27a and the patent reference value calculation unit 27b are provided, but the present invention is not limited to this, and the patent power deviation value calculation unit 27a and the patent reference value calculation unit are not limited to this. It may have any of the parts 27b and evaluate the value of the patent by the patent power deviation value and the patent reference value.

Further, the present invention incorporates the patent reference value and / or the patent power deviation value into an existing patent analysis system, and performs analysis in combination with parameters such as bibliographic items and keywords. It can also be commercialized as an analysis device with a patent value added to the analysis method. Furthermore, the patent standard value and / or the patent power deviation value can be applied to the value evaluation of the design and the trademark, and the application of the design and the trademark can be applied with reference to these patent standard value and / or the patent power deviation value. It may be used as reference information for making decisions such as maintenance and waiver of strategies and rights. In the case of value evaluation of a design, the description of the scope of the patent claim of the present invention reads the patent as the design, and in the case of the value evaluation of the trademark, the description of the scope of the patent claim of the present invention is a patent. Should be read as a trademark.

1: Input device 2: Information processing device 3: Storage device 4: Output device 21: Patent data input processing unit 22: Disparity input processing unit 23: Calculation unit 24: Average and distribution calculation unit 25: Reference value total calculation unit 26: Optimization calculation unit 27a: Patent power deviation value calculation unit 27b: Patent reference value calculation unit 28: Graph drawing unit 29: Output unit 31: Population storage unit 32: Evaluation item storage unit 33: Random variable storage unit 34a: Patent power Deviation value storage unit 34b: Patent standard value storage unit

Claims (10)

A population storage unit that stores multiple patents subject to patent evaluation as a population and stores this population,
The evaluation item storage unit that stores the evaluation items and the public information published in the gazettes, etc. regarding each patent belonging to this population are used as evaluation items.
Random variable X _ab (for the _ath patent) determined according to the content of the evaluation item for each patent belonging to the population by reading out the information stored in the population storage unit and the evaluation item storage unit. (Bth evaluation item), this random variable X _ab is multiplied by each patent, and the patent reference value for the ath patent is Π (all evaluation items), where h is the total number of evaluation items. When X _a = X _a1 × X _a2 × ・ ・ ・ × X _ab × ・ ・ ・ × X _ah , and m is a disparity parameter set according to the characteristics of the technical field of the population, Y _ab = and calculates the log (X _ab ^m), converts the random variable _{X ab} to _{Y ab,} average and variance calculating calculates the _{Y ab} average E _{(Y b)} and variance V _{(Y b)} for each evaluation item Department and
The random variable _Yab is set as a reference value so that the average E (Y _b ) is 0 and the variance V (Y _b ) is 1, and the obtained evaluation item reference values are added up for each patent to obtain an evaluation item reference. value sum sigma (total evaluation _{item) Y a = Y a1 + Y} a2 + ··· + Y ab + ··· Y ah, the reference value sum calculator for calculating all patents,
A random variable storage unit that stores the average E (Y _b ) and variance V (Y _b ) of the random variables _Yab before being converted to the reference value, and the total Σ (all evaluation items) Y of the evaluation item reference values.
The degree to which this Σ (all evaluation items) Y deviates from the normal distribution is calculated, one or a plurality of evaluation items are removed from the evaluation items, and Σ (all evaluation items) Y is calculated again to calculate Σ (all evaluation items). Item) An optimization calculation unit that minimizes the degree to which Y deviates from the normal distribution,
Calculate the patent reference value for each patent by Π (all evaluation items) X = {exp (Σ (all evaluation items) Y)} ^{1 / m} based on the total evaluation item reference values approximated to the normal distribution. Patent standard value calculation unit and
The patent standard value storage unit that stores the patent standard value and
A patented value evaluation device characterized by being equipped with. Based on the total evaluation item reference values approximated to the normal distribution, the patent power deviation value for each patent is calculated by the patent power deviation value = 50 + (Σ (all evaluation items) Y) × 10 / h ^0.5. Patent power deviation value calculation unit and
A patent power deviation value storage unit that stores the patent power deviation value,
The patent value evaluation device according to claim 1, wherein the device is characterized by having. A population storage unit that stores multiple patents subject to patent evaluation as a population and stores this population,
The evaluation item storage unit that stores the evaluation items and the public information published in the gazettes, etc. regarding each patent belonging to this population are used as evaluation items.
Random variable X _ab (for the _ath patent) determined according to the content of the evaluation item for each patent belonging to the population by reading out the information stored in the population storage unit and the evaluation item storage unit. (Bth evaluation item), this random variable X _ab is multiplied by each patent, and the patent reference value for the ath patent is Π (all evaluation items), where h is the total number of evaluation items. When X _a = X _a1 × X _a2 × ・ ・ ・ × X _ab × ・ ・ ・ × X _ah , and m is a disparity parameter set according to the characteristics of the technical field of the population, Y _ab = and calculates the log (X _ab ^m), converts the random variable _{X ab} to _{Y ab,} average and variance calculating calculates the _{Y ab} average E _{(Y b)} and variance V _{(Y b)} for each evaluation item Department and
The random variable _Yab is set as a reference value so that the average E (Y _b ) is 0 and the variance V (Y _b ) is 1, and the obtained evaluation item reference values are added up for each patent to obtain an evaluation item reference. value sum sigma (total evaluation _{item) Y a = Y a1 + Y} a2 + ··· + Y ab + ··· Y ah, the reference value sum calculator for calculating all patents,
A random variable storage unit that stores the average E (Y _b ) and variance V (Y _b ) of the random variables _Yab before being converted to the reference value, and the total Σ (all evaluation items) Y of the evaluation item reference values.
The degree to which this Σ (all evaluation items) Y deviates from the normal distribution is calculated, one or a plurality of evaluation items are removed from the evaluation items, and Σ (all evaluation items) Y is calculated again to calculate Σ (all evaluation items). Item) An optimization calculation unit that minimizes the degree to which Y deviates from the normal distribution,
Based on the total evaluation item reference values approximated to the normal distribution, the patent power deviation value for each patent is calculated by the patent power deviation value = 50 + (Σ (all evaluation items) Y) × 10 / h ^0.5. Patent power deviation value calculation unit and
A patent power deviation value storage unit that stores the patent power deviation value,
A patented value evaluation device characterized by being equipped with. The patent value evaluation device according to any one of claims 1 to 3, further comprising a disparity input processing unit capable of inputting the disparity parameter m and changing the disparity parameter m used for calculation. .. The patent value evaluation device according to claim 1, further comprising a graph drawing unit for graphing the patent standard value in order to analyze the patent value based on the patent standard value. The patent value evaluation device according to claim 3, further comprising a graph drawing unit for graphing the patent power deviation value in order to analyze the patent value based on the patent power deviation value. The patent value evaluation device according to claim 5, wherein the graph drawing unit graphs the patent reference value over time. Multiple patents to be evaluated for patent value are used as the population, and public information published in publications, etc. regarding each patent belonging to this population is used as the evaluation item.
For each patent belonging to the population, when the content of the evaluation item is a random variable X _ab (the bth evaluation item for the ath patent), this random variable X _ab is multiplied by each patent. , The patent standard value for the a-th patent is obtained as Π (all evaluation items) X _a = X _a1 × X _a2 × ・ ・ ・ × X _ab × ・ ・ ・ × X _ah , where h is the total number of evaluation items. At the same time, when m is a disparity parameter set according to the characteristics of the technical field of the population, _Yab = log (X _ab ^m ) is calculated, the random variable X _ab is converted to _Yab , and each evaluation is performed. Calculate the mean E (Y _b ) and variance V (Y _b ) of Y _ab for the item.
The random variable _Yab is set as a reference value so that the average E (Y _b ) is 0 and the variance V (Y _b ) is 1, and the obtained evaluation item reference values are added up for each patent to obtain an evaluation item reference. Total value Σ (all evaluation items) Y _a = Y _a1 + Y _a2 + ... + Y _ab + ... Y _ah is calculated for all patents.
The degree to which this Σ (all evaluation items) Y deviates from the normal distribution is calculated, one or a plurality of evaluation items are removed from the evaluation items, and Σ (all evaluation items) Y is calculated again to calculate Σ (all evaluation items). Item) Optimized so that the degree to which Y deviates from the normal distribution is minimized.
Evaluation item A patent value evaluation method characterized by calculating a patent standard value and / or a patent power deviation value based on a total of standard values of evaluation items approximated to a normal distribution. The patent value evaluation method according to claim 8, wherein the patent reference value and / or the patent power deviation value is graphed over time. The patent value evaluation method according to claim 8 or 9, wherein the patent reference value and / or the patent power deviation value is combined with bibliographic items and keywords of patent information to perform information analysis.

Images