Abstract
The author has conducted a study investigating how a company's internal reviews and CNIPA examinations affect the lifetime cost of Chinese patents. A statistical model based on the Lognormal distribution of patent value and supported by empirical data was proposed to analyze the influences. The study considers all costs incurred throughout the innovation stage to the end of patent life. The findings suggest that the lifetime cost of a good patent can be minimized by a particular combination of the internal pass rate and the official grant rate. The study also presents an approach to minimize the lifetime cost during the company's internal review.
1. Introduction
Over the past three decades, China has experienced rapid growth in technology developments, leading to a surge in the number of patents granted to Chinese domestic entities. However, this has also resulted in a substantial financial burden for some large and midsized entities due to the associated expenditures in patent application and maintenance. Additionally, nearly half of the applications filed by Chinese domestic applicants are not granted, resulting in an enormous waste of both finances and human resources. Furthermore, a substantial number of the patents granted to Chinese entities are abandoned every year, generating no return on investment. These issues have raised concerns and criticisms, not only in China but also in the international patent community [1].
Although there have been some general studies and discussions regarding the quality of Chinese patents based on published data, there is a lack of detailed analysis based on real invention processes and expenditure information in Chinese domestic entities [2, 3, 4, 5, 6, 7]. This article aims to study the lifetime cost of a Chinese patent and how two evaluation and screening processes affect it. The first process is an assessment of the invention proposal within the entity, while the second is a rigorous examination by CNIPA examiners. The article proposes a statistical model that accurately estimates the lifetime cost of a Chinese invention patent and provides an approach to minimize it. This quantitative analysis and cost-reduction approach could serve as practical guidance for many Chinese invention entities looking to improve their patent practices.
2. Patent application processes and expense categories
Fig. 1 Patent application processes and expense categories in China
Fig. 1 provides a simplified block diagram of the patent application processes and expense categories in a typical Chinese company. The diagram shows 13 processes and 8 expense categories involved in the process. The R&D staff creates inventions and drafts invention proposals which are denoted as A1 and A2, respectively. Although only two stages (B1 and B2) owned by the IPR (Intellectual Property Rights) department are shown in the diagram almost all processes require the participation of the IPR department. After an invention proposal is submitted to the IPR department through a Patent Life Cycle Management System, it is assigned to an IPR engineer who specializes in the technical field of the invention and is responsible for prior art searching. The IPR department regularly holds an invention review meeting where a batch of invention proposals are examined with the inventors, technical experts, R&D director, and IPR engineers. IPR engineers present their judgments on the novelty and inventiveness of the invention proposals. However, the judgments on the value and practical use of the inventions are usually given by technical experts and R&D managers who have in-depth technical backgrounds and business views. Quantitatively determining the value of an invention is a complex task as it involves technical, market, and legal values. Despite its ambiguity, people are starting to recognize its importance and try to evaluate it meaningfully. Once an invention has passed the internal review process, law firms are entrusted with drafting specifications and drawings. To cover the expenses associated with the drafting work at stages C1 and C2, the law firms collect a legal service fee in advance, often in the form of a lump sum for each entrusted invention application.
For the sake of simplicity, this study will focus on:
1) The costs of a Chinese invention application (no utility model that has lower fees and is not examined at CNIPA) in a regular process of patent prosecution (no PPH, re-exam, and pre-exam).
2) Inventions from domestic Chinese entities. In addition to various translation fees, overseas applicants are unable to receive a 70% discount on CNIPA fees, which are roughly six times more expensive for each patent application.
3) Cumulative patent costs in the ninth year after the grant. Several academic discussions have proposed a so-called most economical patent lifetime defined by the time when the expected economic return equals the cost of ongoing maintenance [8, 9, 10]. The choice of a nine-year lifetime in this study is mainly due to the following three reasons. First, CNIPA annual maintenance fees increase significantly from the 10th year. Second, as core technology patents held by foreign technology pioneers are approaching expiration, those Chinese patents that have been deployed around the core technology patents in an attempt to have strategic deterrent effects are losing their value of existence. Third, due to rapid technology development and breakthroughs, particularly due to the emerging AI, the technology lifecycle is becoming shorter and shorter.
3. Two blind spots in the estimation of patent costs
Estimations of patent cost frequently have two blind spots:
1) The survivorship bias, which only considers the costs of applications that were successfully granted as patents, ignores all the costs incurred on applications that were withdrawn, abandoned, or rejected.
2) A significant number of patents are of little value or vulnerable against invalidity challenges, which are referred to as bad patents hereinafter. We need to reallocate all the costs associated with bad patents to good patents that can withstand invalidity challenges and truly have value in technological, economic, and legal aspects.
These two blind sports will be correctly addressed in this study.
4. Modeling of prior-art searching
Fig. 2 Statistics of searching time in a Chinese display enterprise and a fitting curve of lognormal distribution
When developing an invention proposal, a crucial task is searching for prior art. Fig. 2 shows the statistics of searching time in a Chinese display enterprise and a fitting curve of lognormal distribution. Although the median searching duration is about 13 days, the net searching time dedicated to one invention proposal might be a fraction of that duration, because IPR engineers and inventors usually work on multiple projects in parallel. The integral of the statistical distribution in Fig.2 is called cumulative distribution. It is easy to prove that this cumulative distribution can be well approximated by an exponential function of a time variable,
(1)
where SR (%) is defined as the probability of detecting a prior art of the invention proposal, and therefore (1-SR) is the probability of detection failure. The more time IPR engineers spend on searching, the higher the probability that IPR engineers find the prior arts, following an exponential rule described by search theory [11]. In the above formula, SR0 is the upper limit of the detection probability, determined by the searching capability possessed by the IPR engineer, tS is the searching time, and t0 is a time constant. Assume that searching for two days will achieve 80% of the upper limit SR0, then for a given detection probability SR, the required searching time is derived from (1) as
(2)
For SR0=0.95, searching five days will result in 6.7% detection failure. Longer searching times, of course, will incur more labor costs and risk being preempted by competitors. The IPR manager needs to make an optimal decision by taking into account the search time, detection probability, future grant rate, competitive risks, and overall costs.
Unfortunately, many invention applicants in China do not conduct thorough searches for prior art before their filing in CNIPA. This is because they believe that as long as they entrust their cases to patent agencies, there is no need for them to do any additional work. However, the issue is that patent agencies charge based on the number of cases entrusted, unlike in the USA where patent attorneys charge based on the number of hours spent. Conducting a prior art search takes time and may reduce the chance of drafting a specification, which in turn affects the income of the patent agents.
5. Modeling of Invention Value
Table 1 A scorecard of an invention proposal
Table 1 is a scorecard that is used to evaluate an invention proposal. Chinese Patent Law [12] mandates that three basic attributes, novelty, inventiveness, and practical use, must be evaluated. However, internal review also comprehensively evaluates the invention's value from technological, economic, and legal aspects. The total score determines whether the invention will be filed for a Chinese patent only, or both Chinese and overseas patents, or filed for a utility model patent, kept as a technical secret, held on for improvement, or rejected. Studies have shown that the value of patents follows a log-normal statistical distribution [13, 14, 15, 16, 17], where the variables are greater than zero, and the distribution curves are skewed to the left side and have heavy tails on the right side.
The probability density function (PDF) of a lognormal distribution adopted in this study is expressed by
(3)
where x represents the total score as shown in Table 1, ranging from x0 to 100. The symbols μ and σ are two parameters, with which the mathematic expectation and variance of the lognormal distribution can be obtained by
(4)
and
(5)
Fig. 3 illustrates a histogram that displays the technology value of 26,700 Chinese invention applications, which were granted to over 40 domestic entities from April 2014 to April 2024 and were analyzed by a commercial search engine PATSNAP. A fitting curve with the lognormal distribution is plotted in Fig.3 as well. Except for some deviations in high scores, the statistical distribution of the technology value of China's invention applications can be well described by the lognormal distribution. Notably, no data points were found below 20, which is perhaps due to the imperfection of the algorithm in the patent value model of the search engine.
Fig. 3 Histogram of technology value of Chinese inventions
The technology values of the invention proposals shown in Fig. 3 are not exactly equal to the scores given in Table 1. However, the genetic characteristics of invention proposals are strongly correlated with grant rates and subsequent assessments of patent value during their lifetime. Inventions that score high value may be entrusted to better patent attorneys for drafting specifications and preparing responses to examiners, which may lead to a higher value and strong patent against invalidity litigation (positive assessment leads to positive outcomes). It is therefore reasonable to assume that the statistical characteristics of the invention values were maintained through the two examinations and after patent rights were granted.
6. Empirical data and Modeling
Many studies [13, 14, 15, 16, 17] have confirmed that the higher the patent value, the longer the patent lifespan. A survival probability function SP can be composed by an integration of (3)
(6）
The larger the variable x, the longer a patent survives. Therefore, (6) is adopted to describe the probability of a good patent. On the contrary, the probability of short-lived or bad patents can be expressed as
(7)
The survival ratio of patents can be found by
(8)
Because there is a direct correlation between patent value and its lifespan, a distribution of the survival ratio as a function of time can be obtained with (8) by mapping the patent value x (from 0 to 100 points) to a time coordinate (from 0 to 20 years).
Fig. 4 Survival curve of life expectancy of Chinese patents
Fig. 4 shows the statistics of the life span of Chinese invention patents calculated from the grant date. Unlike Fig. 3, the data in Fig. 4 are not limited to one technology, so the influence of a certain technology lifecycle (the bell curve) is averaged out. The discrete data points represent the average survival ratio of Chinese invention patents (filed and owned by Chinese domestic entities) granted from 2004 to 2016. The solid line is a fitting curve using (8). Except for some deviations near the end of a 20-year lifetime, the actual survival ratio essentially follows the decay trend given by (8). Notably, only 50% of the Chinese invention patents survive in the ninth year.
Fig. 5 Histograms of examination duration
Fig. 5 depicts a set of histograms and the corresponding fitting curves that demonstrate the delay time of 10,000 Chinese invention applications that were filed by BOE Technology and examined between January 2015 and September 2017. The median delay time between filing and examination was 3.5 months, while the median duration of examination for granting was 26.5 months, resulting in a 30-month examination cycle time. Moreover, the median duration from examination to rejection was 38 months, as there are usually several rounds of arguments before the final rejection. These time delays will be used for calculating the compound interests as one of the components of patent lifetime cost. For instance, in the ninth year after the grant, 11.5 years would have passed since the filing expense was incurred.
The time delay from filing to examination follows a narrow normal distribution due to a semi-auto prosecution process. However, an examination comprises several subroutines involving human tasks and interactions. Each subroutine is subject to random behavior from participants such as examiners, corporate IPR staff, and inventors. Hence, the examination duration follows a Cauchy distribution with a heavy tail, similar to travel durations on public transportation [18, 19].
In a diversified enterprise, its inventions may involve multiple technologies and its R&D staff may be located in different cities. Additionally, examiners for the enterprise may come from different technology groups and regional examination centers. As of September 2023, there were eight regional examination centers and over 20,000 examiners in mainland China. Due to the technical complexity and the number of human resources involved, the statistical distribution of the examining duration may have a complex shape or even exhibit multiple peaks.
7. The concepts of good and bad inventions
Generally, the more valuable and stable an invention is, the more likely it is to withstand three tests: corporate internal review, CNIPA examination, and invalidity litigation, and the longer it will last. Therefore, these inventions are defined as good inventions, and vice versa as bad inventions.
Fig. 6 Number of applications examined and granted
Fig.6 plots the number of invention applications filed by Chinese domestic entities and examined from the first half (H1) of 2010 to the second half of 2021. Also plotted in Fig. 6 is the grant rate at the same time. The grant rate is defined as the ratio of the number of granted patents to the number of cases closed in the same duration. Over the past decade, as the application quantity has grown rapidly, the corresponding grant rate has dropped significantly. On average, the grant rate is slightly lower than 50%.
Fig. 7 Number of invalidation suits and success rate
Fig. 7 shows a statistics regarding invalidation suits in China. The success rate is defined as the ratio of the number of patents that either all claims are rejected or voluntarily abandoned to the total number of invalidation suits. From January 2013 to January 2023, a total of 1,691 Chinese invention patents were sued for invalidity, of which 1,019 were eventually judged all the claims invalid or abandoned. Averaged over the past decade, the invalidation success rate of Chinese invention patents is approximately 60%. In other words, the survival probability against invalidation litigation is barely 40%.
Multiplying the 40% survival probability by the 50% grant rate yields 20%, which suggests that only 20% of the applications submitted to CNIPA could become good patents. Assuming that the pass rate of the corporate internal review is 75%, then the proportion of good invention among all the invention proposals, denoted by GY0 (initial good yield) hereinafter equals 40% x 50% x 75% = 15%. GY0 is an important parameter to describe the overall quality of the invention proposal. In other words, only 15 out of 100 invention proposals may have adequate value for subsequent patent applications. Moreover, market competitors would not waste resources to invalidate a junk patent that is of no harm or no value to them. Therefore, the proportion of patents that are truly valuable and legally stable is even smaller.
8. Probability distribution of good and bad inventions
Multiplying ES and SP will result in a probability distribution of good inventions. Similarly, multiplying ES and SQ will result in a probability distribution of bad inventions. These mathematical operations are expressed by
（9）
and
（10）
The functions GD and BD are plotted in Fig. 8 on the right and left sides, respectively. XT is a threshold that invention proposals scored greater than XT will be entrusted to a law firm for drafting the specification, whereas those scored less than XT will be put on hold or simply rejected. Therefore, XT will determine the internal pass rate and the proportion of good inventions among all inventions that pass the review.
Fig. 8 Values of good and bad inventions
However, as shown in Fig. 8, a small tail on the left side of the distribution of the good invention is mistakenly rejected, whereas a small tail on the right side of the distribution of the bad invention is mistakenly allowed. The former is called a false positive, and the latter is called a false negative.
Integrating GD over the entire region results in a proportion of good inventions in the beginning, or the initial good yield GY0 as
(11)
Integrating BD and GD from XT to infinity results in the total percentage of bad inventions P1 and the total percentage of good inventions P2, respectively.
(12）
and
(13)
Therefore, a rate SRN, representing screening efficiency for bad inventions, can be expressed by
(14）
Based on the author's experience, the majority of invention proposals rejected internally were due to similar prior arts found by IPR searching engineers. To determine the required searching time, SR in (2) can be replaced with SRN given in (14). The sum of P1 and P2 is equal to the internal pass rate PR1(XT), that
(15）
for a given PR1, the lower limit of the integral XT can be solved numerically.
The examination conducted by the CNIPA examiner may not follow a similar reviewing protocol as shown in Table I. However, an examination with a semi-quantitative scoring system is highly possible. According to the law of large numbers in statistics, the outcomes of a large number of examinations should exhibit a stationary statistical distribution as well. For simplicity, it is assumed that it still follows a lognormal distribution but with different parameters μ2 and σ2, and is expressed by
（16）
The coefficient γ is a proportion of the filed applications that are selected by applicants to be examined. The average γ value from 2020 to 2023 is 86%. The probability density functions of good invention and bad invention are expressed in the following, respectively,
（17）
and
（18）
Assume that XS is a threshold for screening the inventions to determine whether to grant patent rights or not. By integrating GD2 and BD2 from XS to infinity, the total number of granted patents is expressed by
(19）
The integral includes only the quantity of the good patents for β=0. Hence the proportion of the good patent, defined as the second good yield GY2, is expressed by
（20）
The grant rate PR2(XT, XS), is given by
(21)
For a given value of PR2, the threshold XS can be solved by numerical calculation with (21).
9. Expenses at the invention-creation stage
The first part of the patent cost CP1, which is incurred from stage A1 to stage B2 as shown in Fig. 1 and inflated by a compound interest coefficient RF1 over the years, is expressed by
（22）
In a Chinese domestic enterprise specializing in the development and manufacture of electronic components, the overhead coefficient of cost ranges from 1.2 to 1.5. The overhead coefficients OH1 and OH2 in (22) correspond to variable cost and fixed cost, respectively. NWD is a networking day per year (230 days), calculated by subtracting training and meetings (12 days) and personal leave (7 days) from the stipulated working days in China (250 days). PT and RT are the times spent by inventors and technical experts in stages A1~B2 respectively. CIT, CST, and CRT are the average salaries of inventors, IPR engineers, and technical experts, respectively. CFIX is the fixed cost shared by each invention proposal. According to (2), the cost CP1 is a function of prior art detection probability SR.
10. Expenses at the application stage
In stages C1~C2 and D1~D3, at least three expenses will be incurred: the invention incentive CB1 awarded to inventors, the legal service fee CW paid to law firms, the filing fee CR1, and the examining fee CR2 collected by the CNIPA office. The sum of these costs denoted as CP2, will increase over the years by the compound interest coefficient RF1, and is expressed as
(23)
11. Expenses after the patent grant
The total cost from the grant to the end of the life is given by
(24)
where the factor compound interest coefficient RF2 is smaller than RF1 because of the 2.5-year delay from filing to grant as revealed by the statistical data in Fig. 5, CM is the cumulative maintenance cost that is affected by compound interest as well, factor η is determined by the patent survival rate. Finally, the lifetime cost is expressed by
(25)
where SP is the survival ratio of the patents after 9 years, and the two variables XT and XS are determined by (15) and (21), respectively.
12. Calculation results and discussions
Table 2 Various costs used in the calculation
Assuming that the average annual bank interest rate is 3%, the following variables are considered in the calculation: SP=50%, η=66%, and GY0=15%. In Table 2, various costs are listed in the unit of CNY. Please note that the labor costs listed in Table 2 are average numbers in the display industry in 2023. Other technology fields or industries, such as semiconductor and software industries, may have higher engineer salaries and legal service fees. CP1 (0.9) is defined as the labor cost to create an invention proposal with a 90% probability of being a good invention. Please note that a 70% discount rate is already applied to all official fees.
Fig. 9 Patent Cost in 9-th year
The cost of a patent in its 9th year after being granted is calculated from equation (25) and is shown on a contour plot in Fig 9. The numbers on the contour lines indicate the total cost in units of 10K CNY. However, to make a fair judgment, it is important to consider the patent quality as well.
Fig. 10 Good patent yield GY2
Fig. 10 shows a contour plot of GY2, which represents the proportion of good patents and is an indicator of patent quality. In both figures, a specific location is marked by dotted lines, where the internal pass rate is 40%, the grant rate is 85%, the patent cost is 148,000 CNY, and the good patent yield is 44%.
To improve the quality of patents, the grant rate must be reduced to 30% even if it means an increase in the total patent cost to 400,000 CNY. Any attempts to reduce costs by increasing the internal pass rate or grant rate will result in a decrease in patent quality. Unfortunately, there is no straightforward solution to this problem. The ideal resolution lies in monetizing the patent's value and maximizing ROI (return on investment) on the intellectual property. To avoid uncertainty in pricing patents or inventions, this study will focus on evaluating the cost of a good patent. This is because the cost of a good patent is a reasonable measure of ROI.
13. The actual cost of a good patent
Consider the following hypothetical scenario: when a patent holder decides to abandon some of their patents, they only abandon the bad ones and keep the good ones. Let β = 0 and SP = 1 in the patent cost formula (25), which means that all the costs of bad patents are reallocated to the good patents. The result is shown in Fig. 11 as contour lines.
It is noticeable that the contour line group in Fig. 11 reveals an oval basin area, where the patent cost is the lowest (~174,000 CNY), at the same location marked in Fig. 9 and Fig. 10 where the internal pass rate is 40% and the grant rate is 85%, respectively. Along any horizontal line across the vertical axis, a point with the minimum patent cost can be found. Connecting these points forms the dotted line illustrated in Fig. 11, which is called a low-cost line hereinafter. Alternatively, connecting all the tangent points of the contour lines and horizontal lines will form the low-cost line as well. The patent cost will increase as the coordinates depart from the low-cost line. No entity can determine the grant rate according to its wishes, but the IPR managers in the entity should be able to control their internal pass rate to make the patent cost approach the low-cost line as marked by the dotted line in Fig. 10.
Fig. 11 The actual cost of a good patent in 9-th year
It can be seen from the 175,000 CNY contour line in Fig. 11 that only if the internal pass rate is between 35% and 45%, well below the official grant rate between 75% and 95%, can the cost of a good patent reach its minimum level. The reasoning behind that is like what happens in a product inspection: manufacturers intend to screen out defective parts as early as possible because the more defective parts flow into the backend process, the greater the losses to the manufacturers. Therefore, from a cost perspective, the internal review should be more stringent than what CNIPA examiners do.
Fig.12 Low-cost line and IPF curve
Two curves are plotted in Fig. 12. The dotted line is the low-cost line as shown in Fig. 11, and the solid line is an IPF curve (inverse proportional function) along which the product of the grant rate and the internal pass rate equals 15%. Laminating the IPF curve in Fig. 10 and FIG. 11, it is found that the good patent yields along the IPF curve are more than 65%, but the cost of a good patent increases to 210,000 CNY. Therefore, internal review with overly stringent criteria may not be worthwhile from a cost perspective because rejecting a good invention (false positive) is more costly than allowing a bad one (false negative).
Fig. 13 Patent cost versus internal pass rate
However, the reality is that the average grant rate has been around 50%, as shown in Fig. 6, whereas the average internal pass rates in most Chinese domestic entities, if there are meaningful internal reviews, are around 70%. As illustrated in Fig. 13, the average patent cost is 20,000 CNY more than the lowest cost value. Considering there are 0.82 million Chinese invention patents granted in 2023, the potential savings in capital and talent resources are enormous.
14. Minimum patent cost versus initial good yield
Fig. 14 Minimum cost versus initial good yield
Fig. 14 plots the minimum cost and the final good yield GY2 as functions of the initial good yield GY0. When GY0 increases from 15% to 25%, the minimum cost decreases by 50,000 CNY and the final good yield increases from 43% to 54%. For a Chinese entity holding over 1000 Chinese invention patents, it may save at least 27 million CNY in nine years after the grant.
Therefore, corporations and universities in China (in 2023, about 70 domestic entities including 44 universities were granted more than 1,000 Chinese invention patents) should stop the practice of using the invention quantity as a KPI (Key Performance Index) for their employees and faculties and put more efforts into improving invention quality, to avoid wasting huge capital and talent resources. Moreover, local governments and CNIPA should withdraw subsidy policies based solely on the patent quantities.
15. Conclusions
The author of this work conducted a study on the impact of two evaluation processes of a Chinese invention on the cost of patent lifetime. The first process involves a comprehensive assessment done within the application entity, while the second process is an examination conducted by CNIPA examiners. The author proposed a statistical model to analyze patent lifetime costs and introduced the concepts of good and bad inventions. The study analyzed over 26,700 Chinese patents and confirmed that the value and life expectancy of Chinese invention patents follow lognormal distributions, which are used in the proposed model. In the proposed model, the costs of bad patents and costs incurred on invention applications that were not granted as patents are reallocated to the lifetime costs of good patents. The calculation results based on the proposed model revealed that the lifetime cost of a good patent reaches its minimum level for a certain combination of the internal pass rate and the official grant rate. The methodology of assessing invention cost and the approaches to optimizing the invention process presented in this article are useful not only to increase the ROI of patenting innovation but also applicable to licensing or purchasing patent rights in trading intangible assets and in business mergers and acquisitions.
References
1. Michael D. Frakes & Melissa F. Wasserman, “Does the U.S. Patent and Trademark Office Grant Too Many Bad Patents? Evidence from a Quasi-Experiment”, 67 Stanford Law Review, pp. 613-676, 2015.
2. M. Liang, “Chinese Patent Quality: Running the Numbers and Possible Remedies”, UIC Review of Intellectual Property Law, Vol.11, Issue 3, 478, 2012.
3. D. Yang, “Intellectual Property System in China: A Study of the Grant Lags and Ratios”, Vol. 10, Issue 1, January 2007, pp. 22-52.
4. X. Li, “Behind the recent surge of Chinese patenting: An institutional view”. Research Policy, 41(1), pp. 236–249, 2012.
5. K. Motohashi, “Assessment of technological capability in science industry linkage in China by patent database”, World Patent Information, 30(3), pp. 225–232, 2008.
6. G. Thoma, “Quality and value of Chinese patenting: An international perspective”. Seoul Journal of Economics, 26(1), 2013.
7. G. Zhang and X. Chen, “The value of invention patents in China: Country origin and technology field differences”. China Economic Review, 23(2), pp. 357–370, 2012.
8. A. Pakes and M. Schankerman, “The rate of obsolescence of patents, research gestation lags, and the private rate of return of research resources”, in R&D, Patents and Productivity, edited by Z. Griliches, 98-112. Chicago, Chicago University Press, 1984.
9. V. Denicolo, “The optimal life of a patent when the timing of innovation is stochastic”, Department of Science and Economics, University Bologna.
10. Y. Choi and D. Cho, “A study on the time-dependent changes of the intensities of factors determining patent lifespan from a biological perspective”, World Patent Information, Vol.54, Pages 1-17, Sep. 2018.
11. L. Stone, “Search theory: a mathematic theory for finding lost objects”, Mathematics Magazine, Vol. 50, No.5, Nov.1977, pp. 248-256.
12. Patent Law of People’s Republic of China, https://english.cnipa.gov.cn/col/col3068/index.html.
13. J. Barney, “Method and system for valuing intangible assets”, US 7657476 B2, filed Dec.21, 2006.
14. G. Silverberg and B. Verspagen, “The size distribution of innovations revisited: an application of extreme value statistics to citation and value measures of patent significance”, Maastricht Economic Research Institute of Innovation and Technology, August 2004.
15. F. Scherer, “The size distribution of profits from innovation”, The International Conference on the Economics and Econometrics of Innovation, Strasbourg, June 3, 1996.
16. A. Gambardella, D. Harhoff and B. Verspagen, “The value of European patents”, Discussion Paper No.6848, Centre for Economic Policy Research, June 2008.
17. Y. Xie and D. Giles, “A survival analysis of the approval of U.S. patent applications”, University of Victoria, B.C. Canada, Aug. 2007.
18. D.Ettema, O. Ashiru and J. Polak, “Modeling Timing and Duration of Activities and Trips in Response to Road-Pricing Policies”, Journal of the transportation research board, Vol.1894, Issue 1.
19. A. Pawelek and P. Lichota, “Arriving air traffic separations generalized model identification”, Technical Sciences, Bulletin of the Polish Academy of Sciences, Vol. 70 (2), 2022.
