System Dynamics Modeling and Simulation of Enterprise Patent Management Optimization

Taking the perspective of cybernetics theory, enterprise patent management can be described as a controllable information system which includes various activities and performances. This paper aims to analysis the relationship between enterprise patent activities and patent performance, and to help convert technological innovation achievements into patent rights more effectively and consolidate technology superiority. This paper uses System Dynamics method to construct a multiscale nonlinear enterprise patent management model. Validity of this model is tested, simulation is conducted in VENSIM, sensitivity analysis of this model shows that: Technology Disclosure Sufficiency does not have much influence on patent application quantity, but it has strongly positive effect on patent grant rate and patent portfolio diversity, it can also lower the risk of patent circumvention. Simulation shows that technology disclosure sufficiency is a key determinant to reduce information loss and improve output efficiency of enterprise patent management system. Collaboration of R&D personnel, patent engineers and patent agencies should be enhanced to improve TDS, so as to promote enterprise patent management efficiency. This SD model provides a foundation for future research in patent management optimization.


I. INTRODUCTION
In the era of knowledge economy and globalization, patent legal system has become an international rule with universal binding force. Patent system provides patentee a force to legally monopolize the technology within the term of patent right. Therefore, as a special strategic resource, patent right ownership has become the key to business success especially for high-tech enterprises. To ensure that R&D achievements are effectively under protection has been the core mission of enterprise patent management, as well as a practical challenge for enterprise leader and patent manager.
Current research on enterprise patent management can involve many aspects, including the research on patent filing strategies [1], [2], function and writing skills of patent claim [3], [4], function and standards of patent examination [5], [6], patent quality analysis [7], [8], strategies and effects of patent disclosure [9], [10], patent The associate editor coordinating the review of this manuscript and approving it for publication was Muhammad Zakarya .
Since the goal of enterprise patent management is to transform the R&D achievements into patent rights timely and efficiently, in this angel of view, enterprise patent management can be described as a progress of converting technology innovation achievements into patent rights ownership. The transmission progress requires multiple participators to complete. At the same time, it is influenced by many factors. Therefore, enterprise patent management can be a case of system engineering, and it is suitable to conduct research on enterprise patent management from a perspective of system science. Thus, this paper aims to stablish a system simulation model which could reproduce the actual behavior of enterprise patent management and fill the gap in system engineering research.
Building a system simulation model often comes with two major contributions: VOLUME 9, 2021 This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/ (1) It can give researchers and decision-makers a full view of complex causal relations in the process and help better understand the whole system. In this study, building SD model can help enterprise managers understand the process of patent management with a systematic view from both qualitative and quantitative aspects.
(2) Conducting computer simulation experiments with this model can help optimize management. Model simulation and sensitivity analysis can help better understand the key variables and how the output of the system is influenced. In this study, building SD model can help convert technological innovation achievements into patent rights more effectively.
The construction of this study includes conceptualization, modellization and simulation. Firstly, we analysis the object and problem in enterprise patent management and build a conceptualization model, so as to define the boundary of this study. Secondly, causes and effects in enterprise management are analysed and shown in a causal model, and a stock-flow model is established. Thirdly, the validity of the model is tested by compare the simulate data to the actual data of the sample enterprise, and the sensitivity of main parameters in the SD model are analysed, enterprise patent management suggestions are promoted in the end of this paper. The construction of this study is as shown in Figure 1.

II. METHODS
The method of System Dynamics (SD) is adopted in this study, which can restore operating status of the system and historical data of enterprise's patent performance from both qualitative and quantitative aspects [18], [19]. SD aims to study overall behaviour of the whole system by analysing the feedback structure relationship between various variables in the social and economic system, its application scope lies in: (1) SD is suitable for dealing with long-term and cyclical problems. For example, ecological balance in nature, human life cycle and economic crisis in social problems all show periodic laws and need to be observed through a long historical stage. Many system dynamics models have made scientific explanations for their mechanisms.
(2) It is suitable for study of insufficient data. The problem of insufficient data or difficulty to quantify is often encountered in social and economic modelling research. The causal relationship between various elements of the system dynamic school record and the limited data and certain structure can still be calculated and analysed.
(3) It is suitable for dealing with complex social and economic problems with lower accuracy requirements than physical science. Social and economic systems are usually multiscale nonlinear dynamic systems, it is difficult to describe those systems using general mathematical methods, while can still be reproduced by means of computer simulation.
(4) SD model can be used to conduct conditional prediction. This method emphasizes the conditions that produce the results. Run the model in VENSIM, it can provide results of sensitivity analysis and scenario analysis, so as to help decision-making in system optimization under changing environment.
SD method was also used in researches on patent licensing and patent pool in the past [20], [21]. In this study, our purpose is to describe the dynamic behaviours of enterprise patent management system, so as to help enterprise decision-makers better understand the complexity in patent management, so that to convert technological innovation achievements into patent rights more effectively and consolidate technology superiority. Therefore, it is appropriate and necessary to adopt SD method in the study.

III. CONCEPTUALIZATION
Taking the perspective of cybernetics theory [22], we treat the research object as a whole and regard the enterprise patent management as a controllable information system. The first step is to set boundary of the system. Since the purpose of this study is to help convert technology innovation achievements into patent rights more effectively, the patent management system in this study is mainly about the activities and performance of enterprise R&D department, enterprise patent department and patent office. First of all, R&D department carries out R&D projects to the meet the enterprise's need of technological innovation; Secondly, patent department is responsible for patent drafting and patent application, this process could be done by either enterprise's own patent engineering or patent agent. Thirdly, applications are submitted to Patent Office for examination, those that meet the requirements of the patent eligibility criteria will be granted, and those that do not meet the requirements will be rejected. Valid patent grant gives a legal force to stop others from using the same technology and help consolidate enterprise's technology superiority.
Conceptualization of enterprise patent management in this study is as shown in Figure 2.

IV. MODELLIZATION A. CAUSAL MODEL
A causal model is a qualitative analysis method to understand relationship between different variables in the system. In this study, Causal loop diagram is used to describe enterprise patent management clearly from the qualitative point of view, and to analysis the causes and effects in the system. The interactive relationship among various factors is connected with arrows. If the result of the effect is positive, it is indicated by a plus (+) sign, and if it is negative, it is indicated by a minus (−) sign.
There are two important issues in the process of enterprise patent management. One issue is patent quantity. It takes efforts to convert technology innovation achievements into patent application productivity, these efforts include technology disclosure, patent mining, patent drafting, patent portfolio and patent fling. The other issue is patent quality. On the premise of the creativity, practicability and novelty of the technical achievements, patent drafting quality determines whether the application is granted or rejected, and whether it is validated or nullified in patent reviews filed by competitors. There are also risks in the process of enterprise management, including technology leak risk, novelty loss risk, patent claim risk.
Base on this, a causal model of enterprise patent management in this study is built and as shown in Figure 3. As shown in Figure 3, causal model of enterprise patent management contains 27 factors, covering the process of R&D activity, technology innovation, patent activity and patent performance. In the following paragraphs, we will analysis the cause-tree and use-tree in the model to demonstrate the causal relationship and explain the factors more clearly. It is necessary to explain why and how key variables in this model changes the output of the system in this section. As shown in Figure 3, there are complex causal relationship in enterprise patent management system, each factor of enterprise activity will have influence on enterprise patent performance. However, there are still some key relationships in this model. They will be analyzed in the form of cause tree and use tree in the following paragraphs.

1) CAUSE TREE ANALYSIS OF PATENT FILING PRODUCTIVITY
As shown in Figure 4, patent filing productivity has three causes, and each cause is influenced by 4 to 5 factors. A substantial amount of patent data is freely and publicly available, with proper method, patent mining can discover and extract knowledge from patent repositories [23], [24]. The discovered knowledge by patent mining can not only provide preliminary estimation of technology innovativeness [25], product information [26], technology trends [27], but also help with recognition of competitors and strategy construction for patent application [15], [28]. Patent portfolio strategy leads to not only more patent applications, but also significant promotion of patent drafting quality, so as to maximize value of the patents [13], [14], [29].
There are also risks to control in technology innovation and patent application, including leak risk and novelty loss risk. Novelty, creativity and practicability are the required by patent eligibility criteria, applications that do not conform to the criteria will be rejected by patent office. If the enterprise does not drift and file the patent timely, the creativity of technology might reduce along with time until competitors surpass. There is a bigger chance of technology leak if it takes too long to drift and file a patent.

2) CAUSE TREE ANALYSIS OF PATENT RIGHT OWNERSHIP
As shown in Figure 5, patent right ownership has three causes, and each of the last two causes is influenced by 2 to 3 factors. It is not a hundred percent stable for a patent after it got granted, any patent that has been granted may face validation review filed by any civil subject in the future, if the patent is not in line with the legal standards, it will be nullified. Enterprises usually tend to abandon old patents that are considered useless after new patents are granted, so as to cut down unnecessary patent renewal fee expense.

3) CAUSE TREE ANALYSIS OF PATENT EXCLUSIVE EFFECT
As shown in Figure 6, patent exclusive effect has two causes, and each cause is influenced by 2 to 3 factors. Patent claim amendment is an important part of patent drafting, it provides not only patent eligibility, but also provides legal basis to exclude others, because the technology that a patentee can own is as far as what is described in the patent claim amendments [3], [4], [10], [30], [31]. In other words, badly-drafted patent claim amendments will reduce the quality of the patent, and it will provide a bigger chance for competitor to find a way and circumvent the patent even the patent right is still valid.

4) USE TREE ANALYSIS OF TECHNOLOGY DISCLOSURE SUFFICIENCY
As shown in Figure 7, technology disclosure sufficiency has three uses, and each use influences 3 to 4 factors. Technology disclosure is a task of knowledge transmission which requires cooperation and efforts of patent inventors, patent engineers and patent agents. In this progress, the inventors should expound the technical items, technical effect and technical background entirely and detailly, so that patent engineers and agents could absorb the knowledge as needed.
Technology disclosure is a very important factor of patent management and yet normally ignored. It will promote patent engineer's understanding of creativity, practicality and novelty of the technology, which is the foundation of paten drafting, patent mining and patent portfolio. For example, sufficient technology disclosure will help pinpoint the essential technical features accurately in patent claim writing, and define clearly and concisely the scope of the requested protection. Technology disclosure sufficiency is also to meet the disclosure requirement of patent examination [32]. The disclosure requirement in patent law is designed to reveal knowledge regarding a patented invention to allow proper understanding and utilization of that invention [33]. Patent applications with insufficient disclosure are not in compliance with patent law requirements, and has a high risk of being rejected or nullified [34]. Sufficient technology disclosure inside the enterprise will also markedly improve draft quality of patent claim amendments [35], [36], and allow enterprise to form better patent strategies [1], [29], [37], [38].
In brief, technology disclosure sufficiency is a key determinant to reduce information loss and improve output efficiency of the system. The following use tree demonstrates the importance of technology disclosure sufficiency in the process of enterprise patent management. This model aims to demonstrate the process of enterprise patent activities and performance including R&D, technology innovation, patent mining, patent drafting, patent portfolio, patent application, patent examination, patent authorization, patent rejection, patent nullification and patent circumvention. Stock-flow model of enterprise patent management system is as shown in Figure 8.

C. VARIABLES
Variable selection criteria are representativeness, data comprehension, comparative objectivity, repeatability and simplicity. Model variable name, attribute, initial value and unit are set as in Table 1.

D. EQUATIONS
In this study, VENSIM is used as a computer software platform for SD modellization and simulation. Although system theory is relatively abstruse, VENSIM has provided specialized DYNAMO language, which makes it more convenient and efficient for researchers to establish mathematical relationships between variables and describe the whole system quantitatively. DYNAMO equations in this SD model are as shown in Table 2.
Using SPSS to analysis data of the sample company from 2014 June to 2021 March, equations of this SD model are finally built. This SD model contains 42 equations which are directly exporting from VENSIM. In order to present in a more readable form, we de divided these equations into three categories, including input with initial value, input equations and output equations, as shown as follows:     SD model testing methods usually include structure evaluation test, extreme case test, behavior reproduction test and so on. According to the structural behavior test method proposed by Barlas [39], whether the model structure is composed of factors related to real system is tested through structural verification and extreme case analysis. The behavior of each structural factor is tracked over time, and the amplitude and trend under extreme condition are analyzed, and the abnormal behavior is rescheduled.
The objectivity of the model is verified by comparing the model simulation results with actual data. simulation time was set as 81 months, starting from 2014 June to 2021 March, and the step length was 1 month. Zhuhai Da Hengqin Science and Technology Development Co., Ltd. was chosen as the sample company, which is a state-invested Hi-Tech startup company. Data related to the sample enterprise's patent performance are retrieved through the National Patent Database and IncoPat patent database platform. Data related to the enterprise's patent activities are mainly collected through field investigation according to real situation. Data related to enterprise's patent abilities and risks are obtained through expert consultation, statistical estimation. VENSIM software was used to run the model. Since the purpose of this model is to simulate enterprise's patent activities and performances, data smoothing method is used to modify the value of R&D projects. As shown in Figure 9, Figure 10 and Figure 11, patent filing, patent application cumulant and patent ownership is chosen as the data fitting variables. This SD model can simulate the behavior pattern of enterprise's patent filing. We haven't considered situations like employee vacancy, coronavirus epidemic, work stagnation in this model, and also due to the randomness of periods and data smoothing  method used in the system, the simulated patent filling curve does not need to be exactly identical with the actual patent application curve, but to reflect the pattern of patent filing activity. Despite some limited randomness of patent filing curve, this SD model still fits well with real data of patent application cumulant and patent ownership. In general, data fitting accuracy of true value and simulate value shows that this model is of validity and objectivity, and it is able to reproduce the activities and performance of enterprise patent management.

B. SENSITIVITY ANALYSIS
The sensitivity of main parameters in this model is analyzed in this section. There are some important variables in this SD model that have big influence on enterprise patent performance, which are analyzed in Section IV and shown in Figure 4, Figure 5, Figure 6 and Figure 7. Among them, Technology Disclosure Sufficiency (TDS) is a key determinant of reducing information loss and improving output efficiency in enterprise patent management system and yet often ignored. Adjust the value of TDS to 30%, 60% and 90% and run the model in VENSIM, the results are shown as follows: As shown in figure 12, patent application cumulant of the enterprise is not quite sensitive to the change of TDS, which is reasonable because the purpose of technology disclosure VOLUME 9, 2021 is not to promote the quantity but to improve quality of the patent. As shown in Figure 13, patent grant rate of the enterprise is quite sensitive to the change of TDS, which is reasonable because TDS will make significant improvement of patent drafting quality. At the same time, good drafting quality will make it easier for patent office to examinate the patent application files, so that patent examination period will be shortened. Therefore, higher TDS will increase enterprise's patent grant rate.
As shown in Figure 14, patent ownership of the enterprise is sensitive to the change of TDS. This result is reasonable because higher TDS will increase grate rate of the enterprise, which leads to less under-examination patents, less rejected patents and more granted patents, so that enterprise's patent ownership will be increasing.
As shown in Figure 15, patent portfolio diversity of the enterprise is sensitive to the change of TDS. This result accords with reality because higher TDS will allow patent  engineer and patent agent to do a more thorough work in patent mining, and get more information about current technologies, potential competitors, so as to make a better patent portfolio strategy.
As shown in Figure 16, patent exclusive effect of the enterprise is sensitive to the change of TDS, and become more and more sensitive along with time. This result is reasonable because higher TDS will improve patent drafting quality and better support the content of patent claim amendments, which makes competitors much harder to find a loophole or come up with a substitute technical solution to circumvent the legal force of enterprise's patent rights. A higher TDS will also allow enterprise to form better patent portfolio strategies, which could maximize the exclusive effect of enterprise's patent rights.
As shown in Figure 17, patent circumvented risk of the enterprise is sensitive to the change of TDS. This result if reasonable because higher TDS will allow enterprise to form a better patent strategy including patent drafting, filing and portfolio, so as to cut down the risk of patent circumvention.

VI. CONCLUSION
The novelty of this research can be described as follows: (1) This study stablished a valid SD model of enterprise patent management. This model includes 40 variables and 42 equations, can simulate activities of R&D, technology innovation, patent mining, patent drafting, patent portfolio, patent application, patent examination, patent authorization, and can reproduce performance of patent ownership, patent rejection, patent nullification and patent circumvention. Data fitting accuracy of the model is good.
(2) This study analyzed the importance and effects of TDS in enterprise patent management. Simulation of the model shows that TDS does not have much influence on patent application quantity, but it has strongly positive effect on patent grant rate and patent portfolio diversity. It can also lower the risk of patent circumvention and maximize the exclusive effect of enterprise's patent rights. Collaboration of R&D personnel, patent engineers and patent agencies should be enhanced to improve TDS, so as to promote enterprise patent management efficiency.
With other regulatory factors, this SD model can be used to conduct more simulation experiments and support enterprise patent management optimization. This SD model also provides a foundation for future research in enterprise patent management and decision-making.