KR102632595B1 - Method and apparatus for analyzing patent - Google Patents

Abstract

A patented analysis method and device are disclosed. A patent analysis method according to an embodiment includes the steps of characterizing the network structure of backward citations cited by the target invention, and analyzing and providing the impact of the characterized network structure on technological impact or predictability of future value. Includes.

Description

Patent analysis method and apparatus {Method and apparatus for analyzing patent}

The present invention relates to information retrieval and a database structure for the same. More specifically, it analyzes the mutual citation relationship of patents related to the target invention to be searched from a patent database containing a field containing patent information of prior inventions cited in the patent application. It is about the technology provided.

Inventing is the process of searching and recombining various flows of knowledge. The complexity of today's technological innovation requires the fusion and recombination of more and more knowledge, and from this perspective, the ability to explore the potential of prior inventions and combine them with new ideas is very important for corporate innovation. In this sense, technological innovation is like a continuous process of 'search and recombination', and the value of the created technology depends on the characteristics of the searched prior art and the inventor's own recombination ability.

Until now, many scholars have used patent citations as a major tool to evaluate the value and economic impact of inventions. However, most existing models rely on the individual citation information listed in the patent specification and do not rely on the interrelationship among citations, which serves as a useful source of information for understanding how new inventions are created from cited prior art. has received less attention.

According to one embodiment, we propose a patent analysis method and device that analyzes and evaluates the value of a patent ex ante using a network structure of backward citations.

A patent analysis method according to an embodiment includes the steps of characterizing the network structure of backward citations cited by the target invention, and analyzing and providing the impact of the characterized network structure on technological impact or predictability of future value. Includes.

In the characterization step, at least one of constraints, cohesion, and efficiency can be characterized as network characteristics of backward citations.

In the characterization step, a direct constraint representing the relationship ratio between target invention i and cited invention j and an indirect constraint representing the sum of all indirect paths through which target invention i can reach cited invention j through other cited invention q are used. Thus, the constraint variable can be calculated.

In the characterization stage, the cohesion variable can be calculated using the ratio of the actual connections that exist between the cited inventions to the maximum number of possible ties (ties) between the cited inventions for the target invention.

In the characterization step, an indirect constraint representing the sum of all indirect paths through which target invention i can reach cited invention j through other cited invention q is given, and an indirect constraint is given depending on whether cited invention j is connected to cited invention q. Efficiency can be calculated using binary values.

In the analysis and provision stage, the technical or economic value of the target invention is calculated using network variables and control variables generated by backward citation, and the network variable is of the nth order, including at least one of constraints, cohesion, and efficiency. (n is a positive integer) represents the network characteristics, and the control variables are the number of backward citations, the number of Cooperative Patent Classification (CPC) codes of backward citations, recency and time distribution of backward citations. It may include at least one of (time spread) and number of claims.

In the analysis and provision stage, the impact can be analyzed and provided by separating applicant citations and examiner citations.

A patent analysis device according to another embodiment includes an input unit that receives information on the subject invention subject to patent analysis, and a network structure of backward citations cited by the subject invention, and the characterized network structure determines the technical impact or future value. It includes a control unit that analyzes the impact on predictability and an output unit that provides analysis results.

The control unit may characterize at least one of constraints, cohesion, and efficiency as network characteristics of backward citation.

The control unit calculates the technical or economic value of the target invention using network variables and control variables generated by backward citation, and the network variable represents nth-order network characteristics including at least one of constraints, cohesion, and efficiency. , the control variable is at least one of the number of backward citations, the number of Cooperative Patent Classification (CPC) codes of backward citations, the recency and time spread of backward citations, and the number of claims. It can be included.

According to a patent analysis method and device according to an embodiment, the present invention expands the analysis method for preliminary valuation of patents and inventions by utilizing a secondary backward citation network. Additionally, the patent analysis method and device according to one embodiment provide the first empirical evidence on the relationship between various network measures of backward citations and the future value of patents. Additionally, our model is valuable for both academic and practical purposes. Companies and policy makers can apply our model to identify target technologies or predict promising technology areas. This model allows companies to select target patents to purchase or select strategic R&D projects to pursue in response to competitors' technological innovations and patent strategies. In other words, network analysis of patents can be used to build strategic positioning for a company's R&D investments.

1 is a diagram illustrating the flow of a patent analysis method according to an embodiment of the present invention;
Figure 2 is a diagram illustrating three examples of an ego network according to an embodiment of the present invention;
Figure 3 is a diagram illustrating an ego network to explain the network characteristics of backward citation according to an embodiment of the present invention;
Figure 4 is a diagram showing the configuration of a patent analysis device according to an embodiment of the present invention.

The advantages and features of the present invention and methods for achieving them will become clear by referring to the embodiments described in detail below along with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below and may be implemented in various different forms. The present embodiments are merely provided to ensure that the disclosure of the present invention is complete and to provide common knowledge in the technical field to which the present invention pertains. It is provided to fully inform those who have the scope of the invention, and the present invention is only defined by the scope of the claims. Like reference numerals refer to like elements throughout the specification.

In describing the embodiments of the present invention, if it is judged that a detailed description of a known function or configuration may unnecessarily obscure the gist of the present invention, the detailed description will be omitted, and the terms described below will be used in the embodiments of the present invention. These are terms defined in consideration of the function of , and may vary depending on the intention or custom of the user or operator. Therefore, the definition should be made based on the contents throughout this specification.

The combination of each block in the attached block diagram and each step in the flow chart may be performed by computer program instructions (execution engine), and these computer program instructions can be installed on a processor of a general-purpose computer, special-purpose computer, or other programmable data processing device. Since it can be mounted, the instructions executed through a processor of a computer or other programmable data processing device create a means of performing the functions described in each block of the block diagram or each step of the flow diagram.

These computer program instructions may also be stored in computer-usable or computer-readable memory that can be directed to a computer or other programmable data processing device to implement a function in a particular manner, so that the computer-usable or computer-readable memory The instructions stored in can also produce manufactured items containing instruction means that perform the functions described in each block of the block diagram or each step of the flow diagram.

In addition, computer program instructions can also be mounted on a computer or other programmable data processing device, so a series of operation steps are performed on the computer or other programmable data processing device to generate a process that is executed by the computer or other programmable data processing device. Instructions for performing a data processing device may provide steps for executing functions described in each block of the block diagram and each step of the flow diagram.

Additionally, each block or each step may represent a module, segment, or portion of code containing one or more executable instructions for executing specified logical functions, and in some alternative embodiments, the blocks or steps referred to in the blocks or steps may represent a portion of code. It should be noted that it is possible for functions to occur out of order. For example, it is possible for two blocks or steps shown in succession to be performed substantially at the same time, and it is also possible for the blocks or steps to be performed in the reverse order of the corresponding functions, if necessary.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. However, the embodiments of the present invention illustrated below may be modified into various other forms, and the scope of the present invention is not limited to the embodiments detailed below. Examples of the present invention are provided to more completely explain the present invention to those skilled in the art.

Figure 1 is a diagram showing the flow of a patent analysis method according to an embodiment of the present invention. Referring to Figure 1, the patent analysis device characterizes the network structure of backward citations of the target invention (110 ). Network structure can be characterized by variables such as constraints, cohesion, and efficiency.

Subsequently, the patent analysis device analyzes the impact of the characterized network structure on the technological impact or predictability of future value (120) and provides analysis results (130).

Inventing is the process of searching and recombining various flows of knowledge. The value of an invention is related not only to the number of prior arts or inventions cited, but also to how these prior arts or inventions are connected to each other. The present invention expands the analysis method for ex ante valuation of patents by using network analysis techniques widely used in social capital theory. As an example, using US pharmaceutical industry patent data, we analyzed whether the network characteristics of backward citations of prior art cited in patents have a significant impact on the value of patents and inventions. The results of the empirical analysis show that the network characteristics of backward citations measured by constraints, cohesion, and efficiency have a statistically significant correlation with both the level of patent value and the depreciation rate.

The present invention takes advantage of the fact that in social capital theory, the performance of the ego is determined not only by its own characteristics but also by its relationships with alters. Social capital theory emphasizes that the relationships between egos and alters should be viewed as part of a larger network of relationships formed by interactions between alters and other alters. We take this social capital theory as our theoretical framework to understand whether the structure of prior art (as identified by the citation network among cited inventions) is related to the future value of inventions. In particular, we focus on the trade-off between dense prior knowledge sets and recombining disparate existing knowledge sets. Patents based on similar prior knowledge may generate redundant information, so there is a high risk of being replaced by many similar alternatives, and the lack of originality of knowledge has a limited impact on the value creation of subsequent patents. However, conversely, the similarity of existing knowledge cited can have advantages in terms of technology ramifications. This is because if inventions rely on a dense knowledge system, the burden of combining heterogeneous knowledge can be alleviated and the reliability and adaptability of the created knowledge can be increased. Empirical evidence from existing research on these conflicting views is rare, and the present invention presents empirical evidence for the first time and a new level of analysis method for this issue.

For empirical analysis, we take US pharmaceutical patent data from 2001 to 2005 as an example. Referring to network measurements from social network analysis, we introduce three network variables - constraints, cohesion, and efficiency - that can effectively characterize the structural patterns hidden in citation relationships between prior inventions, and examine how they are related to the value of patents. . The variable is calculated from the backward citation ego network of each patent generated by first and second backward citations (i.e., citations between cited documents) and further by nth backward citations. Considering that applicant citations are more related to knowledge flow inventions and examiner citations are more related to the private value of inventions, we separate applicant citations and examiner citations in the network variable measurement.

The value of an invention is defined from two perspectives: the private value to the owner and the social impact or technological spread of subsequent inventions, and is measured by forward citations and patent renewal. In the case of forward citations, two variants are considered: simple citations and survival probability to investigate the level and depreciation rate of patent value. To estimate the dependent variable, negative-binomial model, Cox-Hazard ratio model, log-normal survival probability, and log-normal survival probability were used. model) is used.

Empirical results show that the structural characteristics of prior art have a statistically significant correlation with patent value. For example, the three network variables of constraints, cohesion, and efficiency have statistically significant effects on the number of citations and the rate of citation decline in both levels and depreciation rates. The estimated effect is significant. For example, all else being equal, a 1% increase in constraints reduces the expected number of examiner citations by 7.20% (7.09% for total citations) and increases the examiner citation dropout rate by 4.85%. (6.91% of total forward citations). The network structure of prior art citations also shows a significant correlation with the possibility of patent renewal. When the network constraints of prior art increase by 1%, the odds of renewal of a patent decrease by 2.58%. These results indicate that the network structure of backward citations holds important information in predicting the future value of patents, and its absence leads to severe overestimation or underestimation.

The main problem of the present invention is to characterize the network structure of patent prior art in a multidimensional manner and analyze and provide whether the prior art network structure affects the technological impact or predictability of future value. The present invention is motivated by structural social capital research using a network approach, where structural relationships in networks are a key element in understanding behavior and performance.

Empirical Model

The main goal of the present invention is to characterize the network structure of the prior art cited in the patent for the invention and to ensure that the characterized network structure provides useful information for estimating the value (ex-post value) of the patent. Equation (1) represents the regression equation used in the present invention. dependent variable represents the ex-post value of patent i at t + t assigned to inventor k in year t . The prediction time window t is a fixed constant for all t to solve the truncation problem over time. NW represents network variables created by backward quoting, and is one of the main concerns in the present invention. The network features of backward citation represent nth order (n is a positive integer) network characteristics including at least one of constraint, cohesion, and efficiency. Where , including the number of claims. represents the patent class dummy variable represents the year fixed effects dummy.

(One)

In addition to examining the impact on patent value, we also perform survival analysis to analyze and provide depreciation rates of patent value. A decline in patent value over time indicates the obsolescence of the technology. For survival analysis, the Cox hazard model and parametric survival model with lognormal distribution are used. Survival analysis estimates the hazard rate of termination of citation at a specific time t. The risk ratio, called the risk ratio, is modeled as follows:

(2)

In equation (2), H ₀ (t) is the baseline risk when all independent variables are 0. This equation estimates the impact of network variables on the ratio of risk of citation termination relative to baseline risk.

Measurements for patent value

As a method of measuring the patent value index, it will generally be the simplest method if the monetary value of the patent can be reasonably quantified. This is because the ultimate goal of promoting invention is to pursue financial returns. However, accurate assessment of expected financial returns is not easy. Commercialization of inventions often involves a multi-step process and is subject to highly uncertain market environments. Therefore, most valuation approaches exhibit significant bias in statistical analysis.

In the present invention, the value of the invention is defined from two perspectives: 'private value for owners' and 'social impact or technological spillover on subsequent inventions'. To measure the ‘private value of a patent,’ examiner citations and patent renewals are taken into account. Examiner citations better reflect the private value of a patent. Additionally, patent renewal data provides direct information on the private value of ownership protection because the renewal data records the payment record of the periodic renewal fee that the patentee must pay to keep the patent right valid.

From a recombinant perspective on inventions, the social value of patents consists of the ripple effect on subsequent inventions. The ripple effect of an invention can be measured by the degree to which it influences subsequent inventions. A good proxy for quantifying the extent of leakage is the number of citations. Citations have been widely accepted as a successful proxy variable for estimating the impact of patents on future technology. According to this tradition is measured as the total number of citations that patent i received during a given period t. Because we consider social value to encompass a wide range of aspects that impact both applicants and examiners, we do not isolate examiner citations here. This measure of social value effectively complements the use of examiner citation counts.

Additionally, we analyze the survival risk ratio of citations as an additional dependent variable. Risk ratio is interpreted as the likelihood of an event occurring compared to the baseline risk. In the present invention, the survival risk ratio represents the risk that citation will cease in the next period. To complement proportional hazards estimation, survival probability is estimated by assuming that survival time follows a lognormal probability distribution. Proportional hazards models provide only the proportional ratio of the effect of a covariate on the baseline risk, whereas hazard models using a lognormal probability distribution can test whether a covariate has an accelerating or decelerating effect on the timing of the event of interest.

Prior art structure analysis of patents: alters-to-alters citation network

The structure of a patent's prior art can be represented by an ego network of backward citations. The ego network of a patent consists of 1) primary citations to previous inventions (i.e., ego-to-alters citations) as well as 2) secondary citations among cited inventions (i.e., alters-to-alters citations). While ego-to-alters citations simply identify the list of previous patents considered by the inventor, alters-to-alters citations characterize the network functionality of the cited invention by considering not only alter properties but also their relational features. The present invention shows that alters-to-alters citations play an important role in characterizing the backward citation network structure.

Hereinafter, the network characteristics of backward citation will be described with reference to the drawings described later.

Figure 2 is a diagram illustrating three examples of an ego network according to an embodiment of the present invention.

Referring to Figure 2, although the same number (20) of previous patents are cited, alters in each network are connected in different ways. The alters in the leftmost network (A) are highly clustered, while the alters in the rightmost network (C) are completely isolated. To identify differences in ego network structures, we formally introduce three network indicators: constraint, cohesion, and efficiency. In Figure 2, all patents have the same number of backward citations: 20. Patent A can confirm that the citation network structure among the alters themselves is more cohesive.

Figure 3 is a diagram illustrating an ego network to explain the network characteristics of backward citation according to an embodiment of the present invention.

constraint

A constraint is a dyadic index defined for each alter connected to the ego. This indicator measures the extent to which the ego's efforts to maintain the relationship are focused on each alter. In the case of ego networks, network constraints are defined as the set of ego's binary constraints for each alter. Formally, the constraints of ego i (patent i) can be defined as follows.

(3)

Constraints consist of two components: a direct network and an indirect network. Direct constraints are expressed only as P _ij, which represents the relationship ratio between i and j. In an unweighted network, i is assumed to have the same effect on n alters, so P _ij becomes 1/n. The indirect constraint represents the sum of all indirect paths that can reach j through other alters. Accordingly, the constraint of ego i on j becomes greater when j is more connected to i's other alters, denoted by the subscript q in equation (3).

For example, Figure 3 shows an ego network with six alters. The constraints for each alter are part of the matrix obtained from the matrix calculation using all edges. In the network, ego i has a maximum constraint of 0.44 on alter E, but a minimum constraint of 0.09 on alter A. This is because the indirect connection between alters and ego is maximized for alter E, but minimized for alter A.

Constraint refers to how much the ego's learning opportunities are limited due to network closure between alters. In Figure 3, alter E is closely related to other alters, which places a burden on learning from alter E due to other alters communicating with alter E. The amount of these alters increases in proportion to the degree to which the alters are clustered. In a backward citation network, constraints reflect the degree of limitation imposed on prior knowledge at each citation as a recombination candidate. From a structural gap perspective, restricted networks are believed to have a negative impact on performance. In this regard, it can be argued that in a highly restricted ego network of backward citations, each knowledge source makes a similar or overlapping contribution to the invention, and thus there are fewer opportunities for inventors to generate new ideas.

Cohesion

Cohesion refers to the strength of relationships between network members. To measure cohesion, we consider the clustering coefficient, which is widely used in existing network analysis. The clustering coefficient is the ratio of the actual connections that exist between alters to the maximum number of possible ties. Clustering coefficients can be calculated at both network (global) and node (local) levels. The global clustering coefficient is the number of closed triples (complete triangles) compared to the total number of triples in the network. The local clustering coefficient is defined for each node and is given as the ratio of links between nodes divided by the maximum link that can exist within the neighboring nodes of the focal node (i.e. ego). Therefore, the local clustering coefficient of a node becomes equal to the density of the network without the focal node. In ego networks, the local clustering coefficient is more relevant because it measures the cohesion between alters from the ego's perspective. It is calculated as follows:

(4)

In equation (4), n is the number of alters and t is the number of citations among them. The denominator is the maximum number of connections possible between n alters. Unlike measures that aggregate measures at a two-dimensional level, the clustering coefficient is defined relative to the focal node (i.e. ego) and measures the overall cohesion of the ego network. In the example of Figure 3, the maximum possible connections between alters is 15 (= 6 * 5/2), and the actual number of connections between alters is 4. Therefore, the clustering coefficient is 0.26 (=4/15).

High cohesion in citation analysis means that there is pressure to form a closed network among cited inventions. From the perspective of a closed network, high cohesion improves homogeneity and uniformity among knowledge sources. This means that the knowledge belonging to a closed closed network is less unique, but is closely related or overlapping with each other, and greater dependence and homogeneity can be expected between them. Therefore, it can be argued that inventions that arise from highly cohesive networks of previous inventions are more likely to be incremental, but that inventions have lower technological risk and are more likely to collaborate with other technologies.

Efficiency

Efficiency refers to the rate of nonoverlapping knowledge flowing from alters to the ego. If an alter can be reached through many other existing alters, a connection to the alter may be considered redundant and therefore inefficient. Therefore, the actual amount of prior knowledge that the ego effectively acquires (i.e., the effective size of the network) must be reduced by the amount of redundancy. The effective size of node i's ego network is measured as follows.

(5)

In equation (5), m _qj represents the marginal strength of the relationship between node j and node q, which measures the amount of interaction between node q and node j divided by the strongest interaction of node j. For a binary network, such as a citation network, the strength of all links is 1, so the value of m is 0 or 1 depending on whether j is connected to q. Sum represents the degree of overlap in the relationship between node j and node i because it becomes larger as node i can reach node j through other nodes. If the sum is 0 for each alter j, then the effective size is 1.

In the example of Figure 3, the effective size of the tie with alternator A remains the same as the original size of 1, since A is not associated with any other alters. However, because there are three additional indirect connections between Alter E and Ego I, the effective size of the connection to Alter E is reduced by 3/6 (1-3/6). The sum of the effective size of each connection becomes the effective size of the network, which is the number of direct connections of alters depending on the degree of overlap. Efficiency is the ratio of the effective size of the ego network to its initial size.

Network efficiency normalizes the effective size of the ego network and tells how much unique knowledge the ego has gained from each alter. If there is no connection between alters, the knowledge received from alters does not completely overlap. In this case, the effective size of each alter is 1 and the effective size of the network is the number of alters. The associated efficiency of the network is equal to 1. Conversely, if all alters are fully connected to each other, the ego's connections are completely "redundant" and the ego gains no unique knowledge from the alters. In this case, the effective size of each alter is 1- (n-1)/n and the effective size of the network is 1. The associated efficiency of the network is equal to 1/n. The definition of efficiency represents the unique information present in the network of cited inventions, separating out 'redundant' knowledge in the network.

In Figure 3, Ego has a maximum constraint of 0.44, but since Alter E has the most indirect path from Ego to other alters, the minimum efficiency for Alter E is 3/6.

Other variables for prior art characteristics

To complement the network measurement, we add several variables that capture additional information from prior art, such as technology scope, temporal scope, and recency. By controlling these variables, we can capture the net effect of network structure on patent value. Technology coverage refers to the extent to which backward citations are spread across various technology sectors. This is measured by the number of unique technology classes to which each cited patent belongs. Temporal range measures the annual dispersion of backward citations and is calculated as the standard deviation of years of citation lag. Recency indicates how recent the patent's referenced knowledge is and is calculated as the average number of years of lag in backward citations. Variables controlling features unique to the patent itself are also considered. It includes the number of claims (# claims), number of backward citations (# backward citations), and number of technology classes to which the patent belongs (# classes). To control for unobserved fixed effects, year dummies and patent class dummies are also included in the regression model.

Applicant vs. examiner backward citations

Examiners tend to block or limit a claimed invention by citing prior art that the inventor missed reporting. In this case, backward citation counting may not effectively represent knowledge leakage from prior art to the inventor. Taking this issue into account, we use applicant backward citations to calculate network variables and other independent variables. By excluding examiner backward citations when measuring prior art structure, the proposed variable can represent the actual knowledge flow of prior art.

However, it is also important to note that applicant citations are far from a perfect tool for indicating the exact structure of the relevant prior invention. Since 41% of all citations to U.S. patents between 2001 and 2003 came from examiners rather than applicants, the examiner citation rate is quite high. In this case, the comprehensive structure of the previous invention can be better identified through backward citation aggregation. This is because the examiner, as an expert in the field, adds a significant portion of backward citations to a given invention that are considered relevant prior art. Backward citations added by the examiner limit the scope of the claimed invention and play an important role in determining the patentability of the invention. Since the main interest of the present invention is to analyze and provide whether the network structure of backward citations provides meaningful information about future patent value, total backward citations including examiner citations are used, and only the applicant's backward citations are used. Compare with the results used.

Figure 4 is a diagram showing the configuration of a patent analysis device according to an embodiment of the present invention.

Referring to FIG. 4, the patent analysis device 1 includes an input unit 10, a control unit 12, an output unit 14, and a database 16.

The patent analysis device 1 analyzes and provides mutual citation relationships of patents related to the target invention to be searched from the patent database 16, which has a field containing patent information of prior inventions cited in the patent application.

The input unit 10 can receive information on the target invention that is the subject of patent analysis. For example, at least one of the application number, publication number, and registration number of the target invention (hereinafter referred to as 'patent number') can be entered.

The control unit 12 analyzes backward cited patents and cited patents among prior inventions of the target invention for which one or more patent numbers have been entered. A backward-cited patent is a patent cited by the subject invention among prior inventions at the time of application. A cited patent is a patent that cites the subject invention.

The control unit 12 characterizes the network structure of backward citations and analyzes and provides the impact of the characterized network structure on technological impact or predictability of future value. Network characteristics of backward citation include constraints, cohesion, and efficiency.

The constraint can be calculated using a direct constraint, which represents the relationship ratio between target invention i and cited invention j, and an indirect constraint, which represents the sum of all indirect paths through which target invention i can reach cited invention j through other cited invention q. You can. Cohesion can be calculated using the ratio of the actual connections that exist between the cited inventions to the maximum number of possible ties (ties) between the cited inventions for the target invention. Efficiency is an indirect constraint representing the sum of all indirect paths through which target invention i can reach cited invention j through other cited invention q, and a binary value given depending on whether cited invention j is connected to cited invention q. It can be calculated using .

The control unit 12 can calculate the technical or economic value of the target invention using network variables and control variables generated by backward citation. At this time, the network variable represents an nth-order network characteristic including at least one of constraints, cohesion, and efficiency, and the control variable is the number of backward citations, the number of Cooperative Patent Classification (CPC) codes of backward citations, It may include at least one of the recency and time spread of backward citations and the number of claims.

The control unit 12 can analyze and provide impact by separating applicant citations and examiner citations.

The output unit 14 provides analysis results through the control unit 12.

As described above with reference to Figures 1 to 4, modern innovation is a process of 'searching and recombining' prior inventions, and the value of an invention is related not only to the number of previous inventions cited but also to the way they are connected to each other. There is. We use social network analysis, widely used in social capital theory, to analyze and provide whether the hidden network structure of prior art has a significant impact on future patent value. Using both primary (ego-to-alter) and secondary alter-to-alter networks, we consider network variables such as constraints, cohesion, and efficiency, and determine how these network characteristics It uses the fact that there is a statistically significant correlation with the future value of the invention. The citation network reveals key properties regarding the knowledge structure of the prior art. In the patent citation network, ‘constraint’ reflects the degree of limitation imposed on prior knowledge as a recombination candidate. 'Cohesion' measures the degree of relationship between network members and implies that there is pressure to create greater homogeneity among knowledge sources. 'Efficiency' measures the rate of non-overlapping knowledge flow from alter to ego and determines the amount of unique knowledge that the invention obtains from each unit of prior art. We assume that network variables measured by applicant backward citations characterize the knowledge flow of inventions. Recognizing the inventor's limited ability to list all relevant prior art and the examiner's complementary role in adding missed prior art, we use a variable measured by backward citations. We expect the variable to identify the strategic location of inventions through more comprehensive information about the technological trajectory of prior art.

Using U.S. pharmaceutical patent data, we show that the network structure of backward citations has a significant impact on both the posthumous value and lifespan of an invention, as well as its private value and social impact. Patents with more limited, more cohesive, and less efficient backward citation networks may have lower patent value and lower technological impact. If an invention is successful enough to be patented, these results support the hypothesis of a ‘structural gap’ among two contrasting perspectives on social capital: a closed network versus a structural gap perspective. Under the structural gap hypothesis, highly restricted, cohesive, and inefficient citations may result in similar or overlapping contributions to the invention, with less opportunity to generate new ideas from them.

The present invention can expand the dimension of analysis for patent evaluation by utilizing a second-order (or expandable to n-th order) backward citation network. Additionally, it can provide the first empirical evidence on the relationship between various network measures of backward citations and the future value of patents. Additionally, our model is valuable for both academic and practical purposes. Companies and policy makers can apply our model to identify target technologies or predict promising technology areas. This model allows companies to select target patents to purchase or select R&D projects to develop on competitors' patents. Network analysis can be used to determine strategic positioning for a company's R&D investments.

So far, the present invention has been examined focusing on its embodiments. A person skilled in the art to which the present invention pertains will understand that the present invention can be implemented in a modified form without departing from the essential characteristics of the present invention. Therefore, the disclosed embodiments should be considered from an illustrative rather than a restrictive perspective. The scope of the present invention is indicated in the claims rather than the foregoing description, and all differences within the equivalent scope should be construed as being included in the present invention.

Claims (10)

In the patent analysis method using a patent analysis device,
Characterizing the network structure of backward citations cited by the target invention; and
Analyzing and providing the impact of the characterized network structure on the technical impact or predictability of future value; Includes,
The steps for characterizing
As network features of backward citations, constraints, cohesion and efficiency are calculated;
The steps to analyze and provide are
Analyze the private value of the patent applicant and the social value of subsequent inventions. When analyzing the private value of the patent applicant, separate the applicant citations and examiner citations and use examiner citations, and do not separate the applicant citations and examiner citations when analyzing social value. A patented analysis method characterized by being used without. delete The method of claim 1, wherein the step of characterizing
Calculate constraint variables using a direct constraint representing the relationship ratio between target invention i and cited invention j, and an indirect constraint representing the sum of all indirect paths through which target invention i can reach cited invention j through other cited invention q. A patent analysis method characterized by: The method of claim 1, wherein the step of characterizing
A patent analysis method characterized by calculating the cohesion variable using the ratio of actual connections that exist between cited inventions to the maximum possible number of ties (ties) between cited inventions for the target invention. The method of claim 1, wherein the step of characterizing
Using an indirect constraint representing the sum of all indirect paths through which target invention i can reach cited invention j through other cited invention q, and a binary value given depending on whether cited invention j is connected to cited invention q, Patent analysis method characterized by calculating efficiency. The method of claim 1, wherein the analyzing and providing steps include
Calculate the technical or economic value of the target invention using network variables and control variables generated by backward citation,
The network variable represents an nth order (n is a positive integer) network characteristic including at least one of constraints, cohesion, and efficiency,
Control variables include at least one of the number of backward citations, the number of Cooperative Patent Classification (CPC) codes of backward citations, the recency and time spread of backward citations, and the number of claims. A patent analysis method characterized by: delete An input unit that receives information on the target invention subject to patent analysis;
A control unit that characterizes the network structure of backward citations cited by the target invention and analyzes the impact of the characterized network structure on technological impact or predictability of future value; and
An output unit that provides analysis results; Includes,
The control part
As network features of backward citations, constraints, cohesion and efficiency are calculated;
Analyze the private value of the patent applicant and the social value of subsequent inventions. When analyzing the private value of a patent applicant, separate applicant citations and examiner citations and use examiner citations, and do not separate applicant citations and examiner citations when analyzing social value. A patented analysis device characterized by being used without using it. delete The method of claim 8, wherein the control unit
Calculate the technical or economic value of the target invention using network variables and control variables generated by backward citation,
The network variables represent nth-order network characteristics including at least one of constraints, cohesion, and efficiency,
Control variables include at least one of the number of backward citations, the number of Cooperative Patent Classification (CPC) codes of backward citations, the recency and time spread of backward citations, and the number of claims. A patented analysis device characterized by: