Patent technology field layout organization management method and system

Abstract

The invention relates to a layout organization management method and system in the technical field of patents, and the method comprises the steps: obtaining patent data related to the power industry, carrying out the preprocessing, obtaining a reference relation between patents based on the preprocessed patent data, and constructing a reference network based on the reference relation; calculating a structural hole index based on the reference network, optimizing an LDA model based on the structural hole index, generating a technical field classification result by using the optimized LAD model, and obtaining the number of patents in each type of technical field; acquiring a quality coefficient of each patent in each technical field by utilizing a text similarity model based on the preprocessed patent data, and acquiring a layout index in the corresponding technical field in combination with the quantity data; and generating a patent technology field layout organization management strategy based on the layout indexes. Compared with the prior art, the invention provides a method capable of arranging patents in innovative fields.

Description

Technical Field

[0001] The present invention relates to intellectual property technology, and in particular to a patent technology field layout organization management method and system. Background Art

[0002] As an important form of intellectual property protection, patents play a vital role in promoting technological innovation. Patents can effectively promote the development of scientific and technological innovation. By protecting the achievements of innovators through patents, innovators can not only obtain economic benefits, but also gain reputation and recognition, thereby further motivating innovators to continuously invest more energy and resources in innovation. At the same time, patents can promote technology transfer and diffusion, thereby promoting technological progress in the entire industry. The patent authorization rate can intuitively reflect the innovation ability of the corresponding technical field, that is, for technical fields with strong innovation capabilities, more patents can be deployed, and vice versa, the number of deployed patents can be reduced. The Chinese patent application "CN119941457A" discloses a multi-objective data analysis and strategy optimization method for patent layout. By introducing a multi-objective data analysis and strategy optimization system into patent layout, a multi-objective optimization model for patent layout is established. Although it solves the problem of insufficient quantification of patent layout, it has the following shortcomings:

[0003] First, by identifying gaps in technological fields through the distribution of patent numbers, the actual relationships between pieces of technology are ignored, resulting in a layout strategy focused on technological fields with a small number of patents but isolated technologies and weak innovation capabilities, rather than cross-disciplinary fields with strong innovation capabilities.

[0004] Second, modeling directly based on the statistics of the three-level segmented technologies provided by experts may ignore the technical fields with implicit innovation and affect the probability of successful patent layout.

[0005] Therefore, providing a method that can be deployed in innovative fields is a technical problem that needs to be solved. Summary of the Invention

[0006] The purpose of the present invention is to overcome the defects of the above-mentioned existing technologies and provide a patent technology field layout organization management method and system, which classifies all patents in the power industry into technical fields using the LDA model optimized with structural hole indicators, and calculates the layout indicators based on the classification results and the calculated dynamic weights.

[0007] The purpose of the present invention can be achieved by the following technical solutions:

[0008] According to a first aspect of the present invention, a method for managing layout organization in the field of patent technology is provided, the method comprising:

[0009] Acquire and preprocess patent data related to the power industry, obtain citation relationships between patents based on the preprocessed patent data, and construct a citation network based on the citation relationships;

[0010] Calculate the structural hole index based on the citation network, optimize the LDA model based on the structural hole index, use the optimized LDA model to generate technology field classification results, and obtain the number of patents in each technology field;

[0011] Based on the pre-processed patent data, a text similarity model is used to obtain the quality coefficient of each patent in each technical field, and the layout index in the corresponding technical field is obtained by combining the quantitative data.

[0012] Generate a layout organization management strategy for the patent technology field based on the layout indicators.

[0013] As a preferred technical solution, the structural hole indicators include: efficiency, constraint and betweenness centrality.

[0014] Among them, the measurement value of utilizing non-redundant connections between patents is efficiency, which is expressed as:

[0015]

[0016] p ik Indicates patent v i and neighbors v k The proportion of connection strength; m jk Indicates neighbor v j and neighbors v k If there is a connection, the value is 1, otherwise it is 0; Indicates patent v i The total number of connections; ik Indicates neighbor v k Compared with patent v i The time decay coefficient of jk Indicates neighbor v k Compared to the neighbor v j Time decay coefficient;

[0017] The dependency value of a patent on its neighboring patents is the constraint, and its expression is:

[0018]

[0019] p ij Indicates patent v i and neighbors v j The proportion of connection strength; p ik Indicates patent v i and neighbors v k The proportion of connection strength; p jkIndicates patent v j and neighbors v k The proportion of connection strength; p ik p jl Indicates patent v k Patent v i and v j The control value of the relationship; ij Indicates patent v j Compared to the neighbor v i The time decay coefficient of

[0020] The strength of a patent as an intermediate point between connected patents is called betweenness centrality, which is expressed as:

[0021]

[0022] σ st Indicates patent v s To Patent v t The total number of shortest paths; σ st ((v i ) indicates patent v s To Patent v t The shortest path through patent v i The number of paths; st Indicates patent v t Compared with patent v s The time decay coefficient of

[0023] The structural hole index is a weighted sum of the efficiency and betweenness centrality, and the inverse value of the constraint.

[0024] As a preferred technical solution, the method for optimizing the LDA model is:

[0025] In the LDA model, the structural hole weight is introduced in the process of Gibbs sampling to update the topic assignment, and its expression is:

[0026]

[0027] Among them, z d,n represents the technical field assignment result of the nth segmentation word in patent d; k represents the technical field index; z -d,n represents the technical field allocation results of other segmentations in the patent except the nth segmentation in patent d; w d,n Indicates the technical field assignment result of the current word; represents the number of patents in patent d that are assigned to technical field k except the current word; α and β are both distribution parameters; represents the number of times word type t appears in technical field k, except for the word type t corresponding to the current n; λ represents the hyperparameter; SHd represents the structural hole index of patent d; z d Indicates the technical field to which patent d belongs; w k represents the structural hole weight.

[0028] As a preferred technical solution, the method for obtaining the classification results of the patent technology field includes:

[0029] A1. Acquire text data based on the preprocessed patent data, construct a text-word matrix based on the text data, initialize the LDA model, and randomly assign a technical field to each word in each text data;

[0030] A2. Based on the allocation of technical fields, calculate the structural hole weight value of each technical field based on the structural hole index;

[0031] A3. performing the Gibbs sampling update topic allocation based on the structural hole weight value;

[0032] A4. Calculate the perplexity after the Gibbs sampling update topic assignment. If the perplexity meets the iteration termination condition, execute A5. Otherwise, execute A2 to A3.

[0033] A5. Output the probability distribution matrix of each patent's technical field and its corresponding classification label.

[0034] As a preferred technical solution, the method for calculating the structural hole weight value is:

[0035]

[0036] Among them, N k represents the number of patents in the technical field k; SH d represents the structural hole index of patent d; z d Indicates the technical field to which patent d belongs.

[0037] As a preferred technical solution, the method for obtaining the layout index includes:

[0038] For each patent in the same technical field, text features are extracted based on the preprocessed patent data, and the text features are used as input to the text similarity model to obtain the similarity of the patents in the corresponding technical field;

[0039] Calculate the quality coefficient of each patent in the corresponding technical field based on the similarity, and calculate the average quality coefficient of the corresponding technical field;

[0040] Calculate the dynamic weight based on the quantitative data; the quantitative data includes the total number of patents, the number of granted patents, and the number of rejected patents in each technical field;

[0041] After normalizing the total number, the layout index of the corresponding technical field is calculated by combining the quality coefficient mean and dynamic weight.

[0042] As a preferred technical solution, the method for calculating the dynamic weight is:

[0043] Calculate the authorization rate, the expression is: Calculate the rejection rate, the expression is: Among them, A k represents the number of authorizations in the technical field k; R k N represents the number of rejections in technical field k; k represents the total number of patents in the technical field k; τ represents the mean difference between the current time and the patent application date;

[0044] The difference between the authorization rate and the rejection rate is calculated, and the difference is the dynamic weight value.

[0045] As a preferred technical solution, the calculation expression of the layout index is:

[0046]

[0047] Among them, γ represents the dynamic weight; It represents the mean value of the quality coefficient of the technical field k; represents the total number of technical fields k after normalization.

[0048] As a preferred technical solution, the loss function of the optimized LDA model is:

[0049]

[0050] Where M represents the number of patents; d represents patent d; w d represents the word in patent d; α′ and β′ both represent prior parameters.

[0051] According to a second aspect of the present invention, a patent technology field layout organization management system is provided for implementing the above method.

[0052] Compared with the prior art, the present invention has the following beneficial effects:

[0053] 1) This invention calculates the structural hole index based on the patent's citation network, and uses the structural hole index to optimize the LDA model to classify technical fields. It reconstructs the technical field value of patents from the dimension of technical association, avoiding the limitations caused by using the number of patents to fill the technical field. By calculating the efficiency, constraints and betweenness centrality between patents, it captures the technical dependence and innovation transmission path of patent keys, avoiding the misjudgment of "technological islands" as blank areas in high-value technical fields, and can achieve the priority identification of technical fields with related innovations.

[0054] 2) The present invention calculates dynamic weights based on the number of authorizations and rejections in each technical field, and introduces dynamic weights to calculate patent layout indicators, avoiding the difficulty in discovering cross-domain emerging technical fields caused by the traditional three-level subdivision technology classification. It also balances the number and quality of patents through dynamic weights, thereby achieving a more scientific patent layout. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 Flow chart of the method of the present invention. DETAILED DESCRIPTION

[0056] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0057] Unless otherwise defined, the technical or scientific terms used in this application should have the ordinary meaning understood by a person of ordinary skill in the technical field to which this application belongs. The words "one", "a", "the" and the like used in this application do not indicate a limit on quantity and may indicate the singular or plural. The terms "include", "comprise", "have" and any variations thereof used in this application are intended to cover non-exclusive inclusions; for example, a process, method, system, product or device that includes a series of steps or modules (units) is not limited to the listed steps or units, but may also include steps or units that are not listed, or may also include other steps or units that are inherent to these processes, methods, products or devices. The words "connect", "connected", "coupled" and the like used in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The word "multiple" used in this application refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships can exist. For example, "A and / or B" can mean: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the objects before and after are in an "or" relationship. The terms "first", "second", "third", etc. involved in this application are only used to distinguish similar objects and do not represent a specific order for the objects.

[0058] Example 1

[0059] In order to solve the technical problems existing in the prior art, the present invention provides a layout organization management method in the field of patent technology, the process of which is as follows: Figure 1 Shown, including:

[0060] S1. Obtain and preprocess patent data related to the power industry, obtain citation relationships between patents based on the preprocessed patent data, and construct a citation network based on the citation relationships.

[0061] S11. Obtain patent data from the patent database of the power industry, including: title, abstract, claims, IPC classification number, cited / citing patent ID, application date and legal status.

[0062] S12. Preprocess patent data, including word segmentation, stop word removal, and stemming, and construct a text-word matrix.

[0063] S13. Obtain citation relationships between patents based on patent data and construct a citation network, in which nodes are patents and edges are citation relationships between patents.

[0064] S2. Calculate the structural hole index based on the citation network, optimize the LDA model based on the structural hole index, use the optimized LAD model to generate technology field classification results, and obtain the number of patents in each technology field.

[0065] S21. Calculation of structural hole index:

[0066] Structural holes are a core concept in social network analysis. They refer to the "gap" between two nodes in a network that lacks a direct connection and must be indirectly connected through a third node. Nodes occupying structural holes (called "bridge nodes") can obtain more non-redundant information and have higher innovation potential. Specifically, in the field of patent technology, structural holes can be used to discover innovative knowledge hidden in patent networks. If a certain technical field is connected to multiple other technical fields, and the corresponding technical field occupies a structural hole, it indicates that it is likely to contain cross-domain innovation points. Therefore, in this invention, the structural hole indicator is introduced to enhance the mining of cross-domain innovative technical fields.

[0067] S211. In detail, the structural hole indicators include efficiency, constraint, and betweenness centrality. Among them, the measurement value of the use of non-redundant connections between patents is efficiency, which is expressed as:

[0068]

[0069] p ik Indicates patent v i and neighbors v k The proportion of connection strength; m jk Indicates neighbor v j and neighbors v k If there is a connection, the value is 1, otherwise it is 0; Indicates patent v i The total number of connections; ik Indicates neighbor v k Compared with patent v i The time decay coefficient of jk Indicates neighbor v k Compared to the neighbor v j Time decay coefficient; ∑ j ((1-ζ ik ζ jk ∑ k p ik m jk ) represents the effective size of the patent, which is used to measure the ability of the patent to influence the network through non-redundant connections. The higher the value, the more significant the structural hole occupied by the node. This value is used to measure the ability of the node to influence the network through non-redundant connections. The closer the value is to 1, the less redundancy there is.

[0070] The dependency value of a patent on its neighboring patents is the constraint, and its expression is:

[0071] C vi =∑ j ((ζ ij p ij +ζ ik ζ jk ∑ k p ik p jk ),

[0072] p ij Indicates patent v i and neighbors v j The proportion of connection strength; p ik Indicates patent v i and neighbors v k The proportion of connection strength; p jk Indicates patent v j and neighbors v k The proportion of connection strength; p ik p jk Indicates patent v k Patent v i and v j The control value of the relationship; ij Indicates patent v j Compared to the neighbor v i The time decay coefficient of the paper; the higher the value, the stronger the paper's dependence on its few neighboring patents and the fewer structural holes.

[0073] The strength of a patent as an intermediate point between connected patents is called betweenness centrality, which is expressed as:

[0074]

[0075] σ st Indicates patent v s To Patent v t The total number of shortest paths; σ st ((v i ) indicates patent v s To Patent v t The shortest path through patent v i The number of paths; st Indicates patent v t Compared with patent v s The time decay coefficient of the patent; the higher the value, the stronger the role of the corresponding patent as a "bridge" in the network, which is related to the location of the structural hole.

[0076] S212. The constraint calculated above is negated, and the weighted sum of the negated value, efficiency, and betweenness centrality is calculated. This sum is the structural hole index, and its calculation expression is:

[0077]

[0078] Wherein, w1, w2 and w3 all represent weights, and w1+w2+w3=1.

[0079] S22. Generate classification results in the technical field:

[0080] A1. Obtain the text-word matrix and initialize the LDA model, randomly assigning a technology field to each word in each text data.

[0081] In detail, in the LAD model, the word distribution of each technical field k follows a Dirichlet distribution, which is expressed as: k ~Dirichlet(β), k=1,2,3,..., generates a technology field distribution for each patent, and the technology field distribution follows the Dirichlet distribution θ d ~Dirichlet(α),d=1,2,3,..., sampling a technology field z from the distribution of patent technology fields d,n , and z d,n ~Multinomial(θ d ), sample a word w from the word distribution corresponding to the domain d,n ,and

[0082] When initializing the model, the structural hole index of each patent is set to 0 and updated in subsequent iterations.

[0083] A2. Based on the allocation of technical fields, calculate the structural hole weight value of each technical field based on the structural hole index.

[0084] Specifically, the method for calculating the weight value of the structural hole is:

[0085]

[0086] Among them, N k represents the number of patents in the technical field k; SH d represents the structural hole index of patent d; z d Indicates the technical field to which patent d belongs.

[0087] A3. Update topic allocation by Gibbs sampling based on the structural hole weight value.

[0088] The probability of the traditional Gibbs sampling updating topic assignment topic is: Relying solely on the text features of patents to identify technical fields ignores the association and citation association between patents, which will lead to ignoring the actual association between technical fields when clustering technical fields. To address this shortcoming, the present invention incorporates structural hole indicators into the LDA model when allocating technical fields. Specifically, the structural hole weight is introduced, and its expression is:

[0089]

[0090] Among them, z d,n represents the technical field assignment result of the nth segmentation word in patent d; k represents the technical field index; z -d,n represents the technical field allocation results of other segmentations in the patent except the nth segmentation in patent d; w d,n Indicates the technical field assignment result of the current word; represents the number of patents in patent d that are assigned to technical field k except the current word; α and β are both distribution parameters; represents the number of times word type t appears in technical field k, except for the word type t corresponding to the current n; λ represents the hyperparameter; SH d represents the structural hole index of patent d; z d Indicates the technical field to which patent d belongs; w k represents the structural hole weight.

[0091] A4. Calculate the perplexity after Gibbs sampling updates the topic assignments. If the perplexity meets the iteration termination criteria, execute A5. Otherwise, execute A2-A3.

[0092] In detail, the calculation formula of perplexity is: Where M is the number of patents, w d,n is the word in patent d, N d is the number of words in patent d, p(w d ) is the likelihood probability of patent d.

[0093] The iteration is terminated when the perplexity calculated during multiple consecutive iterations (10 consecutive iterations in this embodiment) decreases by less than a threshold.

[0094] In addition, when the number of iterations reaches the maximum value, the iteration ends.

[0095] A5. Output the probability distribution matrix of each patent's technical field and its corresponding classification label.

[0096] In this step, the structural hole weights evolve dynamically during the iteration process, thereby guiding the distribution of technology fields to converge towards areas with high innovation potential, forming a coordinated optimization of text semantics and technology network structure. Through this iterative mechanism, the structural hole constraints are no longer static priors, but dynamic guiding signals that are continuously adjusted with the optimization of topic distribution, ultimately making the model output a technology field classification result that satisfies both text consistency and technological innovation.

[0097] S3. Based on the pre-processed patent data, the text similarity model is used to obtain the quality coefficient of each patent in each technical field, and the layout index in the corresponding technical field is obtained by combining the quantitative data.

[0098] S31. For each patent in the same technical field, text features are extracted based on the preprocessed patent data, and the text features are used as input to the text similarity model to obtain the similarity of the patents in the corresponding technical field.

[0099] In the present invention, the n-gram model is used as the text similarity model. In addition, a bag-of-words model, machine learning or deep learning can also be used.

[0100] S32. Calculate the quality coefficient of each patent in the corresponding technical field based on the similarity, and calculate the average quality coefficient of the corresponding technical field.

[0101] Specifically, if the similarity of a patent is negatively correlated with the quality of the patent, that is, the higher the similarity, the lower the quality of the patent. Therefore, the calculation expression of its quality coefficient is: q k,d =1-s k,d , where s k,d represents the similarity of patent d in technical field k; q k,d represents the quality coefficient of patent d in technical field k.

[0102] S33. Calculate dynamic weights based on quantitative data, where the quantitative data includes the total number of patents, the number of authorizations, and the number of rejections in each technical field.

[0103] S331. Calculate the authorization rate, which is expressed as: Calculate the rejection rate, the expression is: Among them, A k represents the number of authorizations in the technical field k; R k N represents the number of rejections in technical field k; k represents the total number of patents in the technical field k; τ represents the mean difference between the current time and the patent application date;

[0104] S332. Calculate the difference between the authorization rate and the rejection rate. This difference is the dynamic weight value.

[0105] S34. After normalizing the total number, calculate the layout index of the corresponding technical field by combining the mean value of the quality coefficient and the dynamic weight.

[0106] When conducting patent layout, if there are fewer patents in a certain technical field and the legal status of most patents is authorized, then more patents can be laid out in this technical field. The corresponding layout index value of this technical field is larger, that is, the layout index is positively correlated with patent quality and negatively correlated with the number of patents. Its calculation expression can be:

[0107]

[0108] Among them, γ represents the dynamic weight; It represents the mean value of the quality coefficient of the technical field k; represents the total number of technical fields k after normalization.

[0109] S4. Generate the layout organization and management strategy for the patent technology field based on the layout indicators.

[0110] Example 2

[0111] This embodiment provides a training method for an optimized LDA model, including:

[0112] B1. Construct a model training dataset and preprocess the patent data in the training dataset.

[0113] B2. Construct a patent citation network based on the preprocessed training dataset and calculate the structural hole index.

[0114] B3. Divide the preprocessed training dataset and the corresponding structural hole indicators into a training set and a test set in a ratio of 8:2.

[0115] B4. Use the training set data to train the optimized LDA model.

[0116] B41. Model initialization: Set the number of potential technical fields to K′, set the initial values ​​of the model hyperparameters and the structural hole constraint coefficients, and divide the technical fields for each word in each patent in each training set according to the topic distribution rules.

[0117] Iterate B42 to B45 until the loss function value stabilizes:

[0118] B42. Remove the topic count of the current word. The detailed removal rules are as follows: in, represents the number of patents assigned to technical field z d,n The number of words; Indicates the technical field z d,n Chinese word wd,n Number of occurrences.

[0119] B43. Calculate the structural hole indexes of all patents in each technical field after processing in B42, and calculate the structural hole weights based on the structural hole indexes of all patents.

[0120] B44. Introduce the structural hole weight into the calculation of the probability of updating the topic assignment using Gibbs sampling to obtain the joint probability. Based on this joint probability, sample new technology areas and update the topic count, which is:

[0121] B45. Calculate the loss function, which is expressed as:

[0122]

[0123] Where M represents the number of patents; d represents patent d; w d represents the word in patent d; α′ and β′ both represent prior parameters.

[0124] B5. After training, verify the performance of the model in the test set. If the performance is excellent, the current model will be used as the final model. Otherwise, re-execute B4.

[0125] Example 3

[0126] The present embodiment provides a layout organization management system in the field of patent technology, which is used to implement the above method, including a central processing unit (CPU), which can perform various appropriate actions and processes according to computer program instructions stored in a read-only memory (ROM) or computer program instructions loaded from a storage unit into a random access memory (RAM). In RAM, various programs and data required for device operation can also be stored. The CPU, ROM, and RAM are connected to each other via a bus. The input / output (I / O) interface is also connected to the bus.

[0127] Many components in a device are connected to the I / O interface, including: input units, such as a keyboard and mouse; output units, such as various types of displays and speakers; storage units, such as magnetic disks and optical disks; and communication units, such as network cards, modems, and wireless communication transceivers. The communication unit allows the device to exchange information / data with other devices via computer networks such as the Internet and / or various telecommunication networks.

[0128] The processing unit performs the various methods and processes described above, such as methods S1 to S3, A1 to A5, and B1 to B5. For example, in some embodiments, methods S1 to S3, A1 to A5, and B1 to B5 may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit. In some embodiments, part or all of the computer program may be loaded and / or installed onto the device via a ROM and / or a communication unit. When the computer program is loaded into the RAM and executed by the CPU, one or more steps of methods S1 to S3, A1 to A5, and B1 to B5 described above may be performed. Alternatively, in other embodiments, the CPU may be configured to execute methods S1 to S3, A1 to A5, and B1 to B5 by any other appropriate means (e.g., by means of firmware).

[0129] The functions described above herein may be performed, at least in part, by one or more hardware logic components. For example, and without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chip (SOCs), complex programmable logic devices (CPLDs), and the like.

[0130] The program code for implementing the method of the present invention can be written in any combination of one or more programming languages. Such program code can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device so that when the program code is executed by the processor or controller, the functions / operations specified in the flow chart and / or block diagram are implemented. The program code can be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0131] In the context of the present invention, machine-readable medium can be a tangible medium that can contain or store a program for use with an instruction execution system, device or equipment or used in combination with an instruction execution system, device or equipment. Machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. More specific examples of machine-readable storage media can include electrical connections based on one or more lines, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0132] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and such modifications or substitutions are intended to be within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be subject to the scope of protection of the claims.

Claims

1. A patent technology field layout organization management method, characterized by: The method comprises: Acquire and preprocess patent data related to the power industry, obtain citation relationships between patents based on the preprocessed patent data, and construct a citation network based on the citation relationships; Calculate the structural hole index based on the citation network, optimize the LDA model based on the structural hole index, use the optimized LDA model to generate technology field classification results, and obtain the number of patents in each technology field; Based on the pre-processed patent data, a text similarity model is used to obtain the quality coefficient of each patent in each technical field, and the layout index in the corresponding technical field is obtained by combining the quantitative data. Generate a layout organization management strategy for the patent technology field based on the layout indicators.

2. A patented technology field layout organization management method according to claim 1, characterized in that: The structural hole indicators include: efficiency, constraint and betweenness centrality, Among them, the measurement value of utilizing non-redundant connections between patents is efficiency, which is expressed as: p ik Indicates patent v i and neighbors v k The proportion of connection strength; m jk Indicates neighbor v j and neighbors v k If there is a connection, the value is 1, otherwise it is 0; Indicates patent v i The total number of connections; ik Indicates neighbor v k Compared with patent v i The time decay coefficient of jk Indicates neighbor v k Compared to the neighbor v j Time decay coefficient; The dependency value of a patent on its neighboring patents is the constraint, and its expression is: p ij Indicates patent v i and neighbors v j The proportion of connection strength; p ik Indicates patent v i and neighbors v k The proportion of connection strength; p jk Indicates patent v j and neighbors v k The proportion of connection strength; p ik p jk Indicates patent v k Patent v i and v j The control value of the relationship; ij Indicates patent v j Compared to the neighbor v i The time decay coefficient of The strength of a patent as an intermediate point between connected patents is called betweenness centrality, which is expressed as: σ st Indicates patent v s To Patent v t The total number of shortest paths; σ st (v i ) indicates patent v s To Patent v t The shortest path through patent v i The number of paths; st Indicates patent v t Compared with patent v s The time decay coefficient of The structural hole index is a weighted sum of the efficiency and betweenness centrality, and the inverse value of the constraint.

3. A patented technology field layout organization management method according to claim 1, characterized in that: The method to optimize the LDA model is: In the LDA model, the structural hole weight is introduced in the process of Gibbs sampling to update the topic assignment, and its expression is: Among them, z d,n represents the technical field assignment result of the nth segmentation word in patent d; k represents the technical field index; z -d,n represents the technical field allocation results of other segmentations in the patent except the nth segmentation in patent d; w d,n Indicates the technical field assignment result of the current word; represents the number of patents in patent d that are assigned to technical field k except the current word; α and β are both distribution parameters; represents the number of times word type t appears in technical field k, except for the word type t corresponding to the current n; λ represents the hyperparameter; SH d represents the structural hole index of patent d; z d Indicates the technical field to which patent d belongs; w k represents the structural hole weight.

4. A patented technology field layout organization management method according to claim 3, characterized in that: The method for obtaining the classification results of the patent technology field includes: A1. Acquire text data based on the preprocessed patent data, construct a text-word matrix based on the text data, initialize the LDA model, and randomly assign a technical field to each word in each text data; A2. Based on the allocation of technical fields, calculate the structural hole weight value of each technical field based on the structural hole index; A3. performing the Gibbs sampling update topic allocation based on the structural hole weight value; A4. Calculate the perplexity after the Gibbs sampling update topic assignment. If the perplexity meets the iteration termination condition, execute A5. Otherwise, execute A2 to A3. A5. Output the probability distribution matrix of each patent's technical field and its corresponding classification label.

5. A patented technology field layout organization management method according to claim 4, characterized in that: The method for calculating the structural hole weight value is: Among them, N k represents the number of patents in the technical field k; SH d represents the structural hole index of patent d; z d Indicates the technical field to which patent d belongs.

6. A patented technology field layout organization management method according to claim 1, characterized in that: The method for obtaining the layout index includes: For each patent in the same technical field, text features are extracted based on the preprocessed patent data, and the text features are used as input to the text similarity model to obtain the similarity of the patents in the corresponding technical field; Calculate the quality coefficient of each patent in the corresponding technical field based on the similarity, and calculate the average quality coefficient of the corresponding technical field; Calculate the dynamic weight based on the quantitative data; the quantitative data includes the total number of patents, the number of granted patents, and the number of rejected patents in each technical field; After normalizing the total number, the layout index of the corresponding technical field is calculated in combination with the quality coefficient mean and dynamic weight.

7. A patented technology field layout organization management method according to claim 6, characterized in that: The method for calculating the dynamic weight is: Calculate the authorization rate, the expression is: Calculate the rejection rate, the expression is: Among them, A k represents the number of authorizations in the technical field k; R k N represents the number of rejections in technical field k; k represents the total number of patents in the technical field k; τ represents the mean difference between the current time and the patent application date; The difference between the authorization rate and the rejection rate is calculated, and the difference is the dynamic weight value.

8. A patented technology field layout organization management method according to claim 6, characterized in that: The calculation expression of the layout index is: Among them, γ represents the dynamic weight; It represents the mean value of the quality coefficient of the technical field k; represents the total number of technical fields k after normalization.

9. A patented technology field layout organization management method according to claim 1, characterized in that: The loss function of the optimized LDA model is: Where M represents the number of patents; d represents patent d; w d represents the word in patent d; α′ and β′ both represent prior parameters.

10. A patented technology field layout organization management system, characterized by: The system is used to implement the method according to any one of claims 1 to 9.

Images