Air traffic control field patent clustering analysis and evaluation recommendation method

Abstract

The invention discloses an air traffic control field patent clustering analysis and evaluation recommendation method. The method comprises the following steps: firstly, establishing an air traffic control field patent evaluation index system; then determining a measurement index weight based on an entropy weight method; performing patent clustering by using Euclidean distance, clustering and combining the patent sets into patent combinations with similar parameter characteristics, performing patent evaluation from legal dimension, technical dimension and economic dimension, and measuring and calculating patent scores; and finally, generating three-dimensional recommendation schemes, and taking the first ten percent of the three patents with high scores as key recommendation schemes. The method provides support for systematic development of patent evaluation and recommendation in the air traffic control field.

Description

Technical Field

[0001] The present invention belongs to the technical field of patent cluster analysis, and in particular to a patent cluster analysis and evaluation recommendation method in the field of air traffic control. Background Art

[0002] As an important part of the national air defense system and air transportation system, the air traffic management system is the command center for ensuring flight safety, implementing air traffic management and services, and realizing efficient and orderly operation of air transportation. In recent years, air flights have become increasingly complex, and problems such as flight conflicts and large-scale airspace congestion have occurred frequently. Foreign systems are not adapted to the local environment, and the technology is uncontrollable, making it difficult to adapt to the application needs of the rapid development of domestic air traffic control. Therefore, there is an urgent need for an air traffic management system with core key technologies with independent intellectual property rights and independent research and production capabilities.

[0003] The global research on air traffic control technology started in 1920 and entered the technology accumulation period in 1995. Since 2015, the number of patent applications in this field has increased significantly, and has made great progress and breakthroughs in the fields of flight delays and traffic management. The accumulation of domestic air traffic control patents began in 2010. Since 2015, the number of patent applications in this field has increased rapidly, and it is expected to maintain an upward trend in the next few years.

[0004] The accumulation of intellectual property rights has given rise to the need for accurate patent recommendations. From the perspective of enterprises, recommending patents for their own patent reserves can help to activate existing stocks, make good use of incremental stocks, and enhance the core competitiveness of enterprises; by evaluating competitors' patent collections, it is possible to analyze the competitive situation of competitors in related technical fields and provide guidance for subsequent technology research and development and market competition.

[0005] At present, there are no reports on the research results of building patent recommendation models in the field of air traffic control. The relevant research results that can be referenced are as follows:

[0006] "A patent recommendation algorithm based on topic classification and semantic similarity" (Wireless Internet Technology, 2021, (21), 105-107) proposed a patent recommendation algorithm based on topic classification and semantic similarity. By introducing the neural network Bert, the patent title and abstract are subjected to keyword extraction and word vector conversion to make patent recommendations. The Chinese patent application number 202210777356.0 discloses a method for recommending related patents based on a technical feature knowledge graph. The implicit technical association improves the coverage of patent recommendations, and the explicit technical association improves the accuracy of patent recommendations. The Chinese patent application number 201910414744.0 discloses a patent recommendation method and system based on patent big data. By obtaining supply and demand information, behavior information, patent comparison information, and determining multiple patents corresponding to each information, the relevance of each patent to the information is calculated, and patent recommendations are provided to users.

[0007] The above research results have not established a patent recommendation model method that can be quantified for analysis from a technical or industry perspective. Summary of the invention

[0008] Purpose of the invention: The technical problem to be solved by the present invention is to provide a patent clustering analysis and evaluation recommendation method in the field of air traffic control in view of the shortcomings of the prior art.

[0009] In order to solve the above technical problems, the present invention discloses a patent clustering analysis and evaluation recommendation method in the field of air traffic control, comprising the following steps:

[0010] Step 1: Establish a patent evaluation index system in the field of air traffic control, which includes three dimensions: legal dimension, technical dimension and economic dimension;

[0011] Step 2, calculate the weights of each indicator in the three dimensions;

[0012] Step 3: Cluster the patent collection into patent portfolios with similar characteristics from three dimensions;

[0013] Step 4: Comprehensively score the patents in each patent portfolio from three dimensions;

[0014] Step 5: Generate recommendation solutions in three dimensions, and select the top ten percent of patent portfolios with the highest comprehensive scores as key recommendation solutions.

[0015] Furthermore, the evaluation indicators of the technical value dimension in the patent value evaluation index system in the air traffic control field in step 1 include aviation risk identification, improving flight operation efficiency, flight operation accuracy data improvement rate, improving flight punctuality, improving anti-destruction and fault tolerance, improving flight safety and saving flight operation costs. The corresponding indicator values ​​are respectively identifying dangerous conflict rate, flight operation efficiency improvement efficiency, flight operation accuracy improvement rate, flight punctuality improvement data, anti-destruction and fault tolerance index, improving flight safety data, and saving flight operation costs.

[0016] Furthermore, the evaluation indicators of the legal value dimension in the patent value evaluation index system in the air traffic control field in step 1 include technology coverage, patent citations, patent family size and patent litigation, and the corresponding indicator values ​​are the number of IPC classification items covered, the number of patent citations, the number of authorized patents in the same family and the patent winning rate.

[0017] Furthermore, in step 1, the evaluation indicators of the economic value dimension in the patent value evaluation index system in the field of air pipes include value realization, and the corresponding indicator values ​​include the market share of patent products and the amount of patent conversion.

[0018] Further, step 2 includes:

[0019] Suppose that m patents in the field of air traffic control are to be evaluated, and the patent set to be evaluated is M = (M 1 ,M 2 ,…,M m ), there are n evaluation indicators for each dimension, and the indicator set for each dimension is D = (D 1 ,D 2 ,…,D n ), the evaluation value of the indicator is recorded as x ij , i = 1, 2, ..., m; j = 1, 2, ..., n, the decision matrix X of each dimension is as follows:

[0020]

[0021] In each dimension, the weight y of the jth indicator of the i-th patent ij for Information entropy of each indicator Then use the indicator information entropy to find the corresponding weight

[0022] Further, step 3 includes:

[0023] Collect patent data of corresponding dimensions;

[0024] Preprocess the collected patent data, including data cleaning and standardization, to eliminate the dimensional effects between different indicators. Obtain the corresponding m patent feature vectors;

[0025] Randomly select k samples from m patent feature vectors as initial cluster centers, 1≤k≤m;

[0026] The Euclidean distance between each sample and each cluster center is calculated, and the samples are classified into the cluster with the closest distance. This method is used to calculate the Euclidean distance between patent feature vectors, thereby identifying close or similar patents and forming a k-patent combination.

[0027] Furthermore, the Euclidean distance calculation formula in step 3 is as follows:

[0028]

[0029] Among them, p and q are the L-dimensional feature vectors of the two patents, p l and q l Represent the values ​​of the lth dimension of the feature vectors p and q respectively.

[0030] Further, step 4 includes: for the k patent combinations in step 3, calculating the comprehensive score Z of the hth patent combination h :

[0031]

[0032] Among them, 1≤h≤k, I h represents the number of patents in the h-th patent portfolio,

[0033] Furthermore, in step 1, the risk conflict identification rate = number of successfully identified risks / total number of risks*100%; flight operation efficiency improvement efficiency = (current efficiency value-base period efficiency value) / base period efficiency value*100%; flight operation accuracy improvement rate = (current accuracy value-base period accuracy value) / base period accuracy value*100%; anti-destruction and fault tolerance index = number of successfully handled anomalies / total number of anomalies*100%.

[0034] Furthermore, the preprocessing of the collected patent data in step 3 also includes: for text data, using word embedding technology to convert the text into a vector representation in a high-dimensional space; for image data, using the output layer of the CNN model as a feature vector.

[0035] Compared with the existing achievements, the present invention has the following significant advantages: 1) Based on the technical features and patent characteristics in the field of air traffic control, the present invention conducts patent clustering analysis to achieve patent classification and combination recommendation of similar patents; 2) The present invention analyzes various patent evaluation indicators in the field of air traffic control, constructs corresponding measurement models and uses the entropy weight method to assign weights, which provides support for the systematic evaluation and recommendation of patents in the field of air traffic control. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments, and the above and / or other advantages of the present invention will become more clear.

[0037] Figure 1 A flowchart of a patent clustering analysis and evaluation recommendation method in the field of air traffic control provided in an embodiment of the present application. DETAILED DESCRIPTION

[0038] The embodiments of the present invention will be described below in conjunction with the accompanying drawings.

[0039] The present application embodiment discloses a patent clustering analysis and evaluation recommendation method in the field of air traffic control, such as Figure 1 As shown, the following steps are included:

[0040] Step 1: Taking into account the patent characteristics and technical features of the air traffic control field, following the principles of comprehensiveness, independence and operability in the construction of evaluation indicators, a patent value evaluation index system in the air traffic control field was established, covering evaluation indicators in three dimensions: technical dimension, economic dimension and legal dimension.

[0041]

[0042]

[0043] Among them, the rate of identifying dangerous conflicts = the number of successfully identified risks / the total number of risks*100%; the rate of improving flight operation efficiency = (current efficiency value-base efficiency value) / base efficiency value*100%; the rate of improving flight operation accuracy = (current accuracy value-base accuracy value) / base accuracy value*100%; the index of invulnerability and fault tolerance = the number of successfully handled anomalies / the total number of anomalies*100%.

[0044] Step 2, calculate the weights of each indicator in the three dimensions;

[0045] Step 3: Cluster the patent collection into patent portfolios with similar characteristics from three dimensions;

[0046] Step 4: Comprehensively score the patents in each patent portfolio from three dimensions;

[0047] Specifically, first, analyze and evaluate the patent collection from a legal perspective.

[0048] 1-1. Use the entropy weight method to calculate the weights of the patent analysis evaluation indicators in the legal dimension. Suppose that m patents in the field of air traffic control are evaluated, and the patent set to be evaluated is M = (M 1 ,M 2 ,…,M m ), the legal dimension evaluation indicators are n 1 The indicator set is D = (D 1 ,D 2 ,…,D n1 ), the evaluation value of the indicator is recorded as x ij (i=1,2,…,m; j=1,2,…,n 1 ), and then get the decision matrix X 1 as follows:

[0049]

[0050] The weight of the jth indicator of the i-th patent for Information entropy of each indicator Then use the indicator information entropy to find the corresponding weight

[0051] 1-2. The legal dimension analysis indicator data of patents can be obtained from the patent database. The collected patent data is preprocessed, including data cleaning and standardization, to eliminate the dimensional effects between different indicators. The preprocessing steps include data cleaning, format unification, and removal of invalid or redundant information. The extracted features are converted into numerical vectors. For text data, word embedding technology (such as Word2Vec or BERT) can be used to convert text into a vector representation in a high-dimensional space. For image data, the output layer of the CNN model can be directly used as the feature vector.

[0052] Preliminary determination of the appropriate number of clusters. Randomly select k 1 samples as initial cluster centers, 1≤k 1 ≤m. Calculate the Euclidean distance between each sample and each cluster center, and classify the samples into the cluster with the closest distance. In this way, calculate the Euclidean distance between patent feature vectors, identify close or similar patents, and form the legal dimension k 1 Patent portfolio.

[0053] The Euclidean distance calculation formula is as follows:

[0054]

[0055] Among them, p and q are the L-dimensional feature vectors of the two patents, p l and q l Represent the values ​​of the lth dimension of the feature vectors p and q respectively.

[0056] 1-3: After determining the weights of the legal dimension measurement indicators, the comprehensive score of the patents in the h-th patent portfolio in the legal dimension is obtained

[0057]

[0058] Among them, 1≤h≤k 1 , I h represents the number of patents in the h-th patent portfolio,

[0059] 2. Analyze and evaluate the patent collection from a technical perspective.

[0060] 2-1: Assume that m patents in the field of air traffic control are to be evaluated, and the patent set to be evaluated is M = (M 1 ,M 2 ,…,M m ), the technical dimension evaluation indicators are n 2 The indicator set is D = (D 1 ,D 2 ,…,D n2 ), the evaluation value of the indicator is recorded as x ij(i=1,2,…,m; j=1,2,…,n 2 ), and then get the decision matrix X 2 as follows:

[0061]

[0062] The weight of the jth indicator of the i-th patent for Information entropy of each indicator Then use the indicator information entropy to find the corresponding weight

[0063] 2-2: The technical value indicator data of patents can be obtained from patent databases, industry reports and professional patent analysis tools. Preprocess the collected patent data, including data cleaning and standardization, to eliminate the dimensional effects between different indicators. Extract key technical features and descriptions based on the information in the patent abstract, specification and claims. The preprocessing steps include data cleaning, format unification and removal of invalid or redundant information. Convert the extracted features into numerical vectors. For text data, word embedding technology (such as Word2Vec or BERT) can be used to convert text into a vector representation in a high-dimensional space. For image data, the output layer of the CNN model can be directly used as the feature vector.

[0064] Preliminary determination of the appropriate number of clusters. Randomly select k 2 samples as initial cluster centers, 1≤k 2 ≤m.

[0065] Calculate the Euclidean distance between each sample and each cluster center, and classify the samples into the cluster with the closest distance. This method calculates the Euclidean distance between patent feature vectors, identifies close or similar patents, and forms the technical dimension k 2 Patent portfolio.

[0066] 2-3: After determining the weights of the technical dimension measurement indicators, the comprehensive score of the patents in the hth patent portfolio in the technical dimension is obtained

[0067]

[0068] Among them, 1≤h≤k 2 , I h represents the number of patents in the h-th patent portfolio,

[0069] 3. Analyze and evaluate patent collections from an economic perspective.

[0070] 3-1: Assume that m patents in the field of air traffic control are to be evaluated, and the patent set to be evaluated is M = (M1 ,M 2 ,…,M m ), the economic dimension evaluation indicators are n 3 The indicator set is D = (D 1 ,D 2 ,…,D n3 ), the evaluation value of the indicator is recorded as x ij (i=1,2,…,m; j=1,2,…,n 3 ), and then get the decision matrix X 3 as follows:

[0071]

[0072] The weight of the jth indicator of the i-th patent for Information entropy of each indicator Then use the indicator information entropy to find the corresponding weight

[0073] 3-2: The economic value indicator data of patents can be obtained from patent databases, industry reports and professional patent analysis tools. Preprocess the collected patent data, including data cleaning and standardization, to eliminate the dimensional effects between different indicators. The preprocessing steps include data cleaning, format unification and removal of invalid or redundant information. Convert the extracted features into numerical vectors. For text data, word embedding technology (such as Word2Vec or BERT) can be used to convert text into vector representation in high-dimensional space. For image data, the output layer of the CNN model can be directly used as the feature vector.

[0074] Preliminary determination of the appropriate number of clusters. Randomly select k 3 samples as initial cluster centers, 1≤k 3 ≤m. Calculate the Euclidean distance between each sample and each cluster center, and classify the samples into the cluster with the closest distance. In this way, calculate the Euclidean distance between patent feature vectors, identify close or similar patents, and form the economic dimension k 3 Patent portfolio.

[0075] 3-3: After determining the weights of the economic dimension measurement indicators, the comprehensive patent score of the hth patent portfolio in the economic dimension is obtained

[0076]

[0077] Among them, 1≤h≤k 3 , I h represents the number of patents in the h-th patent portfolio,

[0078] Step 5. Generate recommendation plans in three dimensions, and select the top ten percent of patents with high scores in the three dimensions as key recommendation plans.

[0079] In a specific implementation, the present application provides a computer storage medium and a corresponding data processing unit, wherein the computer storage medium can store a computer program, and when the computer program is executed by the data processing unit, the invention content of a patent clustering analysis and evaluation recommendation method in the field of air traffic control provided by the present invention and some or all of the steps in each embodiment can be run. The storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc.

[0080] Those skilled in the art can clearly understand that the technical solutions in the embodiments of the present invention can be implemented by means of computer programs and their corresponding general hardware platforms. Based on this understanding, the technical solutions in the embodiments of the present invention are essentially or partly contributed to the prior art can be embodied in the form of a computer program, i.e., a software product, which can be stored in a storage medium and includes several instructions for enabling a device including a data processing unit (which can be a personal computer, a server, a single-chip microcomputer, a MUU or a network device, etc.) to execute the methods described in various embodiments of the present invention or certain parts of the embodiments.

[0081] The present invention provides a patent clustering analysis and evaluation recommendation method in the field of air traffic control. There are many methods and ways to implement the technical solution. The above is only a preferred implementation of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the protection scope of the present invention. All components not specified in this embodiment can be implemented by existing technologies.

Claims

1. A patent clustering analysis and evaluation recommendation method in the field of air traffic control, characterized in that: The following steps are involved: Step 1: Establish a patent evaluation index system in the field of air traffic control, which includes three dimensions: legal dimension, technical dimension and economic dimension; Step 2, calculate the weights of each indicator in the three dimensions; Step 3: Cluster the patent collection into patent portfolios with similar characteristics from three dimensions; Step 4: Comprehensively score the patents in each patent portfolio from three dimensions; Step 5: Generate recommendation solutions in three dimensions, and select the top ten percent of patent portfolios with the highest comprehensive scores as key recommendation solutions.

2. According to claim 1, a patent clustering analysis and evaluation recommendation method in the field of air traffic control is characterized in that: In step 1, the evaluation indicators of the technical value dimension in the patent value evaluation index system in the field of air traffic control include aviation risk identification, improving flight operation efficiency, flight operation accuracy data improvement rate, improving flight punctuality, improving anti-destruction and fault tolerance, improving flight safety and saving flight operation costs. The corresponding indicator values ​​are respectively identifying dangerous conflict rate, flight operation efficiency improvement efficiency, flight operation accuracy improvement rate, flight punctuality improvement data, anti-destruction and fault tolerance index, improving flight safety data, and saving flight operation costs.

3. According to claim 2, a patent clustering analysis and evaluation recommendation method in the field of air traffic control is characterized in that: In step 1, the evaluation indicators of the legal value dimension in the patent value evaluation index system in the field of air traffic control include technology coverage, patent citations, patent family size and patent litigation. The corresponding indicator values ​​are the number of IPC classification items covered, the number of patent citations, the number of authorized patents in the same family and the patent winning rate.

4. According to claim 3, the patent clustering analysis and evaluation recommendation method in the field of air traffic control is characterized in that: In step 1, the evaluation indicators of the economic value dimension in the patent value evaluation index system in the field of air pipes include value realization, and the corresponding indicator values ​​include the market share of patent products and the amount of patent conversion.

5. According to claim 4, a patent clustering analysis and evaluation recommendation method in the field of air traffic control is characterized in that: Step 2 includes: Suppose that m patents in the field of air traffic control are to be evaluated, and the patent set to be evaluated is M = (M1, M2, ..., M m ), there are n evaluation indicators for each dimension, and the indicator set for each dimension is D = (D1, D2, ..., D n ), the evaluation value of the indicator is recorded as x ij , i = 1, 2, ..., m; j = 1, 2, ..., n, the decision matrix X of each dimension is as follows: In each dimension, the weight y of the jth indicator of the i-th patent ij for Information entropy of each indicator Then use the indicator information entropy to find the corresponding weight 6. According to claim 5, a patent clustering analysis and evaluation recommendation method in the field of air traffic control is characterized in that: Step 3 includes: Collect patent data of corresponding dimensions; Preprocessing the collected patent data includes data cleaning and standardization to obtain corresponding m patent feature vectors; Randomly select k samples from m patent feature vectors as initial cluster centers, 1≤k≤m; The Euclidean distance between each sample and each cluster center is calculated, and the samples are classified into the cluster with the closest distance, so as to identify close or similar patents and form a k-patent combination.

7. According to claim 6, a patent clustering analysis and evaluation recommendation method in the field of air traffic control is characterized in that: The Euclidean distance calculation formula in step 3 is as follows: Among them, p and q are the L-dimensional feature vectors of the two patents, p l and q l Represent the values ​​of the lth dimension of the feature vectors p and q respectively.

8. According to claim 7, a patent clustering analysis and evaluation recommendation method in the field of air traffic control is characterized in that: Step 4 includes: for the k patent combinations in step 3, calculating the comprehensive score Z of the hth patent combination h : Among them, 1≤h≤k, I h represents the number of patents in the h-th patent portfolio, 9. The patent clustering analysis and evaluation recommendation method in the field of air traffic control according to claim 8 is characterized in that: The risk conflict identification rate in step 1 = number of successfully identified risks / total number of risks*100%; flight operation efficiency improvement efficiency = (current efficiency value-base period efficiency value) / base period efficiency value*100%; flight operation accuracy improvement rate = (current accuracy value-base period accuracy value) / base period accuracy value*100%; invulnerability and fault tolerance index = number of successfully handled anomalies / total number of anomalies*100%.

10. The patent clustering analysis and evaluation recommendation method in the field of air traffic control according to claim 9 is characterized in that: The preprocessing of the collected patent data described in step 3 also includes: for text data, using word embedding technology to convert the text into a vector representation in a high-dimensional space; for image data, using the output layer of the CNN model as a feature vector.

Images