High-value patent recommendation method for enterprise technology innovation stage

Abstract

The invention discloses a high-value patent recommendation method for an enterprise technology innovation stage, and solves the technical problems that the dimension of an existing enterprise portrait is relatively single, and the enterprise demand cannot be well described. The method comprises the steps of analyzing enterprise technology innovation stage demands; and depicting the enterprise demand through the enterprise portrait. The method comprises the following steps: firstly, determining enterprise portrait dimensions, acquiring data and preprocessing the data; secondly, modeling by using a BERTopic topic, and intelligently outputting a portrait label and a topic cluster by an AI (Artificial Intelligence); retrieving to-be-recommended patents of colleges and scientific research institutions according to the labels, and calculating patent value degrees; and finally, calculating the cosine similarity between the patent to be recommended and the theme cluster according to themes, and recommending the patent by using a top-n principle in combination with the similarity and the value degree. According to the method, the requirements of enterprises in different innovation stages are accurately analyzed, so that university and scientific research institute patents are recommended to the enterprises in a personalized manner, and the patent conversion rate and the enterprise technical level can be improved.

Description

Technical Field

[0001] The present invention relates to the field of patent recommendation technology, and in particular to a method for recommending high-value patents for enterprises in the technological innovation stage. Background Art

[0002] Against the backdrop of globalization and rapid technological development, the rapid spread and transformation of technology and knowledge has become an important force in promoting economic growth and social development. As the most effective carrier of technical information, patents are of great significance to the implementation of intellectual property strategies and the research and development of enterprises and institutions. As an important part of patent transactions, patent recommendation has a great boost to patent transformation. The traditional way of recommending patents to enterprises relies on intermediary services or personal network relationships, which limits the efficiency of transactions to a certain extent. With the development of artificial intelligence technologies such as personalized recommendations, it is possible to recommend patents to enterprises intelligently. Most companies purchase patent technologies from universities or research institutions in order to pursue economic value based on technical solutions. They have different needs at different stages of technological innovation. As a tool for describing corporate needs, corporate portraits are rarely used in intelligent patent recommendations. The existing corporate portrait dimensions are relatively single and cannot describe corporate needs well. Summary of the invention

[0003] The purpose of this invention is to provide a high-value patent recommendation method for the technological innovation stage of an enterprise, so as to solve the technical problem that the existing enterprise portrait dimension is relatively single and cannot well characterize the enterprise needs.

[0004] To achieve the above object, the present invention provides the following technical solutions:

[0005] The present invention provides a method for recommending high-value patents for enterprises in the technological innovation stage, comprising the following steps:

[0006] Step 1: Enterprise technology innovation stage demand analysis: By dividing the enterprise technology innovation demand stages, analyze the technology innovation forms and specific innovation needs;

[0007] Step 2: Determine the dimensions of the enterprise portrait and obtain data;

[0008] Step 3: Preprocess the acquired data to obtain enterprise portrait data;

[0009] Step 4: Extract enterprise portrait tags through enterprise portrait data;

[0010] Step 5: Patent search for universities and scientific research institutions to be recommended: write a search formula based on the image label to search for relevant patents in the patent database, and screen out patents whose applicants are universities and scientific research institutions to obtain patents to be recommended;

[0011] Step 6: Similarity calculation: perform cosine similarity calculation on the patent to be recommended and the corresponding topic cluster by topic to obtain the similarity calculation score;

[0012] Step 7: Calculate the patent value: construct a multi-dimensional patent value evaluation index system, use a variety of feature selection methods to simplify the index, and finally combine the entropy weight method to calculate the comprehensive value of the patent to be recommended;

[0013] Step 8. Generate top-n recommendation list: Based on patent similarity and patent value, with a weight of 0.5 for each, calculate the total recommendation score of the patents to be recommended under each topic and sort them, and finally get the recommended patents.

[0014] Furthermore, the step 2 specifically includes the following steps:

[0015] S201. The portrait dimensions include e-commerce platform product reviews, major product-related news, enterprise patent data and product dynamics. The data include Taobao product review data, Sina platform product-related news, incoPat patent database and Weibo product dynamics;

[0016] S202, setting keywords related to the enterprise's products to collect relevant information in a targeted manner;

[0017] S203, using a crawler program to collect information on a set website, including the original text and posting time, to maximize the retention of the original information;

[0018] S204, using Octopus to crawl target information;

[0019] S205. Download the existing patent information of the target enterprise from the incoPat patent database.

[0020] Furthermore, the specific method of step three is: extract the title and abstract of the patent data as the data source, perform text preprocessing on the crawled original information, including removing abnormal, duplicate and missing data, and obtain preprocessed enterprise portrait data.

[0021] Furthermore, the step 4 specifically includes the following steps:

[0022] S401, use BERTopic topic modeling to perform topic clustering on the portrait data;

[0023] S402. Upload the clustering results to ChatGPT and write promt to organize the portrait labels and topic clusters.

[0024] Furthermore, the specific steps of step seven include:

[0025] S701. Construct a multi-dimensional patent value assessment index system, and calculate and standardize the index values;

[0026] S702, using a filtering method to simplify the indicators of the initially constructed indicator system;

[0027] S703. Based on the simplified patent value evaluation index system, the entropy weight method is used to calculate the weight of each index in the final index system.

[0028] Furthermore, in S701, the specific value of each indicator is calculated based on the constructed indicator system, and the indicator is standardized to eliminate the dimension effect, and the range transformation method is used to normalize the data value. The specific calculation formula is:

[0029] If the jth indicator is a positive indicator

[0030]

[0031] If the jth indicator is a negative indicator

[0032]

[0033] Among them, max(x j ) is the maximum value of the j-th indicator data, min(x j ) is the minimum value in the j-th indicator data.

[0034] Furthermore, the filtering method in S702 specifically includes the following steps:

[0035] S7021. Use the variance filtering method to eliminate indicators with variance close to 0. The indicators with variance close to 0 have small variability and limited contribution to the overall difference assessment. Set the variance threshold to 0.01 and eliminate indicators with variance below the threshold. The calculation formula is as follows:

[0036]

[0037] Among them, S j 2 represents the variance of the j-th indicator data, b ij represents the standardized value of the jth indicator of the i-th patent, represents the average value of the jth indicator data, and n is the total number of patents to be evaluated;

[0038] S7022. Identify highly correlated pairs of indicators by calculating the Pearson correlation coefficient of each pair of indicators, and set the correlation threshold to 0.8. The calculation formula is as follows:

[0039]

[0040] Among them, xi and i are the observed values ​​of the two indicators, and are the means of indicator x and indicator y respectively;

[0041] S7023. Based on the results of correlation analysis, the VIF method is further used to diagnose multicollinearity. If the variance inflation factor VIF j If the value is greater than 10, it is considered that the indicator has serious multicollinearity and should be considered for elimination. The calculation formula is as follows:

[0042]

[0043] Among them, R j 2 It is the determination coefficient when the j-th indicator is subjected to auxiliary regression with other indicators.

[0044] Furthermore, the specific steps of calculating the patent value in S703 include:

[0045] S7031. Construct the original indicator matrix and construct the judgment matrix X as follows:

[0046]

[0047] Among them, i represents the number of patents, j represents the number of evaluation indicators, and x ij represents the actual value of the jth indicator of the i-th patent;

[0048] S7032. Calculate the characteristic weight P of the index value of the i-th patent under the j-th index ij , the calculation formula is as follows:

[0049]

[0050] Among them, b ij is the index value after standardization in S701. It will affect the weight calculation, so it is necessary to translate the data with a value of 0, that is, when b ij = 0, b ij =b ij +β, set β = 10 -6 ;

[0051] S7033. Calculate the entropy value E of the j-th indicator j :

[0052]

[0053] S7034. Calculate the weight coefficient w of the jth indicator j :

[0054]

[0055] S7035. Calculate the patent value. After determining the weight of each evaluation indicator, calculate the weighted average value of each evaluation indicator data. The calculation formula is:

[0056]

[0057] Among them, C i is the value of the i-th patent, w j is the weight of the jth indicator, b ij is the standardized score of the jth indicator of the ith patent.

[0058] Based on the above technical solution, the embodiments of the present invention can at least produce the following technical effects:

[0059] The high-value patent recommendation method for the technological innovation stage of enterprises provided by the present invention analyzes the needs of enterprises at different innovation stages, extracts enterprise portrait labels through crawler technology and natural language processing technology, and effectively describes the needs of enterprises, so as to provide personalized recommendations for patents of universities and scientific research institutes for enterprises. It can improve the patent conversion rate of universities and scientific research institutions and help enterprises improve their technical level. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying creative work.

[0061] Figure 1 It is a flow chart of the present invention;

[0062] Figure 2 This is the demand analysis flow chart for the innovation stage of the enterprise of the present invention;

[0063] Figure 3 This is a flow chart of enterprise portrait dimension determination and data acquisition in the present invention;

[0064] Figure 4 This is a flow chart of enterprise portrait label extraction in the present invention;

[0065] Figure 5 This is a flow chart for calculating the patent value of the present invention;

[0066] Figure 6 This is a diagram of the patent value assessment index system for the present invention. DETAILED DESCRIPTION

[0067] The technical solutions in the embodiments of the present invention will be described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the ability of ordinary technicians in this field to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0068] A high-value patent recommendation method for enterprises in the technological innovation stage, such as Figure 1 As shown, the following steps are included:

[0069] Step 1: Analyze the needs of the enterprise's technological innovation stage, divide the enterprise's technological innovation stage, analyze the technology needs and innovation forms. In summary, it is found that there are two forms of innovation: incremental innovation and sudden innovation.

[0070] like Figure 2 As shown, step one specifically includes the following steps:

[0071] S101 sets the innovation demand stage: the innovation demand stage is divided into the leading demand stage, the new leading demand stage, the current leading demand stage and the later leading demand stage;

[0072] S102 Distinguish between forms of technological innovation and specific innovation needs.

[0073] In the leading demand period, there is no real product yet, and there is a potential demand for the envisioned product in the market. When the enterprise expects that it will produce the leading demand product, it will plan, organize and train the corresponding management and technology of the product to seek sudden innovation. In the new leading demand period, the enterprise can make incremental innovation of the product through market feedback information, which is usually not the innovation of the theme product, but the innovation of the form product, such as additional functions, etc. In the current leading demand period, enterprises with competitive pressure must also constantly accept market information feedback, make incremental innovation of products, and seize the market segment. In the post-leading demand period, product innovation is no longer carried out, but preparations for sudden technological innovation have been made, and new functional products with new technologies have been launched. According to the two forms of innovation, the recommendation is divided into two parts. The first part corresponds to the leading demand and post-leading demand in the innovation stage. The form of innovation is sudden innovation, which can be considered from the perspective of disruptive technology identification. The second part corresponds to the new leading demand and current leading demand in the innovation stage. The form of innovation is incremental innovation. This method mainly conducts in-depth research on the second part.

[0074] Step 2: Determine the dimensions of the enterprise portrait and obtain data. Since incremental innovation is mainly based on product feedback in the new and existing demand stages, the portrait dimensions are determined to be product reviews on e-commerce platforms, major product-related news, enterprise patent data, and product dynamics. The data are obtained from Taobao product review data, Sina platform product-related news, incoPat patent database, and Weibo product dynamics through crawlers or database downloads.

[0075] like Figure 3 As shown, step 2 specifically includes the following steps:

[0076] S201 portrait dimensions are determined, including e-commerce platform product reviews, major product-related news, enterprise patent data and product dynamics. The data are obtained from Taobao product review data, Sina platform product-related news, incoPat patent database and Weibo product dynamics.

[0077] S202 Keyword setting: set keywords related to the enterprise's products, such as 'edible oil transportation', 'edible oil storage', 'edible oil testing', so as to collect relevant information in a targeted manner;

[0078] S203 Information Collection: Through crawler programs, information is collected on the specified website, including the original text and posting time, to maximize the retention of the original information;

[0079] S204 sets a crawler program: uses Octopus to crawl target information;

[0080] S205 Patent data download: Download the target company’s existing patent information from the incoPat patent database.

[0081] Step 3: Data preprocessing: For patent data, extract its title and abstract as the data source. Perform text preprocessing on the crawled raw information, including removing abnormal, duplicate and missing data, and obtain preprocessed enterprise portrait data.

[0082] Step 4: Extract enterprise portrait tags, use BERTopic topic modeling to cluster portrait data, and write promt to organize portrait tags and topic clusters through Chatgpt.

[0083] like Figure 4 As shown, step 4 specifically includes the following steps:

[0084] S401BERTopic topic modeling: Use BERTopic topic modeling to perform topic clustering on portrait data;

[0085] S4011 uses BERT to embed vector representation and uses BERT to vectorize the portrait data;

[0086] S4012 uses UMAP to reduce the dimension of the vector, involving parameters that affect the clustering results, mainly n_neighbors, n_components, min_dist and random_state;

[0087] S4013 uses HDBSCAN for clustering, involving parameters that affect the clustering results, mainly min_cluster_size, min_samples, and metric;

[0088] S4014 word segmentation, create a custom dictionary, remove stop words, and use Jieba to segment the original text;

[0089] S4015c-TF-IDF extracts topic candidate words;

[0090] S402Chatgpt organizes portrait labels and topic clusters: upload clustering results to ChatGPT, and write promt to organize portrait labels and topic clusters;

[0091] For example, "The file contains the clustering results of a certain company's product feedback and technical information through BERTopic. Now you are an expert in a certain product and related technology. Can you help me organize the clustering results and give each topic a short title and detailed description?" If the sorting results are not good, manual sorting is required, and promt sorting portrait labels and topic clusters are written for each topic.

[0092] Step 5: Search for patents to be recommended from universities and scientific research institutions. Write a search formula based on the portrait tags to search for relevant patents in the incoPat patent database, and screen out patents whose applicants are universities and scientific research institutions to obtain patents to be recommended.

[0093] Step 6: Similarity calculation: perform cosine similarity calculation on the recommended patent and the corresponding topic cluster by topic to obtain the similarity calculation score.

[0094] Step 7: Calculation of patent value. In order to consider the comprehensive value of the recommended patent, a multi-dimensional patent value evaluation index system is constructed, and feature selection methods such as variance filtering, correlation analysis and multicollinearity test are used to simplify the indicators. Finally, the entropy weight method is used to calculate the comprehensive value of the patent to be recommended.

[0095] like Figure 5 As shown, step seven specifically includes the following steps:

[0096] S701 Construct a multi-dimensional patent value assessment index system, and calculate and standardize the index values;

[0097] like Figure 6As shown, in this embodiment, the multi-dimensional patent value assessment index system includes legal dimension, technical dimension and economic dimension;

[0098] In this embodiment, the evaluation indicators of the legal dimension include the remaining validity period of the patent, the number of simple families, the number of countries in the family, the number of extended families, the invalidation status of reexamination, the number of claims, the number of independent claims, the number of dependent claims and the number of pages of literature;

[0099] In this embodiment, the evaluation indicators of the technology dimension include the number of cited patents, technological originality, scientific relevance, number of citations, other-citation rate, number of IPCs, number of IPCs of cited patents, number of IPCs of citing patents, number of subject classifications, proportion of self-cited patents, and technology cycle;

[0100] In this embodiment, the evaluation indicators of the economic dimension include the number of patents of the patent owner, the type of applicant, the number of applicants, the number of inventors, the number of citations of the family, the number of cited applicants of the family, the number of national economic classifications, the number of national economic industries, the transfer license status, the financing and insurance status, and the litigation and arbitration status.

[0101] The specific values ​​of each indicator are calculated based on the constructed indicator system, and the indicators are standardized to eliminate the dimension effect. The range transformation method is used to normalize the data values. The specific calculation formula is:

[0102] If the jth indicator is a positive indicator (the larger the better)

[0103]

[0104] If the jth indicator is a negative indicator (the smaller the better)

[0105]

[0106] Among them, max(x j ) is the maximum value of the j-th indicator data, min(x j ) is the minimum value in the j-th indicator data.

[0107] S702 uses a filtering method to simplify the indicators of the initially constructed indicator system;

[0108] In this embodiment, the filtering method includes variance filtering, correlation analysis and multicollinearity test.

[0109] Specifically, the filtering method in S702 includes the following steps:

[0110] S7021 uses the variance filtering method to remove indicators with variance close to 0. Indicators with variance close to 0 have small variability and limited contribution to the overall difference assessment. Set the variance threshold to 0.01 and remove indicators with variance below the threshold. The calculation formula is as follows:

[0111]

[0112] Among them, S j 2 represents the variance of the j-th indicator data, b ij represents the standardized value of the jth indicator of the i-th patent, represents the average value of the jth indicator data, and n is the total number of patents to be evaluated;

[0113] S7022 identifies highly correlated pairs of indicators by calculating the Pearson correlation coefficient for each pair of indicators and sets the correlation threshold to 0.8. The calculation formula is as follows:

[0114]

[0115] Among them, x i and i are the observed values ​​of the two indicators, and are the means of indicator x and indicator y respectively;

[0116] S7023 Based on the results of correlation analysis, the VIF method is further used for multicollinearity diagnosis. If the variance inflation factor VIF j If the value is greater than 10, it is considered that the indicator has serious multicollinearity and should be considered for elimination. The calculation formula is as follows:

[0117]

[0118] Among them, R j 2 is the determination coefficient when the j-th indicator is subjected to auxiliary regression with other indicators;

[0119] S703 uses the entropy weight method to calculate the weight of each indicator in the final indicator system based on the simplified patent value evaluation indicator system. First, the entropy weight method is used to calculate the weight of the three-level indicators of patent value evaluation, and then the weight of the three-level indicators is added to obtain the weight of the second-level indicators, and finally the patent value is calculated;

[0120] Specifically, the specific steps of calculating the patent value in S703 include:

[0121] S7031 constructs the original indicator matrix. Assuming there are i patents and j evaluation indicators, the judgment matrix X is constructed as follows:

[0122]

[0123] Among them, xij represents the actual value of the jth indicator of the i-th patent;

[0124] S7032 calculates the characteristic weight of the index value of the i-th patent under the j-th index. The calculation formula is as follows:

[0125]

[0126] Among them, b ij is the index value after standardization in S701. It will affect the weight calculation, so it is necessary to translate the data with a value of 0, that is, when b ij = 0, b ij =b ij +β, in order to minimize the impact of the translated data on the results, β should be closest to 0, set β = 10 -6 ;

[0127] S7033 Calculate the entropy value E of the j-th indicator j :

[0128]

[0129] S7034 calculates the weight coefficient of the jth indicator:

[0130]

[0131] S7035 calculates the patent value. After determining the weight of each evaluation indicator, the weighted average value is calculated from the data of each evaluation indicator. The calculation formula is:

[0132]

[0133] Among them, C i is the value of the i-th patent, w j is the weight of the jth indicator, b ij is the standardized score of the jth indicator of the ith patent.

[0134] Step 8: Generate the top-n recommendation list. Based on the patent similarity and patent value, with a weight of 0.5 for each, calculate the total recommendation score of the patents to be recommended under each topic and sort them, and finally get the recommended patents.

[0135] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments, and the above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, and these changes and improvements fall within the scope of the present invention to be protected. The scope of the present invention to be protected is defined by the attached claims and their equivalents.

Claims

1. A method for recommending high-value patents for the technological innovation stage of an enterprise, characterized in that: The following steps are involved: Step 1: Enterprise technology innovation stage demand analysis: By dividing the enterprise technology innovation demand stages, analyze the technology innovation forms and specific innovation needs; Step 2: Determine the dimensions of the enterprise portrait and obtain data; Step 3: Preprocess the acquired data to obtain enterprise portrait data; Step 4: Extract enterprise portrait tags through enterprise portrait data; Step 5: Patent search for universities and scientific research institutions to be recommended: write a search formula based on the image label to search for relevant patents in the patent database, and screen out patents whose applicants are universities and scientific research institutions to obtain patents to be recommended; Step 6: Similarity calculation: perform cosine similarity calculation on the patent to be recommended and the corresponding topic cluster by topic to obtain the similarity calculation score; Step 7: Calculate the patent value: construct a multi-dimensional patent value evaluation index system, use a variety of feature selection methods to simplify the index, and finally combine the entropy weight method to calculate the comprehensive value of the patent to be recommended; Step 8. Generate top-n recommendation list: Based on patent similarity and patent value, with a weight of 0.5 for each, calculate the total recommendation score of the patents to be recommended under each topic and sort them, and finally get the recommended patents.

2. The high-value patent recommendation method for the enterprise technology innovation stage according to claim 1 is characterized in that: The step 2 specifically includes the following steps: S201. The portrait dimensions include e-commerce platform product reviews, major product-related news, enterprise patent data and product dynamics. The data include Taobao product review data, Sina platform product-related news, incoPat patent database and Weibo product dynamics; S202, setting keywords related to the enterprise's products to collect relevant information in a targeted manner; S203, using a crawler program to collect information on a set website, including the original text and posting time, to maximize the retention of the original information; S204, using Octopus to crawl target information; S205. Download the existing patent information of the target enterprise from the incoPat patent database.

3. The high-value patent recommendation method for the enterprise technology innovation stage according to claim 1 is characterized in that: The specific method of step three is: extract the title and abstract of the patent data as the data source, perform text preprocessing on the crawled original information, including removing abnormal, duplicate and missing data, and obtain preprocessed enterprise portrait data.

4. The high-value patent recommendation method for the enterprise technology innovation stage according to claim 1 is characterized in that: The step 4 specifically includes the following steps: S401, use BERTopic topic modeling to perform topic clustering on the portrait data; S402. Upload the clustering results to ChatGPT and write promt to organize the portrait labels and topic clusters.

5. The high-value patent recommendation method for the enterprise technology innovation stage according to claim 1 is characterized in that: The specific steps of step seven include: S701. Construct a multi-dimensional patent value assessment index system, and calculate and standardize the index values; S702, using a filtering method to simplify the indicators of the initially constructed indicator system; S703. Based on the simplified patent value evaluation index system, the entropy weight method is used to calculate the weight of each index in the final index system.

6. The high-value patent recommendation method for the enterprise technology innovation stage according to claim 5 is characterized in that: In S701, the specific value of each indicator is calculated based on the constructed indicator system, and the indicator is standardized to eliminate the dimension effect. The range transformation method is used to normalize the data value. The specific calculation formula is: If the jth indicator is a positive indicator If the jth indicator is a negative indicator Among them, max(x j ) is the maximum value of the j-th indicator data, min(x j ) is the minimum value in the j-th indicator data.

7. The high-value patent recommendation method for the enterprise technology innovation stage according to claim 5 is characterized in that: The specific steps of the filtering method in S702 include: S7021. Use the variance filtering method to eliminate indicators with variance close to 0. The indicators with variance close to 0 have small variability and limited contribution to the overall difference assessment. Set the variance threshold to 0.01 and eliminate indicators with variance below the threshold. The calculation formula is as follows: Among them, S j 2 represents the variance of the j-th indicator data, b ij represents the standardized value of the jth indicator of the i-th patent, represents the average value of the jth indicator data, and n is the total number of patents to be evaluated; S7022. Identify highly correlated pairs of indicators by calculating the Pearson correlation coefficient of each pair of indicators, and set the correlation threshold to 0.

8. The calculation formula is as follows: Among them, x i and i are the observed values ​​of the two indicators, and are the means of indicator x and indicator y respectively; S7023. Based on the results of correlation analysis, the VIF method is further used to diagnose multicollinearity. If the variance inflation factor VIF j If the value is greater than 10, it is considered that the indicator has serious multicollinearity and should be considered for elimination. The calculation formula is as follows: Among them, R j 2 It is the determination coefficient when the j-th indicator is subjected to auxiliary regression with other indicators.

8. The high-value patent recommendation method for the enterprise technology innovation stage according to claim 5 is characterized in that: The specific steps of calculating the patent value in S703 include: S7031. Construct the original indicator matrix and construct the judgment matrix X as follows: Among them, i represents the number of patents, j represents the number of evaluation indicators, and x ij represents the actual value of the jth indicator of the i-th patent; S7032. Calculate the characteristic weight P of the index value of the i-th patent under the j-th index ij , the calculation formula is as follows: Among them, b ij is the index value after standardization in S701. It will affect the weight calculation, so it is necessary to translate the data with a value of 0, that is, when b ij = 0, b ij =b ij +β, set β = 10 -6 ; S7033. Calculate the entropy value E of the j-th indicator j : S7034. Calculate the weight coefficient w of the jth indicator j : S7035. Calculate the patent value. After determining the weight of each evaluation indicator, calculate the weighted average value of each evaluation indicator data. The calculation formula is: Among them, C i is the value of the i-th patent, w j is the weight of the jth indicator, b ij is the standardized score of the jth indicator of the ith patent.

Images