Patent litigation compensation prediction method and device, and storage medium

By constructing a patent litigation compensation prediction model and processing and analyzing patent data, the problem of inaccurate prediction of patent litigation compensation in the existing technology is solved, and the accuracy of the prediction results is improved.

CN118940883AInactive Publication Date: 2024-11-12HUAZHONG UNIV OF SCI & TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410959264.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-17
Publication Date
2024-11-12
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing technology lacks effective methods for predicting patent litigation compensation, and it is difficult to accurately predict compensation for patent infringement litigation cases.

Method used

By acquiring and processing a large amount of patent data, a patent litigation compensation prediction model is built, including data cleaning, coding, preprocessing, significance analysis and model training, and prediction is made using linear regression, decision trees and other models.

Benefits of technology

It improves the accuracy of the prediction results of patent litigation compensation, accurately eliminates irrelevant indicators, and enhances the reliability of the prediction.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118940883A_ABST
    Figure CN118940883A_ABST
Patent Text Reader

Abstract

The invention discloses a patent litigation compensation prediction method, which comprises the steps of extracting patent indexes and data, and inputting the patent indexes and data into a patent litigation compensation prediction model to obtain a compensation prediction result in a patent; the patent litigation compensation prediction model is constructed according to the following steps: obtaining indexes, data and patent litigation results of a plurality of patents; performing data cleaning and normalization on the numerical data, and performing data coding on the classified data; preprocessing the numerical index and the classification index; and constructing a data set by using the preprocessed indexes and data, and training the initial regression model to obtain a patent litigation compensation prediction model. According to the method, a large amount of patent data is acquired and processed to fit the patent litigation compensation prediction model, so that the accuracy of patent litigation compensation prediction can be effectively improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of big data and artificial intelligence technology, and more specifically, to a method, device and storage medium for predicting damages in patent litigation. Background Art

[0002] Patents are both a carrier of technical information and a legal document. With the increasing number of patent infringement lawsuits, whether patents can obtain high compensation in infringement lawsuits has become an issue of increasing concern.

[0003] The prediction of compensation for patents in litigation cases can be regarded as a regression problem. With the development of big data and artificial intelligence technology, it has become possible to use big data to organize, classify and establish regression models for existing patent infringement litigation cases. The common patents related to amount in the prior art are mostly limited to the prediction of compensation in ordinary civil and commercial litigation or the prediction of fines in criminal cases, and there is still a lack of compensation prediction methods for patent cases. Summary of the invention

[0004] In response to at least one defect or improvement need in the prior art, the present invention provides a method, device and storage medium for predicting compensation for patent litigation. By acquiring and processing a large amount of patent data to fit a patent litigation compensation prediction model, the accuracy of the patent litigation compensation prediction results can be effectively improved.

[0005] To achieve the above-mentioned purpose, according to a first aspect of the present invention, a method for predicting damages in patent litigation is provided, the method comprising:

[0006] Extract the numerical index, classification index and corresponding numerical data and classification data of the patent, input them into the patent litigation compensation prediction model, and obtain the compensation prediction result of the patent in a single patent infringement lawsuit; characterized in that the patent litigation compensation prediction model is constructed according to the following steps:

[0007] Obtain numerical indicators and their corresponding numerical data, classification indicators and their corresponding classification data, and patent litigation results of multiple patents;

[0008] Perform data cleaning and normalization on numerical data and data encoding on categorical data;

[0009] Using the numerical data after data cleaning, the coded classification data and the patent litigation results, the numerical indicators and classification indicators are preprocessed to obtain the preprocessed numerical indicators and the preprocessed classification indicators;

[0010] A data set is constructed using the preprocessed numerical indicators and their corresponding numerical data, and the preprocessed classification indicators and their corresponding classification data. The data set is used to train the initial regression model to obtain a patent litigation compensation prediction model.

[0011] Furthermore, the above patent litigation compensation prediction method also includes:

[0012] Data encoding for categorical data specifically includes:

[0013] Encode categorical data into 1 and 0 according to the data type.

[0014] Furthermore, the above patent litigation compensation prediction method also includes:

[0015] Conduct significance analysis on numerical indicators and classification indicators, and eliminate those that are not significant;

[0016] Calculate the variance of the numerical data corresponding to each numerical indicator, and eliminate the numerical indicators whose variance is less than the variance threshold;

[0017] Conduct correlation analysis on numerical indicators and eliminate those that are not relevant.

[0018] Furthermore, the above patent litigation compensation prediction method also includes:

[0019] Calculate the average of all patent litigation results to obtain the patent litigation result benchmark; take the data in the patent litigation results that are greater than the patent litigation result benchmark as "high compensation in patent litigation", and take the data that is less than or equal to the patent litigation result benchmark as "low compensation in patent litigation", take the data in the patent litigation results that are greater than the patent litigation result benchmark as the sample group, and take the data in the patent litigation results that are less than the patent litigation result benchmark as the control group, and let the null hypothesis be "there is no significant difference between high compensation in patent litigation and the corresponding numerical indicator / classification indicator", and the alternative hypothesis be "there is a significant difference between high compensation in patent litigation and the corresponding numerical indicator / classification indicator".

[0020] Furthermore, the above patent litigation compensation prediction method also includes:

[0021] The Mann-Whitney U test was used to calculate the p-value, and the numerical indicators with p-values ​​greater than the first significance threshold were eliminated.

[0022] Furthermore, the above patent litigation compensation prediction method also includes:

[0023] Fisher's exact probability method was used to calculate the p-value, and the classification indicators with p-values ​​greater than the first significance threshold were eliminated.

[0024] Furthermore, the above patent litigation compensation prediction method also includes:

[0025] The Spearman rank correlation coefficient between the numerical indicator and the patent litigation results and its corresponding significance p-value are calculated, and the numerical indicators whose significance p-value of the Spearman rank correlation coefficient is greater than the second significance threshold are eliminated.

[0026] Furthermore, the above patent litigation compensation prediction method also includes:

[0027] The initial regression model is one of a linear regression model, a decision tree model, a gradient boosting tree model and a random forest model.

[0028] According to the second aspect of the present invention, there is also provided a patent litigation compensation prediction device, which includes at least one processing unit and at least one storage unit, wherein the storage unit stores a computer program, and when the computer program is executed by the processing unit, the processing unit executes the steps of any one of the above-mentioned methods.

[0029] According to a third aspect of the present invention, there is also provided a storage medium storing a computer program executable by an access authentication device, wherein when the computer program runs on the access authentication device, the access authentication device executes the steps of any one of the above methods.

[0030] In general, the above technical solutions conceived by the present invention can achieve the following beneficial effects compared with the prior art:

[0031] (1) The patent litigation compensation prediction method provided by the present invention obtains a large amount of patent data and processes it to fit a patent litigation compensation prediction model, which can effectively improve the accuracy of the patent litigation compensation prediction results.

[0032] (2) The patent litigation compensation prediction method provided by the present invention adopts different screening methods for common numerical indicators and classification indicators in patent analysis, accurately eliminates irrelevant indicators, and further improves the accuracy of patent litigation compensation prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments 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 these drawings without creative work.

[0034] Figure 1 A flowchart of a modeling method for predicting a patent litigation compensation amount provided by an embodiment of the present invention;

[0035] Figure 2 A schematic diagram of a decision tree model provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0036] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0037] The terms "first", "second", "third", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not limited to the listed steps or units, but optionally includes steps or units that are not listed, or optionally includes other steps or units inherent to these processes, methods, products or devices.

[0038] As a first embodiment of the present invention, a method for predicting compensation for patent litigation is provided. This embodiment uses past patent litigation cases that have been awarded compensation to establish a patent litigation compensation prediction model, and then inputs the numerical indicators, classification indicators and corresponding numerical data and classification data of a single patent into the patent litigation compensation prediction model to obtain the predicted value of compensation that the patent may receive in a single patent infringement lawsuit.

[0039] For information on how to build a patent litigation damages prediction model, see Figure 1 , this embodiment provides the following steps:

[0040] S1. Obtain the numerical indicators and their corresponding numerical data, classification indicators and their corresponding classification data, and patent litigation results of multiple patents.

[0041] In this embodiment, the various indicators of the patent are divided into numerical indicators and classified indicators. Numerical indicators may include indicators with specific numbers such as the number of first rights and authorization period, and classified indicators may include indicators with only "yes" and "no" such as whether there is an agency and whether the applicant is a natural person. This embodiment also includes collecting data corresponding to the above indicators.

[0042] The results of patent litigation in the present invention specifically refer to the amount of compensation awarded to the plaintiff (i.e., the patent owner) in a patent infringement litigation case in which the plaintiff wins the case. The results of patent litigation can be obtained by first obtaining the corresponding judgment documents on various official websites, such as the "Judgment Documents Network" of the Supreme People's Court, or other commercial databases, such as Peking University Law Treasure and Weil Pioneer, and then processing the relevant fields in the judgment documents. As for how to obtain compensation data from judgment documents, the prior art recorded in patents such as CN108287818A and CN106815203A can be used to obtain it, which will not be repeated in this embodiment. In this embodiment, the specific number of the judgment compensation amount obtained as a result of the patent litigation is recorded in a numerical data format.

[0043] It should be noted that the present invention only focuses on patents that have won patent infringement lawsuits. Therefore, for judgment documents without compensation in the results of patent litigation, they can be directly excluded without being extracted as original data. On the other hand, when the defendant's legitimate source defense is established in a patent infringement lawsuit, the court will usually rule that the defendant only needs to pay the plaintiff's reasonable expenses for defending his rights. The data in this case can also be excluded. The specific exclusion method can be either manual screening and exclusion, or exclusion through methods such as machine semantic recognition, which will not be repeated in this embodiment.

[0044] S2. Perform data cleaning and normalization on numerical data and data encoding on categorical data.

[0045] In this embodiment, not all the original data extracted in the previous step are applicable to the subsequent modeling process of the classification model, and data cleaning and data encoding are required.

[0046] The first step is data cleaning for numerical data. Specifically, it is to detect numerical data, process missing values, remove outliers and duplicate records. The missing values ​​of numerical data are replaced by the median value of the numerical indicator.

[0047] Furthermore, the numerical data needs to be normalized. In this embodiment, the following normalization method can be used, namely:

[0048]

[0049] For categorical data, data encoding is required. Since the results of categorical data usually extracted are "yes" and "no", the classification results can also be considered as "yes" and "no", so the encoding process can be regarded as encoding one of the data as 1 and the other as 0.

[0050] For example, when the classification indicator is "the applicant is a natural person", then in the classification data, "the applicant is a natural person" can be coded as 1, and "the applicant is not a natural person" can be coded as 0. Since the classification results and classification data in this embodiment are binary data, that is, there are only data similar to "yes" and "no", the encoding process only needs to encode one category as 1 and the other category as 0 to distinguish the two categories of data.

[0051] S3. Use the numerical data after data cleaning, the coded classification data and the patent litigation results to preprocess the numerical indicators and classification indicators to obtain the preprocessed numerical indicators and the preprocessed classification indicators.

[0052] Considering that there are many types of numerical indicators and classification indicators extracted in the first step, incorporating too many indicators into the training of the classification model will not only seriously prolong the training time, but also lead to poor training results. Therefore, it is necessary to preprocess various indicators.

[0053] The pre-processing step may include the following steps.

[0054] S31. Perform significance analysis on numerical indicators and classification indicators, and eliminate those that are not significant.

[0055] The traditional significance analysis method requires setting the null hypothesis and the alternative hypothesis. Usually, the null hypothesis is that the two variables are unrelated, and the alternative hypothesis is the opposite of the null hypothesis. The p-value is calculated based on the collected data. If the p-value is less than the significance level (usually set to 0.05), the null hypothesis is rejected and the alternative hypothesis is supported. Otherwise, the null hypothesis is accepted. The p-value reflects the probability of the observed data when the null hypothesis is true. The smaller the p-value, the less likely it is to be caused by random chance. The above is a general method for significance analysis.

[0056] Specifically in this embodiment, significance analysis can be performed by the following method.

[0057] First, the average of all patent litigation results is calculated to obtain the patent litigation result benchmark. The data of patent litigation results that are greater than the patent litigation result benchmark are regarded as "high compensation for patent litigation", and the data that are less than or equal to the patent litigation result benchmark are regarded as "low compensation for patent litigation".

[0058] Furthermore, the data in the patent litigation results that are greater than the patent litigation results benchmark are taken as the sample group, and the data in the patent litigation results that are less than the patent litigation results benchmark are taken as the control group, and the null hypothesis is set as "there is no significant difference between high compensation in patent litigation and the corresponding numerical indicators / classification indicators", and the alternative hypothesis is "there is a significant difference between high compensation in patent litigation and the corresponding numerical indicators / classification indicators".

[0059] Considering the difference between numerical indicators and classification indicators, in this embodiment, different test methods are required for numerical indicators and classification indicators. For numerical indicators, considering the large amount of data and the fact that each indicator does not obey the normal distribution, the Mann-Whitney U test is used for numerical indicators. For classification indicators, Fisher's exact probability method has a better effect, so Fisher's exact probability method is used for classification indicators in this embodiment. Indicators with a calculated p value greater than the first significance threshold can be eliminated, and the first significance threshold can be set to the common 0.05.

[0060] The above-mentioned significance analysis method is used for preprocessing to screen out the numerical indicators and classification indicators with significant differences after preprocessing.

[0061] S32. Calculate the variance of the numerical data corresponding to each numerical indicator, and eliminate the numerical indicators whose variance is less than the variance threshold.

[0062] Variance is a statistical indicator that measures the degree of data change. For an indicator, if its variance is small, it means that the sample value on this indicator does not change much; on the contrary, if the variance is large, it means that the sample value on this indicator changes a lot. Numerical indicators with little change in value contain insufficient information and are not suitable as data input for classification model training.

[0063] S33. Perform correlation analysis on the numerical indicators and eliminate the numerical indicators that are not relevant.

[0064] Correlation analysis can be performed by calculating the Spearman rank correlation coefficient. The Spearman rank correlation coefficient is a statistical indicator used to measure the nonlinear relationship between two variables. By converting the numerical value of each variable into a ranking, the coefficient measures the correlation of the variable ranking rather than the actual numerical value. Its value range is between -1 and 1, 1 represents a complete positive correlation, -1 represents a complete negative correlation, and 0 represents no correlation. In this embodiment, the two variables are the patent litigation results and the numerical indicator.

[0065] The Spearman rank correlation coefficient is calculated, and the corresponding significance p-value of the Spearman rank correlation coefficient is calculated at the same time, and the indicators whose significance p-value of the point-by-point correlation coefficient is greater than the second significance threshold (usually set to 0.05) are eliminated, that is, the numerical indicators with no correlation can be eliminated.

[0066] S4. Use the preprocessed numerical indicators and their corresponding numerical data, the preprocessed classification indicators and their corresponding classification data to construct a data set, and use the data set to train the initial regression model to obtain a patent litigation compensation prediction model.

[0067] Among them, the initial regression model is one of a linear regression model, a decision tree model, a gradient boosting tree model and a random forest model.

[0068] As a second embodiment of the present invention, a specific implementation method of constructing a patent litigation compensation prediction model for application in the first embodiment is also provided.

[0069] In this embodiment, the first step is to obtain data. This embodiment selects patents that have been sued and won as a sample group: since patent infringement lawsuits generally have a claim amount, if the case can be won, there will be a judgment amount of compensation. Therefore, as long as there is a case with a judgment amount of compensation, it belongs to the patents that have been sued and won in this embodiment. First, the 17,705 patent cases that have experienced patent infringement lawsuits in China from January 1, 1999 to December 5, 2023, which were published by the Supreme People's Court on the "Judgment Documents Network", were obtained, and the cases with specific values ​​of the total amount of compensation applied for and the total amount of compensation awarded were further screened. A total of 11,471 patent litigation cases were obtained, and the corresponding data were extracted from them. The relevant data of the patents involved in the above cases were further retrieved through the Wisdom Bud Patent Database to form the initial data set.

[0070] Regarding the selection of indicators, the following common indicators related to patent litigation compensation are selected in this embodiment, which can be seen in Table 1.

[0071] Table 1 Selection of numerical and classification indicators in patent litigation compensation prediction

[0072]

[0073]

[0074] The second step is to clean and normalize the numerical data and encode the categorical data. The encoding method of categorical data can also be referred to Table 1.

[0075] The third step is the screening of indicators. The first step is to conduct a significance analysis. In this embodiment, IBM SPSS Statistics software is used to calculate the mean of various indicators including compensation in 11,471 patent litigation cases, where the mean of compensation is rounded to 100,000 yuan. Then, according to the mean of compensation, the 11,471 patent litigation cases are divided into two groups based on whether the compensation is greater than 100,000 yuan. The sample group is 3,152 patents with compensation greater than 100,000 yuan, which is the high compensation group; the control group is 8,319 patents with compensation less than or equal to 100,000 yuan, which is the low compensation group. Let the original hypothesis be "there is no significant difference in the numerical indicators corresponding to high and low compensation in patent litigation", and the alternative hypothesis is "there is a significant difference in the numerical indicators corresponding to high and low compensation in patent litigation". The results of the significance analysis are shown in Table 2.

[0076] Table 2 Significance analysis of numerical indicators

[0077]

[0078]

[0079] The results show that the p-values ​​of 20 numerical indicators, including the total amount of compensation applied for, the total number of cited patents of simple same family, patent type, the number of litigation judgments, the year from application to filing, the number of inventors, the number of first-right words, the authorization cycle, the total number of cited patents, the total number of cited patents, the number of citations within 3 years, the number of citations within 5 years, the number of document pages, the number of original applicants, the number of IPC classifications, the total number of cited patents of Zhihuiya family, Zhihuiya family, the number of lawsuits, the number of licenses, and the number of pledges, are all significant at the 5% confidence level.

[0080] Therefore, it can be concluded that 20 numerical indicators, including total amount of compensation applied for, total number of cited patents in simple family, patent type, number of litigation judgments, year from application to filing, number of inventors, number of first rights, authorization cycle, total number of cited patents, total number of cited patents, number of citations within 3 years, number of citations within 5 years, number of pages of literature, number of original applicants, number of IPC classifications, total number of cited patents in Zhihuiya family, Zhihuiya family, number of lawsuits, number of licenses, number of pledges, etc., have significant differences under different compensation amounts. It is meaningful to analyze them with different compensation amounts and they will be included in the subsequent analysis. The other two numerical indicators are not significant and will be removed in the next analysis.

[0081] The classification indicators include whether the applicant is a natural person, whether the patentee is related to foreign countries or Hong Kong, Macao and Taiwan, whether the applicant is related to foreign countries or Hong Kong, Macao and Taiwan, whether there is an agent at the time of application, whether the plaintiff and the patent applicant are the same, and whether the court and the patentee are in the same region. In the sample group, the number of patents with natural person applicants accounted for 29.7%, the number of patents with foreign or Hong Kong, Macao and Taiwan patentees accounted for 21.6%, the number of patents with foreign or Hong Kong, Macao and Taiwan applicants accounted for 25.3%, the number of patents with agents at the time of application accounted for 80.4%, the number of patents with the same plaintiff and applicant accounted for 50.1%, and the number of patents with the same court and patentee accounted for 53.7%. In the control group, the number of patents whose applicants were natural persons accounted for 41.9%, the number of patents whose patentees were foreign or involved Hong Kong, Macao and Taiwan accounted for 5.5%, the number of patents whose applicants were foreign or involved Hong Kong, Macao and Taiwan accounted for 7.0%, the number of patents whose applications had agents accounted for 77.0%, the number of patents whose plaintiffs and applicants were the same accounted for 68.5%, and the number of patents whose courts and patentees were in the same region accounted for 66.6%. Fisher's exact probability method was used for the above data, and the results of significance analysis are shown in Table 3.

[0082] Table 3 Significance analysis of classification indicators

[0083]

[0084] Note: *** indicates significance at the 0.01 confidence level, ** indicates significance at the 0.05 confidence level, and * indicates significance at the 0.1 confidence level.

[0085] The above table shows that the test result of judging whether the applicant is a natural person is significant at the 10% confidence level. That is, in the two groups of patents in the high compensation group and the low compensation group, the proportion of applicants who are natural persons is significantly different, which is an important indicator and needs to be introduced in the next analysis. The other five classification indicators, patent owners involving foreign countries or Hong Kong, Macao and Taiwan, applicants involving foreign countries or Hong Kong, Macao and Taiwan, the court and patent owners are in the same region, whether the plaintiff and the applicant are the same, and whether there is an agency at the time of application, do not have significant differences in the high compensation group and the low compensation group, and are removed in the next analysis.

[0086] The next step is to calculate the variance of the numerical data corresponding to each numerical indicator, as shown in Table 4.

[0087] Table 4 Variance of each numerical index

[0088]

[0089] The variance threshold selected in this embodiment is 10. As can be seen from Table 4, the variances of the 13 indicators, including the total amount of compensation applied for, the total number of cited patents of the simple same family, the number of litigation judgments, the year from application to filing, the number of first rights, the authorization cycle, the total number of cited patents, the number of citations within 3 years, the number of citations within 5 years, the number of pages of literature, the total number of cited patents of the same family of Zhihuiya, the total number of cited patents of the same family of Zhihuiya, and the number of litigations, are greater than 10, and can be introduced into the next step of analysis.

[0090] The correlation between the numerical index and the classification results is further analyzed. Since the significance and correlation of the classification index are the same, the classification index does not need to calculate the correlation. The Spearman rank correlation coefficient analysis method is used to obtain the correlation calculation results of the numerical index as shown in Table 5.

[0091] Table 5 Correlation analysis of numerical indicators

[0092]

[0093] As shown in Table 5, the total amount of compensation applied for, the number of first-right words, the authorization cycle, and the number of pages of literature are positively correlated with high patent compensation; the total number of cited patents in the simple same family, the number of litigation judgments, the year from application to filing, the total number of cited patents, the number of citations within 3 years, the number of citations within 5 years, and the number of lawsuits are negatively correlated with high patent compensation; the test results of the total number of cited patents in the same family of Zhihuiya and the total number of cited patents in the same family of Zhihuiya are not significant at the 5% confidence level, so the total number of cited patents in the same family of Zhihuiya and the total number of cited patents in the same family of Zhihuiya have no obvious correlation with patent damages. Therefore, 11 numerical indicators such as the total amount of compensation applied for, the total number of cited patents in the simple same family, the number of litigation judgments, the year from application to filing, the number of first-right words, the authorization cycle, the total number of cited patents, the number of citations within 3 years, the number of citations within 5 years, the number of pages of literature, and the number of lawsuits, and the categorical indicator that the applicant is a natural person, a total of 12 related indicators will be introduced into the next step of modeling.

[0094] In this embodiment, the above 11 numerical indicators and 1 classification indicator, as well as the corresponding classification data and numerical data, are combined into a data set, and the data set is divided into a training set and a test set in a ratio of 4:1 for model training and testing.

[0095] In the model selection phase, a variety of different types of regression models were considered to ensure that possible complex relationships could be captured. To this end, linear regression models, decision tree models, gradient boosted tree models, and random forest models were selected as initial regression models and experiments were conducted.

[0096] The accuracy of the model is explained by the values ​​of the coefficient of determination and the mean square error. The coefficient of determination reflects the proportion of the total variance of the target variable explained by the model. A coefficient of determination value of 0 means that the model fails to explain any variance of the target variable, while a coefficient of determination value of 1 means that the model fully explains all variance of the target variable. Therefore, the closer the coefficient of determination is to 1, the stronger the model's ability to explain the target variable. The mean square error is used to measure the prediction error of the model, that is, the average of the squared differences between the model's prediction of the actual observed value and the actual value, so the smaller the better.

[0097] The training set is used to train the above four initial regression models, and the test set is used to verify the training performance of each model. The results are shown in Table 6.

[0098] Table 6 Training performance of different models

[0099]

[0100] The experiment obtains that the linear regression coefficient of determination is 0.0098, and the mean square error is 6.8509E-06, the gradient boosting tree coefficient of determination is 0.7347, and the mean square error is 1.8354E-06, the decision tree coefficient of determination is 0.7522, and the mean square error is 1.7143E-06, and the random forest coefficient of determination is 0.6051, and the mean square error is 2.7323E-06. By comparing the coefficient of determination and the mean square error of these models, it is observed that the decision tree model has achieved the highest accuracy. Therefore, the initial regression model selected by the present embodiment is a decision tree model, and further modeling research is carried out.

[0101] Furthermore, this embodiment also includes calling the DecisionTreeRegressor class in the Scikit-learn library of Python and adjusting the key hyperparameters of the model:

[0102] max_features (maximum number of features): Increase the randomness of the model and prevent the model from being too dependent on certain features.

[0103] max_depth (maximum depth of decision tree): limits the depth of decision tree to prevent overfitting.

[0104] min_samples_split (the minimum number of samples a split node must contain): used to prevent overfitting and limit the minimum number of samples for each leaf node.

[0105] min_samples_leaf (minimum number of samples required on a leaf node): limits the growth of the number of samples on each leaf node to make the leaf nodes more generalizable.

[0106] The final trained decision tree model can be found in Figure 2 ,Since the decision tree model is relatively large, only the first three rows are shown to illustrate the decision path of the model. This helps to intuitively understand how the model makes decisions under different feature conditions and affects the prediction results of patent damages.

[0107] As a third embodiment of the present invention, a patent litigation compensation prediction device is also provided, which includes at least one processing unit and at least one storage unit, wherein the storage unit stores a computer program, and when the computer program is executed by the processing unit, the processing unit executes the steps of any one of the above-mentioned methods.

[0108] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which implements the steps of the above method when executed by a processor. The computer-readable storage medium may include, but is not limited to, any type of disk, including a floppy disk, an optical disk, a DVD, a CD-ROM, a micro drive, and a magneto-optical disk, a ROM, a RAM, an EPROM, an EEPROM, a DRAM, a VRAM, a flash memory device, a magnetic card or an optical card, a nanosystem (including a molecular memory IC), or any type of medium or device suitable for storing instructions and / or data.

[0109] This invention was funded by the China 21st Century Agenda Management Center Commissioned Project: Research on Key Technology Equipment List and Intellectual Property Rights for Carbon Capture, Utilization and Storage (CCUS) (20170986) and the Ministry of Education Humanities and Social Sciences Fund Youth Fund Project (20YJC630201): Research on Patent Path Lock-in Effect and Breakthrough Countermeasures for Key Core Technologies.

[0110] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0111] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A patent litigation compensation prediction method, comprising extracting numerical indicators, classification indicators and corresponding numerical data and classification data of a patent, inputting them into a patent litigation compensation prediction model, and obtaining the compensation prediction result of the patent in a single patent infringement lawsuit; characterized in that: The patent litigation compensation prediction model is constructed according to the following steps: Obtain numerical indicators and their corresponding numerical data, classification indicators and their corresponding classification data, and patent litigation results of multiple patents; Perform data cleaning and normalization on numerical data and data encoding on categorical data; Using the numerical data after data cleaning, the coded classification data and the patent litigation results, the numerical indicators and classification indicators are preprocessed to obtain the preprocessed numerical indicators and the preprocessed classification indicators; A data set is constructed using the preprocessed numerical indicators and their corresponding numerical data, and the preprocessed classification indicators and their corresponding classification data. The data set is used to train the initial regression model to obtain a patent litigation compensation prediction model.

2. The patent litigation damages prediction method according to claim 1, characterized in that: Data encoding for categorical data specifically includes: Encode categorical data into 1 and 0 according to the data type.

3. The patent litigation damages prediction method according to claim 1, characterized in that: Preprocessing specifically includes: Conduct significance analysis on numerical indicators and classification indicators, and eliminate those that are not significant; Calculate the variance of the numerical data corresponding to each numerical indicator, and eliminate the numerical indicators whose variance is less than the variance threshold; Conduct correlation analysis on numerical indicators and eliminate those that are not relevant.

4. The patent litigation damages prediction method according to claim 3, characterized in that: The significance analysis included: Calculate the average of all patent litigation results to obtain the patent litigation result benchmark; take the data in the patent litigation results that are greater than the patent litigation result benchmark as "high compensation in patent litigation", and take the data that are less than or equal to the patent litigation result benchmark as "low compensation in patent litigation", take the data in the patent litigation results that are greater than the patent litigation result benchmark as the sample group, and take the data in the patent litigation results that are less than the patent litigation result benchmark as the control group, and let the original hypothesis be "there is no significant difference between high compensation in patent litigation and the corresponding numerical indicator / classification indicator", and the alternative hypothesis be "there is a significant difference between high compensation in patent litigation and the corresponding numerical indicator / classification indicator".

5. The patent litigation damages prediction method according to claim 4, characterized in that: The significance analysis of numerical indicators also includes: The Mann-Whitney U test was used to calculate the p-value, and the numerical indicators with p-values ​​greater than the first significance threshold were eliminated.

6. The patent litigation damages prediction method according to claim 4, characterized in that: The significance analysis of classification indicators also includes: Fisher's exact probability method was used to calculate the p-value, and the classification indicators with p-values ​​greater than the first significance threshold were eliminated.

7. The patent litigation damages prediction method according to claim 3, characterized in that: The correlation analysis of numerical indicators includes: The Spearman rank correlation coefficient between the numerical indicator and the patent litigation results and its corresponding significance p-value are calculated, and the numerical indicators whose significance p-value of the Spearman rank correlation coefficient is greater than the second significance threshold are eliminated.

8. The method for predicting damages in patent litigation according to claim 1, characterized in that: The initial regression model is one of a linear regression model, a decision tree model, a gradient boosting tree model and a random forest model.

9. A patent litigation damages prediction device, characterized in that: The method comprises at least one processing unit and at least one storage unit, wherein the storage unit stores a computer program, and when the computer program is executed by the processing unit, the processing unit executes the steps of the method according to any one of claims 1 to 8.

10. A storage medium, characterized in that: It stores a computer program executable by an access authentication device. When the computer program runs on the access authentication device, the access authentication device executes the steps of the method according to any one of claims 1 to 8.