Hatching achievement value evaluation method of intangible asset quantification algorithm

Abstract

The invention relates to the technical field of intangible asset valuation, and discloses an intangible asset quantification algorithm incubation achievement value evaluation method comprising the following steps: S1, collecting and preprocessing multi-source data related to intangible asset quantification algorithm achievements to form a multi-dimensional feature set; s2, constructing a dynamic adaptive evaluation framework, updating value evaluation parameters based on a real-time data stream and a reinforcement learning technology, and generating value prediction dynamically adjusted along with time; s3, calculating a fusion value through a multi-dimensional value fusion model; s4, performing uncertainty quantification on the fusion value, generating value distribution, and calibrating a final value according to an uncertainty factor; and S5, outputting the final value. According to the method, the dynamic self-adaptive evaluation framework is constructed, the value evaluation parameters are updated by using the real-time data flow and reinforcement learning technology, and the value prediction adjusted along with time can be generated according to market price fluctuation, technical index updating or legal protection state change.

Description

Technical Field

[0001] The present invention relates to the technical field of intangible asset valuation, and particularly to a method for evaluating the value of the incubation results of intangible asset quantification algorithms. Background Art

[0002] The importance of intangible assets (such as intellectual property, algorithm models, brand value) in the modern economy is increasing day by day, and their value evaluation has become a key link in corporate decision-making, patent transactions, and legal proceedings. In the prior art, the commonly used methods for evaluating the value of intangible assets include the cost method, the market method, and the income method. Among them, the cost method calculates the value based on the replacement or development cost of the asset, the market method conducts valuation by comparing the transaction prices of similar assets, and the income method determines the value by predicting future cash flows and discounting them to the present value. In addition, in recent years, machine learning technology has been introduced into the field of intangible asset valuation. For example, decision trees, support vector machines, or neural network models are used to predict the market value of patents or algorithm results through training with historical data. These methods have been widely used in accounting, patent management, and corporate asset evaluation.

[0003] However, there are several limitations in the prior art, which affect its applicability in evaluating the value of intangible asset quantification algorithm results. First, traditional methods such as the cost method, the market method, and the income method are mostly based on static data and cannot timely reflect the impact of market fluctuations, technological progress, or changes in the legal environment on value. For example, it is difficult to adjust the valuation when the patent technology rapidly depreciates due to the breakthrough of competitors. Second, although existing machine learning models can improve the prediction accuracy, they mostly focus on a single dimension (such as economic benefits) and ignore multi-dimensional factors such as technological innovation degree or legal protection intensity, resulting in insufficient comprehensiveness of the evaluation results. In addition, these methods usually only output fixed values and lack the quantification and analysis of the uncertainty of the prediction results, which limits the credibility and transparency of the evaluation, especially in scenarios that require high-reliability support (such as investment decisions or court evidence). Summary of the Invention

[0004] To make up for the above deficiencies, the present invention provides a method for evaluating the value of the incubation results of intangible asset quantification algorithms, aiming to improve the problem that traditional methods such as the cost method, the market method, and the income method are mostly based on static data and cannot timely reflect the impact of market fluctuations, technological progress, or changes in the legal environment on value.

[0005] In a first aspect, the present invention provides the following technical solution. A method for evaluating the value of the incubation results of intangible asset quantification algorithms includes the following steps: S1. Collect multi-source data related to the results of intangible asset quantification algorithms and perform preprocessing to form a multi-dimensional feature set, where the multi-source data includes economic data, technical data, and legal data; S2. Construct a dynamic adaptive evaluation framework, update the value evaluation parameters based on real-time data streams and reinforcement learning techniques, and generate a value prediction that dynamically adjusts over time. S3. Through a multi-dimensional value fusion model, comprehensively consider the economic value, technical value, and legal value of the multi-source data, and calculate the fusion value. S4. Quantify the uncertainty of the fusion value, generate a value distribution, and calibrate the final value according to the uncertainty factor. S5. Output the final value and its uncertainty distribution as the evaluation result.

[0006] Preferably, the economic data includes market transaction records and expected revenue streams, the technical data includes algorithm innovation degree and technology leadership indicators, the legal data includes patent protection period and litigation risks, and the preprocessing includes data cleaning, normalization, and feature extraction.

[0007] Preferably, the economic data includes market transaction records and expected revenue streams, the technical data includes algorithm innovation degree and technology leadership indicators, the legal data includes patent protection period and litigation risks, and the preprocessing includes data cleaning, normalization, and feature extraction:

[0008] Where s is the current state, a is the evaluation action, r is the immediate reward, α is the learning rate, and γ is the discount factor.

[0009] Preferably, in step S3, the multi-dimensional value fusion model calculates the fusion value through the following formula:

[0010] Where E is the economic value, T is the technical value, L is the legal value, and w1, w2, w3 are dynamic weights that are adjusted in real time according to the data importance by the entropy method.

[0011] Preferably, in step S4, the uncertainty quantification generates a value distribution through Monte Carlo simulation and calculates the uncertainty factor U:

[0012] Where σ is the standard deviation of the value distribution and μ is the mean; the final value calibration formula is:

[0013] Where k is the calibration coefficient, and the range is from 0 to 1.

[0014] Preferably, in step S5, the evaluation result includes the final value V final and its 95% confidence interval, and is output in a visual manner.

[0015] Preferably, it further includes updating the parameters of the dynamic adaptive evaluation framework and the multi-dimensional value fusion model through online learning technology according to the feedback of actual applications to optimize the evaluation accuracy.

[0016] In a second aspect, the present invention provides the following technical solution, an evaluation system for the value of the incubation results of an intangible asset quantification algorithm, including: A data input module for performing step S1 of multi-source data collection and preprocessing; A dynamic evaluation module for performing step S2 of dynamic adaptive evaluation; A fusion calculation module for performing step S3 of multi-dimensional value fusion; An uncertainty analysis module for performing step S4 of uncertainty quantification and value calibration; An output module for performing step S5 of outputting the evaluation result.

[0017] In a third aspect, the present invention provides the following technical solution, a computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein when the processor executes the computer program, it implements the above-mentioned method for evaluating the value of the incubation results of the intangible asset quantification algorithm.

[0018] In a fourth aspect, the present invention provides the following technical solution, a readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the above-mentioned method for evaluating the value of the incubation results of the intangible asset quantification algorithm.

[0019] The present invention has the following beneficial effects: 1. In the present invention, by constructing a dynamic adaptive evaluation framework and using real-time data streams and reinforcement learning techniques (such as Q-learning algorithm) to update the value evaluation parameters, it is possible to generate a value prediction that adjusts over time according to market price fluctuations, technical indicator updates, or changes in legal protection status; compared with the traditional static valuation method that only calculates based on fixed-point data, this method introduces a continuous learning mechanism in step S2, enabling the evaluation result to reflect changes in the external environment, thereby improving the applicability of the evaluation result in a dynamic environment.

[0020] 2. In the present invention, in step S3, a multi-dimensional value fusion model is adopted. By quantifying economic value (such as income stream based on the net present value method), technical value (such as patent citation rate and technical complexity score), and legal value (such as protection period and litigation risk weighting), and using dynamic weights (such as calculated by the entropy method) for fusion, a comprehensive value result is generated. By integrating multi-source data features, the one-sidedness of the evaluation result is avoided, enabling the value evaluation to simultaneously reflect the contributions of intangible assets at the commercial, technical, and legal levels.

[0021] 3. In the present invention, in step S4, the uncertainty output by the Monte Carlo simulation quantization algorithm is used to generate a value distribution and calculate a confidence interval (such as a 95% confidence level), and the final value is calibrated according to the uncertainty factor. Not only the specific value is output, but also the statistical characteristics of the uncertainty distribution (such as the mean and standard deviation) are provided, and presented in a visual form in step S5, making the reliability of the evaluation process and results easier to verify. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 It is a method flowchart of the evaluation method for the incubation achievement value of the intangible asset quantization algorithm proposed by the present invention; Figure 2 It is a system framework diagram of the evaluation system for the incubation achievement value of the intangible asset quantization algorithm proposed by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0023] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0024] Embodiment 1 Refer to Figure 1 , in the first embodiment of the present invention, the present invention provides an evaluation method for the incubation achievement value of an intangible asset quantization algorithm, including the following steps: S1. Collect multi-source data related to the intangible asset quantization algorithm achievement and perform preprocessing to form a multi-dimensional feature set, where the multi-source data includes economic data, technical data, and legal data; S2. Construct a dynamic adaptive evaluation framework, update the value evaluation parameters based on real-time data streams and reinforcement learning techniques, and generate a value prediction that dynamically adjusts over time; S3. Through a multi-dimensional value fusion model, comprehensively calculate the economic value, technical value, and legal value of the multi-source data to calculate the fusion value; S4. Quantify the uncertainty of the fusion value, generate a value distribution, and calibrate the final value according to the uncertainty factor; S5. Output the final value and its uncertainty distribution as the evaluation result.

[0025] Specifically, S1: The collection of multi-source data is achieved through various methods such as web crawlers, database interfaces, and manual entry. Among them, economic data includes historical transaction records, current market prices, future earnings forecasts (such as annualized returns obtained based on time series analysis), industry average return rates, etc.; technical data includes the originality score of algorithms (based on citation frequencies in patent databases and reviews by technical field experts), technical complexity (such as the computational complexity O(n) or space complexity of algorithms), technical leadership indicators (such as performance comparison data with similar technologies, for example, the percentage increase in accuracy); legal data includes patent registration status, protection period (accurate to the year, month, and day), number of license agreements, potential litigation risk assessment (such as probability statistics based on historical litigation cases), geographical coverage of intellectual property rights (such as China, the United States, the European Union, etc.); the preprocessing process specifically includes removing noise points in the data (such as outlier removal using the 3σ criterion), filling in missing values (using mean imputation or KNN interpolation method), and normalizing data with different dimensions (such as Min-Max normalization or Z-score normalization), and finally forming a multi-dimensional feature vector containing at least three types of features (economic, technical, legal) for subsequent evaluation.

[0026] S2: The construction of the dynamic adaptive evaluation framework is based on the input of real-time data streams. For example, the market price is updated hourly through an API interface, or the technical indicators are updated quarterly through a technical monitoring system; the reinforcement learning technology uses the Q-learning algorithm. When initializing the Q-table, the initial state and action space are set according to historical data. Among them, the state s includes the current market conditions, technical maturity, and legal protection strength, and the action a includes increasing or decreasing the evaluation weight of a certain dimension value; the introduction of real-time data streams is achieved by subscribing to financial market data sources (such as Bloomberg or Reuters) and updating technical patent databases (such as GooglePatents); during the update process, the learning rate α can be set in the range of 0.1 to 0.5 and decreased with the number of training rounds, and the discount factor γ is set according to the expected life cycle of the asset (such as taking 0.8 for short-term assets and 0.95 for long-term assets); the dynamically adjusted value prediction V′(t) is indexed by the timestamp, and a prediction value is generated every minute or hour, which is applicable to scenarios of high-frequency trading or rapid technological iteration.

[0027] S3: The specific implementation of the multi-dimensional value fusion model includes first calculating the economic value E using the net present value method (NPV), and the formula is (where CF t$CF_t$ is the cash flow in the $t$-th period, and $r$ is the discount rate (determined based on the industry average cost of capital); the technology value $T$ is quantified by constructing a technology scoring system, for example, based on the number of patent citations (1 point added for each citation), the difficulty of technology implementation (scored by experts from 1 to 10), and the scarcity of technology (the reciprocal of the number of similar technologies in the market); the legal value $L$ adopts the weighted scoring method, and the weight factors include the proportion of the protection period (remaining years / total years) and the litigation risk coefficient (based on the historical winning rate, such as in the range of 0 - 1); during the fusion, the dynamic weights $w_1$, $w_2$, $w_3$ are calculated by the entropy method. The specific steps are to standardize the data of each dimension, calculate the entropy value ($k$ is a constant, $p$ ij is the proportion of the $i$-th sample in the $j$-th dimension), and determine the weights ; the fusion value $V$ fusion is taken as the weighted sum and updated daily to reflect the latest data.

[0028] S4. The quantification of uncertainty is achieved through Monte Carlo simulation. The specific operation is to set a random distribution (such as normal distribution or uniform distribution) for the input parameters (such as the income stream and technology scores), run at least 10,000 simulations to generate a value distribution curve, and calculate the mean $\mu$ and standard deviation $\sigma$; the uncertainty factor $U$ represents the relative volatility. If $U > 0.3$, it is considered that the uncertainty is relatively high; during the calibration process, the calibration coefficient $k$ can be adjusted according to the application scenario. For example, in a risk-sensitive scenario (such as financial investment), it is set to 0.8, and in a general scenario, it is set to 0.2; the final value $V$ final is calculated considering the conservative correction of the value by uncertainty to ensure that the result is not overly optimistic; in addition, sensitivity analysis can be introduced to evaluate the independent impact of each input parameter on the final value (such as the value fluctuation when the income stream changes by ±10%).

[0029] S5. The output of the final value includes the numerical result $V$ final in the unit of RMB or other currencies, accurate to two decimal places), the statistical characteristics of the uncertainty distribution (such as mean, variance, skewness, kurtosis), the confidence interval (default 95%, calculated by assuming a normal distribution for the upper and lower limits, such as ($\mu\pm1.96\sigma$)); the visualization methods include a line chart showing the change of value over time, a bar chart showing the contribution ratio of each dimension, and a probability density chart showing the uncertainty distribution; the output format supports PDF reports, Excel tables, and interactive web interfaces. Users can choose to view detailed data or a brief summary; in addition, the system retains historical evaluation records for subsequent comparative analysis.

[0030] Economic data includes market transaction records and expected income streams, technical data includes algorithm innovation and technology leadership indicators, legal data includes patent protection periods and litigation risks, and preprocessing includes data cleaning, normalization, and feature extraction.

[0031] Specifically, market transaction records are derived from public market data (such as the buying and selling prices recorded on patent trading platforms) or internal corporate financial statements. Expected revenue streams are generated through time series forecasting models (such as ARIMA or LSTM), and the forecast period can be 1 to 10 years. The algorithm innovation is determined by counting the number of citations in the patent database (the percentage of citations in the past 5 years) and comparative analysis of similar technologies (such as the performance gap with existing algorithms, and the quantitative improvement rate, such as 10%-50%). Technology leadership indicators include algorithm operation efficiency (such as the amount of data processed per second) and the breadth of application scenario coverage (the number of industries supported, such as finance, medical care, etc.). The patent protection period is accurate. By a specific date (e.g. 20 years from March 18, 2025), litigation risk is analyzed through historical case database analysis (e.g. litigation success rate in the same industry in the past 10 years, assigned a value of 0-1 in combination with expert opinions); data cleaning operations include removing duplicate records, deleting fields with a missing rate of more than 30%, and filling the remaining missing values ​​with the mean or median; normalization uses the Min-Max method to map the data to the [0,1] interval, or the Z-score method to ensure zero mean and unit variance; feature extraction includes principal component analysis (PCA) to retain features with 90% variance, or correlation analysis to remove redundant variables, and finally generating a feature vector of at least 10 dimensions.

[0032] Economic data includes market transaction records and expected revenue flows, technical data includes algorithm innovation and technology leadership indicators, legal data includes patent protection period and litigation risks, and preprocessing includes data cleaning, normalization, and feature extraction:

[0033] Among them, s is the current state, a is the evaluation action, r is the immediate return, α is the learning rate, and γ is the discount factor.

[0034] Specifically, the implementation of the Q-learning algorithm includes constructing a state space (which includes at least three dimensions: market state, technical state, and legal state. Each dimension is divided into three levels: high, medium, and low, with a total of 27 combinations), and an action space (including increasing or decreasing the evaluation weights of economic, technical, and legal values, with a step size of 0.05 - 0.2); the initial Q-table is preset through expert experience or historical data. For example, the economic weight is set to 0.5 in the high market state; the immediate reward r is calculated based on actual feedback, such as the difference between the predicted value and the market transaction price (a positive value represents a positive reward, and a negative value represents a negative reward); the learning rate α is initialized to 0.3 and decays exponentially with the number of training rounds (e.g., halving every 100 rounds), and the discount factor γ is dynamically adjusted according to the asset type (e.g., taking 0.85 for software algorithms and 0.9 for hardware patents); the real-time data stream is updated every 5 minutes through the WebSocket interface or updated daily through batch processing; after each update, a new value prediction V′(t) is generated and compared with the previous prediction. If the change exceeds 5%, an alarm mechanism is triggered to notify the user; in addition, an exploration rate ϵ can be introduced (initially 0.1, decreasing to 0.01 over time), and the ϵ-greedy strategy is adopted to balance exploration and exploitation to ensure the adaptability of the algorithm.

[0035] In step S3, the multi-dimensional value fusion model calculates the fusion value through the following formula:

[0036] where E is the economic value, T is the technical value, L is the legal value, and w1, w2, w3 are dynamic weights, which are adjusted in real time according to the importance of data by the entropy method.

[0037] Specifically, the calculation of the economic value E is based on the net present value method, and the cash flow CF t is estimated through market research and historical revenue data. The discount rate r is determined by combining the risk-free rate (such as the 3% yield of national bonds) and the industry risk premium (such as adding 5% for the technology industry). The calculation period can be 5 years, 10 years, or the asset life cycle; the technical value T is calculated through multi-index weighted scoring. For example, the proportion of patent citation times accounts for 40% (each citation adds 0.5 points, with a maximum of 50 points), the proportion of technical difficulty accounts for 30% (expert scoring from 1 to 10), and the proportion of market scarcity accounts for 30% (the reciprocal of the number of similar technologies, ranging from 0 to 20). The total score is normalized to [0, 1000]; the legal value L is calculated through the protection period score (each remaining year adds 10 points, with a maximum of 200 points), the litigation risk deduction (risk coefficient from 0 to 1, each 0.1 deducts 20 points), and the license income addition (each license contract adds 50 points); the implementation of the entropy method for the dynamic weights w1, w2, w3 includes data standardization (linear proportional transformation), entropy calculation (considering the distribution of at least 100 samples), and weight normalization (ensuring w1 + w2 + w3 = 1). The weight update frequency is daily or weekly; the fusion value V fusionThe results are in ten thousand yuan, rounded to two decimal places, and record the contribution ratios of each dimension (such as 60% for economy, 25% for technology, and 15% for law), which is convenient for analyzing the value sources.

[0038] In step S4, uncertainty quantification generates a value distribution through Monte Carlo simulation and calculates the uncertainty factor U:

[0039] where σ is the standard deviation of the value distribution and μ is the mean; the final value calibration formula is:

[0040] where k is the calibration coefficient, and the range is from 0 to 1.

[0041] Specifically, the specific operation of Monte Carlo simulation is to set a random distribution for the input parameters. For example, the revenue stream follows a normal distribution (the mean is the predicted value, and the standard deviation is 1.5 times the historical volatility), the technology score follows a uniform distribution (the range is 10% above and below the expert score), and the protection period follows a Poisson distribution (based on historical patent survival data); each simulation generates 10,000 - 50,000 samples, and calculates the statistical characteristics of the value distribution, including the mean μ (as the benchmark value), the standard deviation σ (reflecting volatility), skewness (evaluating the symmetry of the distribution), and kurtosis (evaluating the possibility of extreme values); the calculation result range of the uncertainty factor U is from 0 to 1. If U < 0.1, it indicates high certainty, and if U > 0.5, it means that the input data needs to be further verified; the calibration coefficient k is set according to the user's risk preference. For example, a conservative user sets it to 0.9 (significantly reducing the value), and an aggressive user sets it to 0.1 (slightly adjusting); the final value V final The calibration process records the intermediate results of each simulation and generates an uncertainty report, including the probability density curve and the cumulative distribution function (CDF); in addition, the Bootstrap method can be introduced to verify the stability of the Monte Carlo results and ensure that the error is less than 5%.

[0042] In step S5, the evaluation results include the final value V final and its 95% confidence interval, and are output in a visual way.

[0043] Specifically, the final value V finalOutput in specific currency units (e.g., ten thousand yuan in RMB, accurate to two decimal places, e.g., 1050.25 ten thousand yuan). The 95% confidence interval is calculated through the normal distribution hypothesis (μ ± 1.96σ), or directly take the 2.5% and 97.5% quantiles through Monte Carlo simulation (e.g., [1000.50, 1100.75] ten thousand yuan); the output of the uncertainty distribution includes a probability density plot (the horizontal axis is value, the vertical axis is probability density, and a smooth curve is used with kernel density estimation), a box plot (showing the median, interquartile range, and outliers), and a line plot (showing the trend of value changing over time, with a time granularity of day, week, or month); visualization tools support Matplotlib, Plotly, or Tableau, and users can interactively adjust the view (e.g., zoom in on a certain time period, switch the dimensional perspective); the output report formats include PDF (containing charts and text explanations), CSV (containing raw data and statistical results), JSON (supporting API calls); in addition, the system provides a historical comparison function, and users can view the changes in the evaluation results in the past 30 days or 1 year, and it supports exporting to an Excel table for easy financial analysis or legal filing.

[0044] It also includes updating the parameters of the dynamic adaptive evaluation framework and the multi-dimensional value fusion model through online learning technology according to the actual application feedback to optimize the evaluation accuracy.

[0045] Specifically, the actual application feedback includes market transaction verification data (such as the difference between the final transaction price and the predicted value, in ten thousand yuan), user subjective evaluation (such as a satisfaction score of 1 - 5), and third-party audit results (such as the valuation report of an accounting firm); the online learning technology adopts an incremental update strategy. For example, each time feedback is received, the Q-table value of Q-learning is adjusted using the gradient descent method (the step size is 0.01 - 0.05), or the weights w1, w2, w3 of the fusion model are updated (recalculated based on the entropy value of the feedback data); the update frequency can be set to daily, weekly, or trigger-based (automatically updated when the feedback error exceeds 10%); the specific indicators for optimizing the evaluation accuracy include the mean square error (MSE, required to be less than 5%), the prediction coverage rate (the proportion of the confidence interval covering the actual value, required to be greater than 90%), and the model convergence time (less than 10 seconds after each update); in addition, the system records the parameter change log for each update (such as the learning rate decreasing from 0.3 to 0.25), and supports a rollback function. If the accuracy decreases after the update, it can be restored to the previous version parameters; users can view the real-time status of the optimization process (such as the current MSE value, convergence curve) through the interface.

[0046] Embodiment 2: Refer to Figure 2 , in the second embodiment of the present invention, the present invention provides an intangible asset quantification algorithm-based incubation achievement value evaluation system, including: A data input module for performing step S1: multi-source data collection and preprocessing; A dynamic evaluation module for performing step S2: dynamic adaptive evaluation; A fusion calculation module for performing step S3: multi-dimensional value fusion; An uncertainty analysis module for performing step S4: uncertainty quantification and value calibration; An output module for performing step S5: evaluation result output.

[0047] Specifically, the data input module supports multiple data formats (such as CSV, JSON, SQL databases), collects data in parallel through multi-threaded technology (each thread processes one data source, such as market data, technical data), and is equipped with a data verification function (checking data integrity, passing if the missing rate is less than 10%); the dynamic evaluation module runs in a GPU-accelerated environment (such as Q-learning calculation supported by NVIDIA CUDA), processes at least 1000 state updates per second, and the memory occupancy does not exceed 2GB; the fusion calculation module incorporates the entropy method algorithm, supports value fusion of at least 10 dimensions, the calculation time is less than 1 second, and the results are stored in local cache (Redis or SQLite); the uncertainty analysis module integrates a Monte Carlo simulation library (such as NumPy or SciPy in Python), supports customizing the number of simulations (1000 - 100000 times, adjustable by the user), and generates detailed logs (recording the seed values and result distributions of each simulation); the output module supports multi-terminal display (PC, mobile phone, tablet), provides API interfaces (RESTful or GraphQL) for external systems to call, and the output data is encrypted and stored (using the AES-256 algorithm) to ensure the security of business-sensitive information; the overall system is deployed on a cloud server (such as Alibaba Cloud ECS), supports high-concurrency access (processing 100 user requests per second), and is equipped with a fault tolerance mechanism (such as automatically restarting after a module crashes).

[0048] Embodiment Three In the third embodiment of the present invention, based on the same inventive concept, a computer-readable storage medium is proposed. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, it implements the steps of the method for evaluating the value of the incubation results of the intangible asset quantification algorithm in the above embodiment.

[0049] Embodiment Four In the fourth embodiment of the present invention, based on the same inventive concept, a computer device is proposed. The terminal includes: a processor and a memory; the processor and the memory communicate with each other; the memory is used to store instructions; the processor is used to execute the instructions in the memory to execute the method for evaluating the value of the incubation results of the intangible asset quantification algorithm in the above embodiment.

[0050] It should be understood that each part of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logic functions on data signals, application specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.

[0051] Finally, it should be noted that the above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. The incubation results value assessment method of the intangible asset quantification algorithm is characterized by: The following steps are involved: S1. Collect and pre-process multi-source data related to the results of the intangible asset quantification algorithm to form a multi-dimensional feature set, wherein the multi-source data includes economic data, technical data and legal data; S2. Build a dynamic adaptive evaluation framework to update value evaluation parameters based on real-time data streams and reinforcement learning techniques to generate value predictions that are dynamically adjusted over time; S3. Calculate the fusion value by integrating the economic value, technical value and legal value of the multi-source data through a multi-dimensional value fusion model; S4, quantifying the uncertainty of the fusion value, generating a value distribution, and calibrating the final value according to the uncertainty factor; S5. Output the final value and its uncertainty distribution as the evaluation result.

2. The method for evaluating the value of incubation results of intangible asset quantification algorithm according to claim 1 is characterized in that: The economic data includes market transaction records and expected profit flows, the technical data includes algorithm innovation and technology leadership indicators, the legal data includes patent protection period and litigation risks, and the preprocessing includes data cleaning, normalization and feature extraction.

3. The method for evaluating the value of incubation results of intangible asset quantification algorithm according to claim 1 is characterized in that: The economic data includes market transaction records and expected revenue streams, the technical data includes algorithm innovation and technical leadership indicators, the legal data includes patent protection period and litigation risks, and the preprocessing includes data cleaning, normalization and feature extraction: ; Where s is the current state, a is the evaluation action, r is the immediate return, α is the learning rate, and γ is the discount factor.

4. The method for evaluating the value of incubation results of intangible asset quantification algorithm according to claim 1, characterized in that: In step S3, the multi-dimensional value fusion model calculates the fusion value by the following formula: ; Among them, E is economic value, T is technical value, L is legal value, w1, w2, w3 are dynamic weights, which are adjusted in real time according to the importance of data through the entropy method.

5. The method for evaluating the value of incubation results of intangible asset quantification algorithm according to claim 1, characterized in that: In step S4, the uncertainty quantification generates a value distribution through Monte Carlo simulation and calculates the uncertainty factor U: ; Where σ is the standard deviation of the value distribution and μ is the mean; the final value calibration formula is: ; where k is the calibration coefficient, ranging from 0 to 1.

6. The method for evaluating the value of incubation results of intangible asset quantification algorithm according to claim 1 is characterized in that: In step S5, the evaluation result includes a final value V final and its 95% confidence interval, and output them visually.

7. The method for evaluating the value of incubation results of intangible asset quantification algorithm according to claim 1 is characterized in that: It also includes updating the parameters of the dynamic adaptive evaluation framework and the multi-dimensional value fusion model through online learning technology based on actual application feedback to optimize the evaluation accuracy.

8. The incubation achievement value assessment system of the intangible asset quantification algorithm is characterized by: The method for evaluating the value of incubation results of the intangible asset quantification algorithm according to any one of claims 1 to 7 comprises: A data input module, used to perform multi-source data collection and preprocessing in step S1; A dynamic evaluation module, used to perform the dynamic adaptive evaluation in step S2; A fusion calculation module, used to perform multi-dimensional value fusion in step S3; Uncertainty analysis module, used to perform uncertainty quantification and value calibration in step S4; The output module is used to execute step S5 to output the evaluation results.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the incubation results value assessment method of the intangible asset quantification algorithm according to any one of claims 1 to 7 is implemented.

10. A readable storage medium, characterized in that: The readable storage medium stores a computer program, and when the computer program is executed by a processor, the incubation results value assessment method of the intangible asset quantification algorithm according to any one of claims 1 to 7 is implemented.

Images