Scientific and technological achievement value prediction method and device, program product and electronic equipment

Abstract

The invention relates to the technical field of computers, and provides a scientific and technological achievement value prediction method and device, a program product and electronic equipment. The method comprises the following steps: selecting a target achievement value feature of a to-be-predicted scientific and technological achievement from full achievement value features according to a to-be-predicted scene and an achievement type of the to-be-predicted scientific and technological achievement; obtaining a first target input feature value based on the achievement type coding feature value of the to-be-predicted scientific and technological achievement, the scene coding feature value of the to-be-predicted scene and the first feature value of the target achievement value feature of the to-be-predicted scientific and technological achievement; and inputting the first target input feature value into a pre-trained scientific and technological achievement value prediction model, and determining the achievement value of the scientific and technological achievement to be predicted in the scene to be predicted according to the output of the scientific and technological achievement value prediction model. According to the technical scheme, the value prediction accuracy of scientific and technological achievements in different scenes can be improved.

Description

Technical Field

[0001] This disclosure relates to the field of computer technology, and more specifically, to a method for predicting the value of scientific and technological achievements, a device for predicting the value of scientific and technological achievements, a computer program product, and an electronic device. Background Technology

[0002] Predicting the value of scientific and technological achievements can provide accurate scientific and quantitative basis for their transformation and management. For example, it can provide quantitative standards for performance evaluation and equity incentives, promote a virtuous cycle of scientific research and innovation, and accelerate the process of bringing scientific and technological achievements from the laboratory to the market.

[0003] The prediction of the value of scientific and technological achievements in related technologies does not distinguish between prediction scenarios and achievement types. It uses the same characteristics and uniform evaluation standards to adapt to all needs, resulting in low accuracy and reliability in predicting the scientific and technological value of scientific and technological achievements.

[0004] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this disclosure is to provide a method and apparatus for predicting the value of scientific and technological achievements, a computer program product and an electronic device, thereby improving the accuracy of value prediction for different types of scientific and technological achievements in different scenarios to at least a certain extent.

[0006] Other features and advantages of this disclosure will become apparent from the following detailed description, or may be learned in part from practice of this disclosure.

[0007] According to a first aspect of this disclosure, a method for predicting the value of scientific and technological achievements is provided, comprising: selecting target achievement value features of the scientific and technological achievement to be predicted from all achievement value features based on the scenario to be predicted and the achievement type of the scientific and technological achievement to be predicted; obtaining a first target input feature value based on the achievement type encoding feature value of the scientific and technological achievement to be predicted, the scenario encoding feature value of the scenario to be predicted, and a first feature value of the target achievement value feature of the scientific and technological achievement to be predicted; inputting the first target input feature value into a pre-trained scientific and technological achievement value prediction model, and determining the achievement value of the scientific and technological achievement to be predicted in the scenario to be predicted based on the output of the scientific and technological achievement value prediction model.

[0008] According to a second aspect of this disclosure, a device for predicting the value of scientific and technological achievements is provided, comprising: a feature selection module configured to select a target achievement value feature of the scientific and technological achievement to be predicted from a full set of achievement value features based on a scenario to be predicted and the achievement type of the scientific and technological achievement to be predicted; an input feature value determination module configured to obtain a first target input feature value based on an achievement type encoding feature value of the scientific and technological achievement to be predicted, a scenario encoding feature value of the scenario to be predicted, and a first feature value of the target achievement value feature of the scientific and technological achievement to be predicted; and a value prediction module configured to input the first target input feature value into a pre-trained model for predicting the value of scientific and technological achievements, and determine the achievement value of the scientific and technological achievement to be predicted in the scenario to be predicted based on the output of the model for predicting the value of the scientific and technological achievement to be predicted.

[0009] According to a third aspect of this disclosure, a computer program product comprising instructions is provided, which, when run on a computer, causes the computer to perform the steps of the method for predicting the value of scientific and technological achievements as described in the first aspect.

[0010] According to a fourth aspect of this disclosure, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the method for predicting the value of scientific and technological achievements as described in the first aspect of the above embodiments.

[0011] According to a fifth aspect of this disclosure, an electronic device is provided, comprising: a processor; and a storage device for storing one or more programs, which, when executed by the one or more processors, cause the one or more processors to implement the scientific and technological achievement value prediction method as described in the first aspect of the above embodiments.

[0012] As can be seen from the above technical solutions, the method and apparatus for predicting the value of scientific and technological achievements in the exemplary embodiments of this disclosure, as well as the computer program product and electronic device for implementing the method for predicting the value of scientific and technological achievements, have at least the following advantages and positive effects: In the technical solutions provided by some embodiments of this disclosure, on the one hand, the target value characteristics of the scientific and technological achievement to be predicted are selected according to the scenario to be predicted and the type of achievement to be predicted. Based on the adaptation characteristics of each achievement to be predicted and the prediction scenario, accurate prediction of different types of scientific and technological achievements in different prediction scenarios can be achieved, thereby improving the accuracy of scientific and technological achievement value prediction. On the other hand, this disclosure can achieve accurate value prediction of different types of scientific and technological achievements in different prediction scenarios by deploying only one model, thereby reducing the cost of model training, iteration and maintenance.

[0013] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0014] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure. It is obvious that the drawings described below are merely some embodiments of this disclosure, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.

[0015] Figure 1 A flowchart illustrating a method for predicting the value of scientific and technological achievements according to an exemplary embodiment of this disclosure is shown. Figure 2 A graphical user interface diagram illustrating the data entry and collection of application information of a scientific and technological achievement according to an exemplary embodiment of this disclosure is shown. Figure 3 A graphical user interface diagram illustrating the information filling and collection of another scientific and technological achievement in an exemplary embodiment of this disclosure is shown. Figure 4 A flowchart illustrating a method for determining a first preset mapping relationship in an exemplary embodiment of this disclosure is shown. Figure 5 This diagram illustrates a flowchart of a method for determining a pre-trained scientific and technological achievement value prediction model according to an exemplary embodiment of this disclosure. Figure 6 This diagram illustrates a flowchart of a method for determining the value of a scientific and technological achievement to be predicted in a scenario to be predicted, according to an exemplary embodiment of this disclosure. Figure 7 This diagram illustrates the composition of a technology achievement value prediction device according to an exemplary embodiment of the present disclosure. Figure 8 A schematic diagram of the structure of an electronic device in an exemplary embodiment of this disclosure is shown. Detailed Implementation

[0016] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided to make this disclosure more comprehensive and complete, and to fully convey the concept of the example embodiments to those skilled in the art. The described features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. In the following description, numerous specific details are provided to give a full understanding of embodiments of this disclosure. However, those skilled in the art will recognize that the technical solutions of this disclosure can be practiced with one or more of the specific details omitted, or other methods, components, apparatus, steps, etc., can be employed. In other instances, well-known technical solutions are not shown or described in detail to avoid obscuring various aspects of this disclosure.

[0017] The terms “a,” “an,” “the,” and “the” are used in this specification to indicate the presence of one or more elements / components / etc.; the terms “including” and “having” are used to indicate an open-ended inclusion and to mean that there may be other elements / components / etc. in addition to the listed elements / components / etc.; the terms “first” and “second” are used only as markings and are not a limitation on the number of objects.

[0018] Furthermore, the accompanying drawings are merely illustrative of this disclosure and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0019] In related technologies, the same characteristics and unified evaluation standards are used to evaluate the value of scientific and technological achievements for different prediction scenarios and different types of scientific and technological achievements, which leads to inaccurate value prediction results for some scientific and technological achievements in certain scenarios.

[0020] Figure 1 This diagram illustrates a flowchart of a method for predicting the value of scientific and technological achievements according to an exemplary embodiment of this disclosure. (Reference) Figure 1 The method includes: Step S110: Based on the scenario to be predicted and the type of the scientific and technological achievement to be predicted, select the target achievement value feature of the scientific and technological achievement to be predicted from the value features of all achievements. Step S120: Based on the result type coding feature value of the scientific and technological achievement to be predicted, the scene coding feature value of the scenario to be predicted, and the first feature value of the target result value feature of the scientific and technological achievement to be predicted, the first target input feature value is obtained. Step S130: Input the first target input feature value into the pre-trained scientific and technological achievement value prediction model, and determine the achievement value of the scientific and technological achievement to be predicted in the scenario to be predicted based on the output of the scientific and technological achievement value prediction model.

[0021] exist Figure 1 In the technical solution provided by the embodiments shown, on the one hand, the present disclosure selects the target result value characteristics of the scientific and technological achievement to be predicted based on the scenario to be predicted and the result type of the scientific and technological achievement to be predicted. It can achieve accurate prediction of different types of scientific and technological achievements in different prediction scenarios based on the adaptation characteristics of each result to be predicted and the prediction scenario, thereby improving the accuracy of scientific and technological achievement value prediction. On the other hand, the present disclosure can achieve accurate prediction of the value of different types of scientific and technological achievements in different prediction scenarios by deploying only one model, thereby reducing the cost of model training, iteration and maintenance.

[0022] Next, a detailed explanation will be given of the specific implementation method of "step S110, selecting the target result value characteristics of the scientific and technological achievement to be predicted from the full set of result value characteristics based on the scenario to be predicted and the result type of the scientific and technological achievement to be predicted".

[0023] In one exemplary implementation, the types of results include one or more of the following: papers, patents, process methods, technical standards, published monographs, and software copyrights; the scenarios to be predicted include one or more of the following: scientific research project initiation scenarios, results transaction scenarios, scientific research performance evaluation scenarios, and science and technology innovation financing scenarios.

[0024] For example, different types of technological achievements have different attributes, so the value prediction characteristics of different types of technological achievements are not the same. Furthermore, the differences in business objectives in different scenarios mean that the value characteristics considered for the same type of technological achievement are not the same in different scenarios.

[0025] For example, in the context of scientific research project initiation, the core objective is to select research results with R&D potential that can solve industry pain points, providing a basis for subsequent R&D funding and project initiation. The focus is primarily on the technological innovation, forward-looking nature, and feasibility of the scientific and technological achievements. In the context of successful transactions, the core objective is to achieve successful pricing and promote technology transfer or licensing. The evaluation focuses on market value, rights stability, and the ability to convert these into profits. In the context of scientific research performance evaluation, the core objective is to accurately and objectively quantify and assess the research performance of researchers or research institutions. The evaluation focuses on the level of achievements, academic or industry influence, and standardization. In the context of science and technology innovation financing, the core objective is to attract investment and obtain financing support. The evaluation focuses on the growth potential, market prospects, and industrialization capabilities of these achievements.

[0026] Taking scientific and technological achievements, namely papers and patents, as examples, in the context of scientific research project approval, for papers, the focus is on their technological innovation, their alignment with cutting-edge technologies, their compatibility with research directions and policies, and the feasibility of experimental verification. For patents, the focus is on the scope of protection of their claims, the innovativeness of their technical solutions, and their technological maturity. In the context of technology transfer, for papers, the focus is on the transformability of their technical solutions, the market demand for derivative products based on the paper's technology, and the technical service capabilities of the paper's authors' research team. For patents, the focus is on the stability of the patent rights, the global layout of patent families, historical licensing and / or transfer status, and the remaining protection period.

[0027] In one exemplary implementation, the overall value characteristics of scientific and technological achievements include basic attribute characteristics, application effect characteristics, and value potential characteristics of different types of scientific and technological achievements. The basic attribute characteristics include one or more of the following: technological novelty, technological maturity, technological stability, and achievement level. The application effect characteristics are determined based on one or more of the following: the frequency of application of the scientific and technological achievement in existing scientific and technological projects, application benefits, and the level of the applied scientific and technological projects. The value potential characteristics of the scientific and technological achievements are determined based on one or more of the following: the degree of matching between the scientific and technological achievement and policy, the potential for integration between the scientific and technological achievement and cutting-edge technologies, and the application scenarios to which the scientific and technological achievement is suitable.

[0028] For example, the total value characteristics of achievements refer to the union of the value characteristics of all types of scientific and technological achievements. For instance, if there are two types of scientific and technological achievements, A and B, the value characteristics of achievement A include value characteristics 1, 2, and 3, and the value characteristics of achievement B include 1, 2, 4, 5, and 6. Then the total value characteristics of achievements include 1, 2, 3, 4, 5, and 6.

[0029] Among these, fundamental attributes are the inherent characteristics of scientific and technological achievements. These fundamental attributes vary depending on the type of achievement. For example, for a paper, fundamental attributes include technological novelty and achievement level; for a patent, they may include technological novelty, technological stability, and achievement level; for a process method, they may include technological maturity and achievement level; for a technical standard, they may include technological novelty, technological stability, and achievement level; for a published monograph, they may include technological stability and achievement level; and for software copyright, they may include technological novelty, technological maturity, and technological stability.

[0030] For the same basic attribute characteristics of different types of scientific and technological achievements, the methods for quantifying their values ​​also differ. The method can be determined based on the characteristics of the achievement itself, and this exemplary implementation does not impose any specific limitations on this. For example, for papers, the technical novelty characteristic can be quantified by the number of innovative points, or the achievement level characteristic can be quantified based on the journal's ranking. For patents, the technical novelty characteristic can be quantified by the number of innovative points in the claims, the technical stability characteristic can be quantified by whether the patent has been granted, whether there are invalidation lawsuits, etc., or the achievement level characteristic can be quantified by the patent type (invention, utility model, design), whether there are overseas patent families, whether patent awards have been received, etc. For technological methods, the technological maturity characteristic can be quantified by the scale of application of the method, such as 10 points for large-scale engineering application, 7 points for passing engineering pilot verification, 4 points for laboratory experimental stage, etc., or the recognition level of the technological method can be quantified. For achievements, the level of achievement characteristics can be quantified. For example, the level of achievement characteristics for national-level technological methods is higher than that for provincial-level technological methods. For technical standards, the novelty of technology can be quantified by the proportion of innovative technology clauses in the standard; the stability of technology can be quantified by the extent of modifications made to the standard after its publication; and the level of achievement characteristics can be quantified based on the level to which the standard belongs. For published monographs, the stability of technology can be quantified by whether the monograph has copyright disputes, academic errors, or plagiarism; and the level of achievement characteristics can be determined by the monograph's publication level, for example, the level of achievement characteristics for monographs published by national-level publishers is higher than that for monographs published by ordinary publishers. For software copyrights, the novelty of technology can be quantified by the degree of differentiated functionality of improved algorithms; the maturity of technology can be quantified by commercial scale; and the stability of technology can be quantified by the presence of code vulnerabilities and copyright disputes.

[0031] In this disclosure, the scientific and technological value of scientific and technological achievements can be predicted from three dimensions: the inherent attributes of the achievements, feedback from practical applications, and future value-added potential. By using characteristics from multiple dimensions, the accuracy of value prediction for scientific and technological achievements can be improved.

[0032] For example, it can be done through Figure 2 and Figure 3 The interface shown collects data on the application effects of the scientific and technological achievements to be predicted, and then extracts feature values ​​of the application effect characteristics based on the collected results. For example, relevant personnel can use methods such as... Figure 2The interface shown allows users to fill in information such as the name of the scientific and technological achievement, the name of the project in which the achievement is applied, and the benefits of the application. Each science and technology manager can then use this interface. Figure 3 The interface shown collects the data filled in by the relevant personnel under its responsibility, and then reviews and confirms the collected data to ensure the authenticity and accuracy of the information collected.

[0033] For example, keywords and other information about scientific and technological achievements can be matched with relevant policies, and the degree of matching between the achievements and policies can be determined based on the number of matched keywords. The integration potential between scientific and technological achievements and cutting-edge technologies can be quantified based on rule mapping and keyword matching. For instance, the integration potential can be quantified by checking whether the achievements explicitly indicate support for cutting-edge technology interfaces, whether the technology field of the achievements belongs to the same broad category as the cutting-edge technologies, and by matching the keywords of the achievements with the keyword database of cutting-edge technologies. The value of the feature of application scenarios that scientific and technological achievements are suitable for can be quantified by the number of application scenarios they are suitable for and whether these application scenarios belong to the key scenarios supported by national policies.

[0034] In one exemplary implementation, a first preset mapping relationship can be pre-constructed to indicate the correspondence between the predicted scenario and the type of result and the value characteristics of the target result. Then, based on the first preset mapping relationship, the target result value characteristics are selected from all the result value characteristics of the scientific and technological results to be predicted, according to the scenario to be predicted and the result type of the scientific and technological results to be predicted.

[0035] For example, Figure 4 This diagram illustrates a flowchart of a method for determining a first preset mapping relationship according to an exemplary embodiment of this disclosure. (See reference...) Figure 4 The method may include steps S410 to S450. Wherein: In step S410, for each outcome type under each prediction scenario, the training sample corresponding to the outcome type and the value label of the training sample under the prediction scenario are obtained.

[0036] For example, we can collect scientific and technological achievements corresponding to different types of achievements, and then determine the value label of each type of scientific and technological achievement in each prediction scenario through manual annotation, thereby generating training samples and value labels for each combination of prediction scenario and achievement type.

[0037] For example, by having experts label the value of each scientific and technological achievement in each prediction scenario, the value of the achievement can be a grade value, such as high, medium, and low, or a specific value value, thereby obtaining the training sample corresponding to each type of achievement and the value label of the training sample in each prediction scenario.

[0038] It should be noted that, Figure 4 The purpose of the method shown is to determine the first preset mapping relationship. In the process of determining the first preset mapping relationship, it is only necessary to identify the feature that is most relevant to the value of each type of scientific and technological achievement in each prediction scenario. Therefore, the value label at this time can be either a hierarchical value label or a quantitative value label.

[0039] In step S420, the second feature value of the full value features of the training samples is extracted, and the first candidate model is trained based on the second feature value and the value label to obtain the first target model.

[0040] For example, for each training sample under each combination of prediction scenario and outcome type, the second feature value of the full value features of the training sample can be extracted. Then, the first candidate model is trained based on the second feature value and the value label to obtain the first target model under that combination of prediction scenario and outcome type.

[0041] For example, if there are 4 prediction scenarios and 5 outcome types, then 20 primary target models can be obtained.

[0042] In step S430, the contribution of each feature is calculated based on the feature importance score output by the first target model.

[0043] In one exemplary implementation, the first candidate model may include a model with feature importance attributes, such as a random forest model. Based on this, the feature importance score is determined according to the feature importance attributes of the first target model.

[0044] For example, after the random forest model is trained, feature importance scores can be output through the feature_importances attribute. For instance, the values ​​in the feature_importances attribute can be used as feature importance scores, allowing the calculation of the contribution of each feature based on these scores.

[0045] The contribution of each feature can be determined based on the proportion of its importance score to the sum of the importance scores of all features.

[0046] In step S440, the feature ranking is determined based on the feature importance score, and the cumulative contribution of the feature is calculated based on the feature ranking and the contribution of each feature.

[0047] For example, feature rankings can be determined based on feature importance scores from highest to lowest, and then the cumulative contribution of each feature can be calculated based on the feature ranking and the contribution of each feature.

[0048] For example, there are 5 features, namely features A, B, C, D, and E, ranked in the order E, C, D, A, B. Feature E contributes 30%, feature C contributes 25%, feature D contributes 20%, feature A contributes 15%, and feature B contributes 10%. Then the cumulative contribution of feature E is 30%, the cumulative contribution of feature C is the sum of the contributions of feature E and feature C, which is 55%, and so on, with feature D contributing 75%.

[0049] In step S450, the first preset mapping relationship is determined based on the cumulative contribution of the features.

[0050] For example, one exemplary implementation of step S450 may include: determining the participation feature for calculation when the cumulative contribution is greater than or equal to a preset threshold, determining the participation feature as the target result value feature corresponding to the result type in the prediction scenario; and establishing the first preset mapping relationship based on the target result value feature corresponding to each result type in each prediction scenario.

[0051] Taking a preset threshold of 75% as an example, in the example of features A, B, C, D, and E above, the ranking order is E, C, D, A, and B. The cumulative contribution of feature D is 75%, which is equal to the preset threshold. When the cumulative contribution is equal to 75%, the participating features in the calculation are feature D and the features ranked before it, namely features E, C, and D. Therefore, the target outcome value features are E, C, and D.

[0052] In another embodiment of this disclosure, in order to make the number of effective features more balanced for different prediction scenarios and different result types and avoid model learning bias, the maximum number of dimensions of the first candidate feature under each combination can be counted. For combinations with insufficient number of dimensions, second candidate features are added according to feature importance, so that the number of value features of the target result tends to be consistent.

[0053] Based on this, another exemplary implementation of step S450 may include: determining the participating features that participate in the calculation when the cumulative contribution is greater than or equal to a preset threshold, and using the participating features as the first candidate features corresponding to the outcome type in the prediction scenario; calculating the maximum value of the number of feature dimensions of the first candidate features corresponding to each outcome type in each prediction scenario; when the maximum value is less than or equal to a preset number, calculating the dimension difference between the number of feature dimensions and the maximum value for outcome types in prediction scenarios where the number of feature dimensions of the first candidate features is less than the maximum value; selecting the top N second candidate features after the first candidate features from the full set of value features according to the feature importance score ranking, where N is equal to the dimension difference; determining the target outcome value feature corresponding to the outcome type in the prediction scenario based on the first candidate feature and the second candidate features; and establishing a first preset mapping relationship based on the target outcome value feature corresponding to each outcome type in each prediction scenario.

[0054] Taking a preset threshold of 75% as an example, in the example of features A, B, C, D, and E above, the ranking order is E, C, D, A, and B. The cumulative contribution of feature D is 75%, which is equal to the preset threshold. When the cumulative contribution is equal to 75%, the participating features in the calculation are feature D and the features ranked before it, namely features E, C, and D. Therefore, the first candidate features are E, C, and D.

[0055] Taking a maximum value of 4 as an example, for the outcome type in the prediction scenario where the number of first candidate features is less than 4, the second candidate feature can be selected from the features ranked after the first candidate feature based on the difference between the current number of first candidate features and 4, until the sum of the number of first candidate features and the second candidate feature reaches 4, thereby determining both the first candidate feature and the second candidate feature as the target outcome value features.

[0056] For example, continuing with the example of features A, B, C, D, and E above, the first candidate features are E, C, and D. Since the number of first candidate features is 3 (3 is less than 4), and the ranking order is E, C, D, A, B, feature A is determined as the second candidate feature. Ultimately, the target achievement value features for this type of scientific and technological achievement in this prediction scenario include features E, C, D, and A.

[0057] After determining the target outcome value characteristics corresponding to each outcome type under each prediction scenario, a first preset mapping relationship can be established based on this correspondence.

[0058] Next, the specific implementation method of "step S120, obtaining the first target input feature value based on the result type coding feature value of the scientific and technological achievement to be predicted, the scene coding feature value of the scenario to be predicted, and the first feature value of the target result value feature of the scientific and technological achievement to be predicted" will be explained in detail.

[0059] For example, one exemplary implementation of step S120 may include: determining the feature values ​​of other features besides the target result value features in the total result value features according to a preset invalid identifier value; extracting the first feature value of the target result value feature of the scientific and technological result to be predicted; and concatenating the result type encoding feature value of the scientific and technological result to be predicted, the scene encoding feature value of the scene to be predicted, the feature values ​​of other features, and the first feature value of the target result value feature of the scientific and technological result to be predicted to obtain the target input feature value.

[0060] For example, the preset invalid identifier value can be customized according to needs or experience. For instance, since the characteristic value of all target outcome value features cannot be -999, the preset invalid identifier value can be -999.

[0061] For example, the target input feature value can be obtained by assembling the result type encoding feature value of the scientific and technological achievement to be predicted, the scene encoding feature value of the scene to be predicted, the feature values ​​of other features, and the first feature value of the target result value feature of the scientific and technological achievement to be predicted, according to a preset order. The preset order can be customized according to requirements, and this exemplary embodiment does not impose any special limitations on it. However, once the preset order is fixed, it is the same for all scientific and technological achievements to be predicted in all prediction scenarios, and the preset order is the same in both the training phase and the actual prediction phase.

[0062] By pre-setting invalidation flags, the dimension of the target input feature values ​​of the scientific and technological achievements to be predicted can be the same under different prediction scenarios and different types of results. This enables multiple prediction scenarios and multiple types of results to share the same value prediction model, avoids building separate models for different scenarios or different types of results, reduces model maintenance costs, and improves prediction versatility and efficiency.

[0063] Next, a detailed description will be given of the specific implementation method of "step S130, inputting the first target input feature value into the pre-trained scientific and technological achievement value prediction model, and determining the achievement value of the scientific and technological achievement to be predicted in the scenario to be predicted based on the output of the scientific and technological achievement value prediction model".

[0064] For example, Figure 5 This diagram illustrates a flowchart of a method for determining a pre-trained model for predicting the value of scientific and technological achievements, as shown in an exemplary embodiment of this disclosure. (Reference) Figure 5The method may include steps S510 to S590. Wherein: In step S510, training samples and sample labels are obtained. The training samples include sample scientific and technological achievements of different achievement types and the label achievement values ​​of the sample scientific and technological achievements under different prediction scenarios.

[0065] In one exemplary implementation, the value of the labeling outcome includes the quantitative value of the labeling outcome and the potential value level of the labeling outcome.

[0066] For example, the labeled research results value in this disclosure includes quantitative value scoring labels for the results. If each result is assigned a value range of 0-100, then the quantitative value of the labeled result can be any value within that range. The result value potential level is used to characterize the value level of the scientific and technological achievement, such as high, medium, and low levels, which can be customized as needed. The labeled result value can be obtained by experts labeling each sample of scientific and technological achievements.

[0067] It should be noted that the difference in the number of scientific and technological achievements of each type of sample needs to be kept within a reasonable range. For example, if the difference in the number of scientific and technological achievements of each type of sample is less than the preset value, the model can avoid learning bias due to the imbalance of sample size and improve the accuracy and generalization ability of the model prediction.

[0068] In step S520, the target result value characteristics of the sample scientific and technological results are selected from the full set of result value characteristics based on the predicted scenario and result type.

[0069] For example, for each sample scientific and technological achievement and its sample value label, the target achievement value feature of the sample scientific and technological achievement can be selected from the full set of achievement value features based on the prediction scenario and achievement type to which the sample value label belongs, and then the second feature value corresponding to the target achievement value feature of the sample scientific and technological achievement can be extracted.

[0070] In step S530, the second target input feature value is obtained based on the result type encoding feature value of the sample scientific and technological achievement, the scene encoding feature value of the predicted scenario, and the second feature value of the target result value feature of the sample scientific and technological achievement.

[0071] For example, the second target input feature value can be obtained by splicing the result type encoding feature value of the sample scientific and technological achievement, the scene encoding feature value of the prediction scenario, and the second feature value of the target result value feature of the sample scientific and technological achievement according to the above preset order.

[0072] In step S540, the second target input feature value is input into the second candidate model, and the quantitative value of the prediction result and the potential level of the prediction result value are determined based on the output of the second candidate model.

[0073] For example, the second candidate model can be any machine learning model, such as a neural network model with a multi-task output structure, etc., and this exemplary embodiment does not impose any special limitations on it. By inputting the second target input feature value into the second candidate model, the second candidate model can predict the quantitative value and value potential level of the sample scientific and technological achievement, thereby obtaining the predicted quantitative value and predicted value potential level of the achievement.

[0074] In step S550, based on the second preset mapping relationship, the first task weight and the second task weight are determined for each combination of prediction scenario and outcome type.

[0075] For example, a second pre-defined mapping relationship can be constructed in advance. This second pre-defined mapping relationship is used to characterize the first task weight and the second task weight corresponding to each combination of prediction scenarios and outcome types. The first task weight is used to characterize the weight of the quantitative value prediction result, and the second task weight is used to characterize the weight of the outcome value potential level prediction result.

[0076] The weights of the first and second tasks can be determined based on experience, and this exemplary implementation does not impose any special limitations on them. For example, in a results transaction scenario, more attention is paid to the specific value of the results, so the weight of the first task can be greater than the weight of the second task; in a scientific research performance evaluation scenario, more attention is paid to the value level classification of the results, so the weight of the second task can be greater than the weight of the first task.

[0077] In step S560, for each subset of sample scientific and technological achievements under each combination of prediction scenario and achievement type, a first degree of difference between the tagged quantitative value of the sample scientific research achievements in the subset and the predicted quantitative value of the results, and a second degree of difference between the tagged value potential level and the predicted value potential level of the results are calculated.

[0078] For example, sample scientific and technological achievements can be classified according to the prediction scenario and the type of achievement to obtain a subset of sample scientific and technological achievements under each combination of prediction scenario and achievement type. For each subset, the mean square error between the label quantitative value of the sample scientific and technological achievements in the subset and the predicted quantitative value of the achievement is calculated to obtain the first degree of difference corresponding to the subset. The cross-entropy loss between the label value potential level and the predicted value potential level of the sample scientific and technological achievements in the subset is calculated to obtain the second degree of difference corresponding to the subset.

[0079] In step S570, for each combination of prediction scenario and outcome type, the sum of the first product between the first task weight and the first degree of difference and the second product between the second task weight and the second degree of difference is calculated to determine the degree of difference.

[0080] For example, the sum and difference of any subset can be obtained using the following formula (1): (1) In formula (1), The first task weight of the subset. The first degree of difference corresponding to the subset. The second task weight of the subset. The second degree of difference corresponding to the subset, where, This can be understood as the sum of the first degree of difference corresponding to all sample scientific and technological achievements in the subset. It can be understood as the sum of the second degree of difference corresponding to all sample scientific and technological achievements in the subset.

[0081] In step S580, the training loss is determined by weighted summation of the sum and degree of difference corresponding to each combination of prediction scenario and outcome type.

[0082] For example, the training loss can be determined by weighting and summing the final losses for all combinations of prediction scenarios and outcome types based on the number of samples. For instance, the larger the number of samples for a combination of prediction scenarios and outcome types, the greater the weight of the degree of difference. Therefore, the training loss can be obtained by weighting and summing the degree of difference for each subset based on this weight.

[0083] In step S590, the second candidate model is iteratively trained based on the training loss until the preset conditions are met, thereby obtaining the pre-trained scientific and technological achievement value prediction model.

[0084] For example, when the training loss is less than a preset value or the number of iterations reaches a preset number, a candidate scientific and technological achievement value prediction model is obtained. The model performance of the candidate scientific and technological achievement value prediction model can be tested based on test samples, such as testing the prediction accuracy and recall rate of the model. If the test is passed, the candidate scientific and technological achievement value prediction model that has passed the test is determined as the pre-trained scientific and technological achievement value prediction model. If the test is not passed, the model is retrained until the test is passed, and the pre-trained scientific and technological achievement value prediction model is obtained.

[0085] By performing multi-task training on the model through steps S510 to S590, the model can learn the correlation between features and values ​​from different perspectives, reducing the risk of overfitting on a single task.

[0086] For example, Figure 6 This diagram illustrates a flowchart of a method for determining the value of a technological achievement to be predicted in a predicted scenario, according to an exemplary embodiment of this disclosure. (See reference...) Figure 6 The method may include steps S610 to S640. Wherein: In step S610, the first target input feature value is input into the pre-trained scientific and technological achievement value prediction model, and the quantitative value prediction value and the potential value prediction level of the scientific and technological achievement to be predicted are obtained according to the output of the scientific and technological achievement value prediction model.

[0087] For example, as mentioned earlier, a pre-trained scientific and technological achievement value prediction model can simultaneously output the quantitative value and value potential level of the scientific and technological achievement to be predicted.

[0088] In step S620, the quantitative value prediction value and the result value potential prediction level are normalized to obtain the first normalized value and the second normalized value.

[0089] For example, both the quantitative value prediction and the outcome value potential prediction level can be normalized to the 0-1 range to obtain the corresponding first normalized value and second normalized value.

[0090] For example, the predicted quantitative value can be divided by 100 to normalize it to the 0-1 range. Based on the preset mapping relationship, the predicted level of the potential value of the achievement can be directly mapped to a value in the 0-1 range, such as high level mapped to 1, medium level mapped to 0.6, and low level mapped to 0.1. Alternatively, it can be normalized to the 0-1 range based on the number of registrations and the preset value corresponding to each level. For example, if the number of levels is 3, the preset value corresponding to high level is 3, the preset value corresponding to medium level is 2, and the preset value corresponding to low level is 1. Each preset value divided by the number of levels 3 is the normalization result, i.e., the second normalized value.

[0091] In step S630, based on the second preset mapping relationship, the first task weight and the second task weight of the scientific and technological achievement to be predicted in the scenario to be predicted are determined.

[0092] The specific implementation of step S630 can be referred to the specific implementation of step S550 above, and will not be repeated here.

[0093] In step S640, the value of the scientific and technological achievement to be predicted in the scenario to be predicted is determined by the sum of the third product between the first task weight and the first normalized value and the fourth product between the second task weight and the second normalized value.

[0094] For example, a third product can be calculated between the first task weight and the first normalized value, and a fourth product can be calculated between the second task weight and the second normalized value. Then, the sum of the third and fourth products can be calculated to obtain the value of the result to be predicted in the scenario to be predicted.

[0095] In this disclosure, the final outcome value can be obtained by weighting the quantitative prediction results and the value potential level prediction results, thereby improving the reliability and accuracy of outcome value prediction. Specifically, the quantitative value and qualitative level prediction results can be mutually calibrated to improve the prediction accuracy of the model.

[0096] During the model training phase, the quantitative value prediction task and the value potential level prediction task can constrain and assist each other, enabling the model to learn more stable and robust value characteristics, thereby improving the overall prediction accuracy.

[0097] In the prediction phase, the quantitative value prediction and the value potential level prediction results are normalized separately, and then weighted and summed according to scenario weights to obtain the final result value that integrates both information. For example, if the quantitative score prediction for a certain scientific and technological achievement is 78 points and the value potential level prediction is "high potential", after normalizing the quantitative score and level separately, they are weighted and fused according to preset weights to obtain the final result value that simultaneously reflects quantitative accuracy and level differentiation. This avoids the situation where the overall evaluation error is large due to the bias of a single prediction result, and achieves comprehensive optimization of the prediction results, thereby improving the reliability of the prediction results.

[0098] In another exemplary implementation, when the quantitative value prediction result and the value potential level prediction result are consistent (e.g., the quantitative value prediction result falls within the range of the value potential level prediction result), the quantitative value prediction and the value potential level prediction result are combined and output together. Users can determine which prediction result to use to represent the final value of the scientific and technological achievement according to their needs. However, if the quantitative value prediction result does not fall within the range of the value potential level prediction result, it indicates a significant difference between the two prediction results, and an anomaly is determined. Feature extraction and prediction can then be re-executed for the scientific and technological achievement to ensure the accuracy of the final value. If the anomaly persists after multiple repetitions, the value of the achievement can be determined manually by experts.

[0099] In this disclosure, quantitative regression and hierarchical classification tasks supervise and reinforce each other during training, enabling the model to learn more robust features and achieve higher accuracy and reliability than single-task models. Furthermore, the model outputs both quantitative value scores and value potential levels in a single step, eliminating the need for multiple independent models, simplifying the system architecture, and improving prediction efficiency. Simultaneously, the inherent correspondence between quantitative scores and levels is utilized to achieve indirect verification between results, preventing complete distortion of the final evaluation due to bias in a single task, thus improving the reliability of the prediction results.

[0100] Furthermore, it should be noted that the above figures are merely illustrative representations of the processes included in the method according to exemplary embodiments of the present invention, and are not intended to be limiting. It is readily understood that the processes shown in the above figures do not indicate or limit the temporal order of these processes. Additionally, it is readily understood that these processes may be executed synchronously or asynchronously, for example, in multiple modules.

[0101] Furthermore, the exemplary embodiments of this disclosure also provide a device for predicting the value of scientific and technological achievements. (See reference...) Figure 7 As shown, the technology achievement value prediction device 700 includes the following program modules: a feature selection module 710, configured to select the target achievement value feature of the technology achievement to be predicted from all achievement value features based on the scenario to be predicted and the achievement type of the technology achievement to be predicted; an input feature value determination module 720, configured to obtain a first target input feature value based on the achievement type encoding feature value of the technology achievement to be predicted, the scenario encoding feature value of the scenario to be predicted, and a first feature value of the target achievement value feature of the technology achievement to be predicted; and a value prediction module 730, configured to input the first target input feature value into a pre-trained technology achievement value prediction model, and determine the achievement value of the technology achievement to be predicted in the scenario to be predicted based on the output of the technology achievement value prediction model.

[0102] In one exemplary implementation, selecting target result value features from the full set of result value features of the scientific and technological achievement to be predicted, based on the scenario to be predicted and the result type of the scientific and technological achievement to be predicted, includes: selecting target result value features from the full set of result value features of the scientific and technological achievement to be predicted, based on a first preset mapping relationship, according to the scenario to be predicted and the result type of the scientific and technological achievement to be predicted, wherein the first preset mapping relationship is used to indicate the correspondence between the prediction scenario and the result type and the target result value feature; wherein, the method for determining the first preset mapping relationship includes: for each result type under each prediction scenario, obtaining training samples corresponding to the result type and value labels of the training samples under the prediction scenario; extracting second feature values ​​of the full set of value features of the training samples, training a first candidate model based on the second feature values ​​and the value labels to obtain a first target model; calculating the contribution of each feature based on the feature importance score output by the first target model; determining the feature ranking based on the feature importance score, calculating the cumulative contribution of the feature based on the feature ranking and the contribution of each feature; and determining the first preset mapping relationship based on the cumulative contribution of the feature.

[0103] In one exemplary implementation, determining the first preset mapping relationship based on the cumulative contribution of the feature includes: determining the participating features that participate in the calculation when the cumulative contribution is greater than or equal to a preset threshold, and using the participating features as the first candidate features corresponding to the outcome type in the prediction scenario; and calculating the maximum value of the number of feature dimensions of the first candidate features corresponding to each outcome type in each prediction scenario. When the maximum value is less than or equal to a preset number, for the outcome type in the prediction scenario where the number of feature dimensions of the first candidate feature is less than the maximum value, the dimension difference between the number of feature dimensions and the maximum value is calculated; based on the feature importance score ranking, the top N second candidate features ranked after the first candidate feature are selected from the full set of value features, where N is equal to the dimension difference; based on the first candidate feature and the second candidate feature, the target outcome value feature corresponding to the outcome type in the prediction scenario is determined; based on the target outcome value feature corresponding to each outcome type in each prediction scenario, the first preset mapping relationship is established.

[0104] In one exemplary implementation, the method for determining the pre-trained scientific and technological achievement value prediction model includes: acquiring training samples and sample labels, wherein the training samples include sample scientific and technological achievements of different achievement types and the labeled achievement values ​​of the sample scientific and technological achievements under different prediction scenarios, wherein the labeled achievement values ​​include the labeled achievement quantitative value and the labeled achievement value potential level; selecting the target achievement value features of the sample scientific and technological achievements from the full set of achievement value features according to the prediction scenario and achievement type; obtaining a second target input feature value based on the achievement type encoding feature value of the sample scientific and technological achievements, the scenario encoding feature value of the prediction scenario, and the second feature value of the target achievement value features of the sample scientific and technological achievements; inputting the second target input feature value into a second candidate model, and determining the predicted achievement quantitative value and the predicted achievement value potential level according to the output of the second candidate model; and determining the predicted achievement quantitative value and the predicted achievement value potential level based on a second preset mapping relationship. The process involves determining the first task weight and the second task weight for each combination of prediction scenarios and outcome types; for each subset of sample scientific and technological achievements under each combination of prediction scenarios and outcome types, calculating the first degree of difference between the labeled quantitative value of the sample scientific and technological achievements and the predicted quantitative value, as well as the second degree of difference between the labeled value potential level and the predicted value potential level; for each combination of prediction scenarios and outcome types, calculating the sum of the first product between the first task weight and the first degree of difference, and the second product between the second task weight and the second degree of difference; weighted summation of the sum of the sums ...

[0105] In one exemplary implementation, the step of inputting the first target input feature value into a pre-trained scientific and technological achievement value prediction model and determining the achievement value of the scientific and technological achievement to be predicted in the scenario to be predicted based on the output of the scientific and technological achievement value prediction model includes: inputting the first target input feature value into the pre-trained scientific and technological achievement value prediction model; obtaining the quantitative value prediction value and the potential value prediction level of the scientific and technological achievement to be predicted based on the output of the scientific and technological achievement value prediction model; normalizing the quantitative value prediction value and the potential value prediction level of the achievement to be predicted to obtain a first normalized value and a second normalized value; determining the first task weight and the second task weight of the scientific and technological achievement to be predicted in the scenario to be predicted based on a second preset mapping relationship; and determining the achievement value of the scientific and technological achievement to be predicted in the scenario to be predicted based on the sum of the third product between the first task weight and the first normalized value and the fourth product between the second task weight and the second normalized value.

[0106] In one exemplary embodiment, the total value characteristics of the achievements include basic attribute characteristics, application effect characteristics, and value potential characteristics of different types of scientific and technological achievements; wherein, the basic attribute characteristics include one or more of the following: technological novelty characteristics, technological maturity characteristics, technological stability characteristics, and achievement level characteristics; the application effect characteristics are determined based on one or more of the following: the frequency of application of the scientific and technological achievements in existing scientific and technological projects, application benefits, and the level of the applied scientific and technological projects; the value potential characteristics of the scientific and technological achievements are determined based on one or more of the following: the matching degree between the scientific and technological achievements and policies, the integration potential between the scientific and technological achievements and cutting-edge technologies, and the application scenarios to which the scientific and technological achievements are adapted.

[0107] In one exemplary implementation, obtaining the first target input feature value based on the result type coding feature value of the scientific and technological achievement to be predicted, the scene coding feature value of the scene to be predicted, and the first feature value of the target result value feature of the scientific and technological achievement to be predicted includes: determining the feature values ​​of other features besides the target result value feature in the full set of result value features according to a preset invalid identifier value; extracting the first feature value of the target result value feature of the scientific and technological achievement to be predicted; and concatenating the result type coding feature value of the scientific and technological achievement to be predicted, the scene coding feature value of the scene to be predicted, the feature values ​​of other features, and the first feature value of the target result value feature of the scientific and technological achievement to be predicted to obtain the first target input feature value.

[0108] The specific details of each part of the above-mentioned device have been described in detail in the method section of the implementation plan. For any undisclosed details, please refer to the implementation plan of the method section, and therefore will not be repeated here.

[0109] It should be noted that although several modules or units for the device used to perform actions have been mentioned in the detailed description above, this division is not mandatory. In fact, according to exemplary embodiments of this disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.

[0110] Furthermore, although the steps of the method in this disclosure are described in a specific order in the accompanying drawings, this does not require or imply that the steps must be performed in that specific order, or that all the steps shown must be performed to achieve the desired result. Additional or alternative steps may be omitted, multiple steps may be combined into one step, and / or a step may be broken down into multiple steps.

[0111] An exemplary embodiment of this disclosure also provides a computer program product. The computer program product includes a computer program that, when executed by a processor, implements the above-described method for predicting the value of scientific and technological achievements.

[0112] In one implementation, the computer program product can be a tangible product containing a computer program, such as a computer-readable storage medium storing the computer program. The readable storage medium can be a storage medium based on electrical, magnetic, optical, electromagnetic, infrared, or other signals, including but not limited to: random access memory (RAM), read-only memory (ROM), magnetic tape, floppy disk, flash memory, hard disk drive (HDD), solid-state drive (SSD), etc. For example, the computer program product can be implemented as a non-volatile storage medium storing a computer program, such as read-only memory, NAND flash memory, etc.

[0113] In one implementation, the computer program product can be an intangible product containing a computer program. For example, the computer program product can be implemented as a virtual digital product, such as an executable file, installation package, or other digital file storing the computer program.

[0114] Computer program code can be written in one or more programming languages. Examples of programming languages ​​include C, Java, C++, and Python. Program code can execute entirely on the user's computing device, partially on the user's computing device, or as a standalone software package. It can also execute partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, such as a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via an internet connection provided by a mobile network operator).

[0115] Computer programs can be carried or transmitted via signals such as electricity, magnetism, light, electromagnetic radiation, and infrared radiation. Electronic devices can convert the signals carrying computer programs into digital signals, thereby running the computer programs. When a computer program runs on an electronic device, its code is used to cause the electronic device to execute (more specifically, to execute by the processor of the electronic device) the method steps of various exemplary embodiments of this disclosure, such as the above-described method for predicting the value of scientific and technological achievements.

[0116] Exemplary embodiments of this disclosure also provide an electronic device, such as a terminal device or a server. The electronic device may include a processor and a memory. The memory stores executable instructions of the processor, such as a computer program. The processor executes the executable instructions to perform the method steps of various exemplary embodiments of this disclosure. Furthermore, the electronic device may also include a display for displaying a graphical user interface.

[0117] The following is for reference. Figure 8 The electronic device is illustrated by way of a general-purpose computing device. It should be understood that... Figure 8 The electronic device 800 shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments disclosed herein.

[0118] like Figure 8 As shown, the electronic device 800 may include: a processor 810, a memory 820, a bus 830, an I / O (input / output) interface 840, a network adapter 850, and a display 860.

[0119] The memory 820 may include volatile memory, such as RAM 821 and cache unit 822, and may also include non-volatile memory, such as ROM 823. The memory 820 may also include one or more program modules 824, including but not limited to: an operating system, one or more application programs, other program modules, and program data. Each or some combination of these examples may include an implementation of a network environment. For example, program module 824 may include the modules described above.

[0120] The processor 810 may include one or more processing units, such as an AP (Application Processor), a modem processor, a GPU (Graphics Processing Unit), an ISP (Image Signal Processor), a controller, an encoder, a decoder, a DSP (Digital Signal Processor), a baseband processor, and / or an NPU (Neural-Network Processing Unit).

[0121] The processor 810 can be used to execute executable instructions stored in the memory 820, such as the aforementioned method for predicting the value of scientific and technological achievements.

[0122] Bus 830 is used to connect different components of electronic device 800 and may include data bus, address bus and control bus.

[0123] Electronic device 800 can communicate with one or more external devices 900 (such as keyboard, mouse, external controller, etc.) through I / O interface 840.

[0124] Electronic device 800 can communicate with one or more networks via network adapter 850. For example, network adapter 850 can provide mobile communication solutions such as 3G / 4G / 5G, or wireless communication solutions such as wireless LAN, Bluetooth, and near-field communication. Network adapter 850 can communicate with other modules of electronic device 800 via bus 830.

[0125] Electronic device 800 can display a graphical user interface, such as a value prediction result display interface, through display 860.

[0126] although Figure 8 As not shown in the diagram, other hardware and / or software modules may also be configured in the electronic device 800, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0127] Those skilled in the art will understand that various aspects of this disclosure can be implemented in the following forms: a completely hardware implementation, a completely software implementation (including firmware, microcode, etc.), or a combination of hardware and software implementations, which may be referred to as "circuit", "module" or "system" respectively.

[0128] It should be understood that this disclosure is not limited to the specific methods, steps, or structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. Those skilled in the art will readily conceive of other embodiments based on the specific implementations provided in this disclosure. Therefore, the specific implementations provided in this disclosure are merely exemplary, and the scope and spirit of this disclosure are indicated by the claims, and should cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary technical means in the art not disclosed in this disclosure.

Claims

1. A method for predicting the value of scientific and technological achievements, characterized in that, include: Based on the scenario to be predicted and the type of the scientific and technological achievement to be predicted, the target achievement value characteristics of the scientific and technological achievement to be predicted are selected from the value characteristics of all achievements. Based on the result type coding feature value of the scientific and technological achievement to be predicted, the scene coding feature value of the scenario to be predicted, and the first feature value of the target result value feature of the scientific and technological achievement to be predicted, the first target input feature value is obtained. The first target input feature value is input into a pre-trained scientific and technological achievement value prediction model, and the achievement value of the scientific and technological achievement to be predicted in the predicted scenario is determined based on the output of the scientific and technological achievement value prediction model.

2. The method according to claim 1, characterized in that, The process of selecting target result value features from the total result value features of the scientific and technological achievements to be predicted, based on the scenario to be predicted and the type of achievement to be predicted, includes: Based on the first preset mapping relationship, according to the scenario to be predicted and the type of the scientific and technological achievement to be predicted, the target achievement value feature is selected from the full set of achievement value features of the scientific and technological achievement to be predicted. The first preset mapping relationship is used to indicate the correspondence between the prediction scenario and achievement type and the target achievement value feature. The method for determining the first preset mapping relationship includes: For each outcome type in each prediction scenario, obtain the training sample corresponding to the outcome type and the value label of the training sample in the prediction scenario; Extract the second feature value of the full value features of the training samples, and train the first candidate model based on the second feature value and the value label to obtain the first target model; Based on the feature importance score output by the first target model, calculate the contribution of each feature; The feature ranking is determined based on the feature importance score, and the cumulative contribution of the feature is calculated based on the feature ranking and the contribution of each feature. The first preset mapping relationship is determined based on the cumulative contribution of the features.

3. The method according to claim 2, characterized in that, Determining the first preset mapping relationship based on the cumulative contribution of the features includes: The participation features that participate in the calculation when the cumulative contribution is greater than or equal to a preset threshold are determined, and the participation features are determined as the target result value features corresponding to the result type in the prediction scenario; The first preset mapping relationship is established based on the target outcome value characteristics corresponding to each outcome type in each prediction scenario.

4. The method according to claim 1, characterized in that, The methods for determining the pre-trained scientific and technological achievement value prediction model include: Acquire training samples and sample labels. The training samples include sample scientific and technological achievements of different achievement types and the label achievement value of the sample scientific and technological achievements under different prediction scenarios. The label achievement value includes the quantitative value of the label achievement and the potential level of the label achievement value. Based on the prediction scenario and the type of outcome, the target outcome value characteristics of the sample scientific and technological achievements are selected from the value characteristics of all outcomes. Based on the result type coding feature value of the sample scientific and technological achievements, the scenario coding feature value of the prediction scenario, and the second feature value of the target result value feature of the sample scientific and technological achievements, the second target input feature value is obtained. The second target input feature value is input into the second candidate model, and the quantitative value of the prediction result and the potential level of the prediction result value are determined based on the output of the second candidate model. Based on the second preset mapping relationship, the first task weight and the second task weight are determined for each combination of prediction scenario and outcome type. For each subset of sample scientific and technological achievements under each combination of prediction scenario and achievement type, calculate the first degree of difference between the tagged quantitative value of the sample scientific research achievements in the subset and the predicted quantitative value, as well as the second degree of difference between the tagged value potential level and the predicted value potential level. For each combination of prediction scenarios and outcome types, calculate the sum of the differences between the first product of the first task weight and the first degree of difference and the second product of the second task weight and the second degree of difference; The training loss is determined by weighting and summing the sums corresponding to each combination of prediction scenarios and outcome types and the degree of difference. The second candidate model is iteratively trained based on the training loss until the preset conditions are met, thereby obtaining the pre-trained scientific and technological achievement value prediction model.

5. The method according to claim 4, characterized in that, The step of inputting the first target input feature value into a pre-trained scientific and technological achievement value prediction model, and determining the achievement value of the scientific and technological achievement to be predicted in the scenario to be predicted based on the output of the scientific and technological achievement value prediction model, includes: The first target input feature value is input into the pre-trained scientific and technological achievement value prediction model, and the quantitative value prediction value and the value potential prediction level of the scientific and technological achievement to be predicted are obtained according to the output of the scientific and technological achievement value prediction model. The quantitative value prediction value and the result value potential prediction level are normalized respectively to obtain the first normalized value and the second normalized value; Based on the second preset mapping relationship, the first task weight and the second task weight of the scientific and technological achievement to be predicted in the scenario to be predicted are determined. The value of the scientific and technological achievement to be predicted in the scenario to be predicted is determined by summing the third product between the first task weight and the first normalized value and the fourth product between the second task weight and the second normalized value.

6. The method according to claim 1, characterized in that, The value characteristics of the total amount of achievements include the basic attribute characteristics, application effect characteristics, and value potential characteristics of different types of scientific and technological achievements; The basic attribute characteristics include one or more of the following: technological novelty characteristics, technological maturity characteristics, technological stability characteristics, and achievement level characteristics of scientific and technological achievements; The application effect characteristics are determined based on one or more of the following: the frequency of application of the scientific and technological achievements in existing scientific and technological projects, the application benefits, and the level of the scientific and technological projects in which they are applied. The value potential characteristics of the scientific and technological achievements are determined based on one or more of the following: the degree of matching between the scientific and technological achievements and policies, the potential for integration between the scientific and technological achievements and cutting-edge technologies, and the application scenarios to which the scientific and technological achievements are adapted.

7. The method according to claim 1, characterized in that, The first target input feature value is obtained by using the result type coding feature value of the scientific and technological achievement to be predicted, the scene coding feature value of the scenario to be predicted, and the first feature value of the target result value feature of the scientific and technological achievement to be predicted. The feature values ​​of other features in the total value features of the results, excluding the target value features, are determined based on the preset invalid identifier values. Extract the first feature value of the target achievement value feature of the scientific and technological achievement to be predicted; The first target input feature value is obtained by concatenating the result type coding feature value of the scientific and technological achievement to be predicted, the scene coding feature value of the scene to be predicted, the feature values ​​of other features, and the first feature value of the target result value feature of the scientific and technological achievement to be predicted.

8. A device for predicting the value of scientific and technological achievements, characterized in that, include: The feature selection module is configured to select the target result value feature of the scientific and technological achievement to be predicted from the full set of result value features based on the scenario to be predicted and the result type of the scientific and technological achievement to be predicted. The input feature value determination module is configured to obtain the first target input feature value based on the result type encoding feature value of the scientific and technological result to be predicted, the scene encoding feature value of the scenario to be predicted, and the first feature value of the target result value feature of the scientific and technological result to be predicted. The value prediction module is configured to input the first target input feature value into a pre-trained scientific and technological achievement value prediction model, and determine the achievement value of the scientific and technological achievement to be predicted in the scenario to be predicted based on the output of the scientific and technological achievement value prediction model.

9. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the method described in any one of claims 1 to 7.

10. An electronic device, characterized in that, include: One or more processors; A storage device for storing one or more programs, which, when executed by one or more processors, cause the one or more processors to implement the method as described in any one of claims 1 to 7.

Images