Enterprise capability assessment method based on neurocognitive diagnosis technology

Abstract

The invention discloses an enterprise capability assessment method based on a neurocognitive diagnosis technology, and the method comprises the steps: 1, carrying out the field division of scientific and technological achievement data, and constructing a negative sample of interaction data; 2, constructing a basic neurocognitive diagnosis enterprise capability evaluation learning module, obtaining embedded vector representations such as field mastering degree, scientific and technological achievement difficulty and distinction degree, and performing representation learning on a target embedded vector; meanwhile, interaction is carried out on embedded vectors, and self-adaptive learning is carried out after the two full-connection layers; 3, constructing a partial order sorting learning module, and constructing BPR loss through the partial order learning module, an expected capability change module and an enterprise-scientific and technological achievement interaction result constraint design module; and 4, optimizing the self-adaptive diagnosis network of the whole model by using cross entropy loss and comparison loss. According to the method, the enterprise-scientific and technological achievement interaction data is utilized to perform adaptive diagnosis on scientific and technological enterprise capability evaluation, so that an effective method is provided for enterprise capability evaluation in different fields.

Description

Technical Field

[0001] The present invention relates to the field of cognitive diagnosis technology in machine learning, artificial intelligence and smart education, as well as the field of scientific and technological achievement data processing and scientific research capability evaluation of different enterprises. Specifically, it is an enterprise capability evaluation method based on neurocognitive diagnosis technology. Background Art

[0002] Cognitive diagnosis (CD) is considered as one of the key tasks of smart education applications, aiming to diagnose the ability status of each examinee based on specific assessment data, which has attracted more attention in many real-world scenarios, such as education and games. In particular, cognitive diagnosis in the field of smart education is to discover students' mastery level of different skills / knowledge concepts, which is similar to our task goal. Existing cognitive diagnosis methods can be roughly divided into two categories: statistical-based methods and deep neural network-based methods. The former is based on item response theory (IRT), multidimensional IRT (MIRT) and (DINA) methods, all of which use manually designed linear functions to simulate the interaction between students and exercises. The latter is based on neural networks to model the complex interaction patterns between students and exercises, while trying to incorporate richer contextual features and prior relationships between knowledge points to enhance the representation learning of students and exercises.

[0003] In the early days, the evaluation of scientific and technological capabilities mainly focused on qualitative analysis of the current status of the development of high-end scientific and technological talents in my country, and quantitative indicators to study academic rising stars. In the early days, the evaluation of the capabilities of listed companies in different fields also adopted quantitative indicators and qualitative analysis, so as to combine the two types of indicators to measure the capabilities of a listed company; the following key problems exist in the evaluation of enterprise-level scientific and technological achievements: previous methods only considered the capabilities of enterprises from qualitative and quantitative perspectives, and could not measure the capabilities of enterprises in a certain field from the perspective of the enterprise level in a fine-grained manner; the different achievements made by enterprises usually belong to different specific research fields, and the completion difficulty and level of the corresponding achievements in each field are also different. Different achievements correspond to different achievement types, and each achievement will have a corresponding difficulty coefficient. Because the scientific research capabilities demonstrated by different enterprises under these fine-grained influencing factors cannot be well quantified, a supervised method combining these influencing factors is urgently needed to solve this problem. At the same time, a large amount of corporate scientific and technological achievements interaction data currently has a long tail problem: a small number of companies interact with a large number of scientific and technological achievements, while a large number of companies interact with very few scientific and technological achievements. The long tail means that the recommendation system tends to recommend popular projects, which is not conducive to overall project recommendations. This data distribution will cause the existing neural network-based recommendation model to be unable to diagnose the ability status of all companies well, thereby affecting the overall prediction effect. Therefore, a method to alleviate the long tail problem is needed to deal with this data distribution phenomenon in scientific and technological achievements. Summary of the invention

[0004] The present invention aims to solve the deficiencies of the existing technologies in the above-mentioned science and technology enterprise capability assessment, and proposes an enterprise capability assessment method based on neurocognitive diagnosis technology, in order to perform adaptive diagnosis on the science and technology enterprise capability assessment, thereby achieving more effective enterprise capability assessment.

[0005] In order to achieve the above-mentioned object of the invention, the present invention adopts the following technical scheme:

[0006] The enterprise capability assessment method based on neurocognitive diagnosis technology of the present invention is characterized in that it is carried out according to the following steps:

[0007] Step 1: Define enterprise technology data and use the FastText model to divide the enterprise technology data into different fields to obtain the technology achievement field matrix. , and construct the total interaction set of enterprises and scientific and technological achievements after negative samples ;

[0008] Step 2: Construct a basic neurocognitive diagnosis enterprise capability assessment learning module, including: initial embedding representation layer, enterprise capability diagnosis layer, and prediction layer, and then perform a matrix of scientific and technological achievements on the matrix. The Scientific and technological achievements Belongs to the pth field Matrix of scientific and technological achievements Process and get the i-th enterprise For the pth field No. Scientific and technological achievements The predicted interaction results , thereby constructing a basic neurocognitive diagnostic enterprise capability assessment network for loss ;

[0009] Step 3: Construct a partial order sorting learning module, including: partial order learning module, expected capability change module, and enterprise-scientific and technological achievement interaction result constraint design module;

[0010] Step 3.1: Construct a partial order learning module and use partial order response sampling to sample the total interaction set of enterprises and scientific and technological achievements after negative samples. Perform partial sampling to obtain the i-th enterprise A reliable set of candidate interactions ;

[0011] Step 3.2: Construct an expected capability change module to calculate reliable candidate interaction sets The i-th enterprise For item u of non-interactive scientific and technological achievements The expected capability change of the interaction triplet ,and After normalization, we get the i-th enterprise For item u of non-interactive scientific and technological achievements The expected capacity change weight ;

[0012] Step 3.3: Design the constraints of enterprise-scientific and technological achievements interaction results and construct the i-th enterprise BPR loss function for all scientific and technological achievements ;

[0013] Step 4: Use formula (19) to construct the i-th enterprise The total loss function for all scientific and technological achievements : (19)

[0014] In formula (19), Hyperparameters to balance the two losses;

[0015] Step 5: Use the Adam optimizer to train the neurocognitive diagnosis enterprise capability assessment network composed of the basic neurocognitive diagnosis enterprise capability assessment learning module and the partial order sorting learning module, and minimize until Until convergence, the optimal neurocognitive diagnosis enterprise capability assessment network is obtained, which is used to diagnose the enterprise's mastery of different fields, thereby obtaining the enterprise's capability assessment results in different fields.

[0016] The enterprise capability assessment method based on neurocognitive diagnosis technology described in the present invention is also characterized in that the step 1 is performed according to the following steps:

[0017] Step 1.1: Define the number of enterprises in the technology data as The number of scientific and technological achievements is , the number of fields is , let the i-th enterprise, the m-th scientific and technological achievement and the p-th field be , and , , , ;

[0018] Let the i-th enterprise For the mth scientific and technological achievement The interaction result is recorded as ,and , represents the total set of enterprise-scientific and technological achievements interactions;

[0019] Step 1.2: Use FastText classification model to Divide the fields and obtain the scientific and technological achievements field matrix ,make The mth scientific and technological achievement Belongs to the pth field The matrix of scientific and technological achievements is recorded as ;

[0020] Step 1.3: Based on the total interaction set of enterprise-scientific and technological achievements Construct negative sample data for all enterprises;

[0021] Define the level of scientific and technological achievements as k, , C represents the maximum level;

[0022] Search the i-th enterprise according to the level k of scientific and technological achievements from high to low For all the scientific and technological achievements made, if there are corresponding scientific and technological achievements at the current level of scientific and technological achievements, no negative samples will be constructed; otherwise, Randomly select The negative samples of each non-interacting field are constructed respectively, so as to obtain the i-th enterprise Constructed negative samples, and ;in, The reverse order of the scientific and technological achievement level of the mth scientific and technological achievement, For the i-th enterprise The total number of all scientific and technological achievements made, To control the threshold of the number of negative samples for each enterprise, To control the threshold of the total number of negative samples for each enterprise;

[0023] Let the total number of scientific and technological achievements after constructing negative samples be , Scientific and technological achievements are recorded as , , the total interaction set of enterprise-scientific and technological achievements after constructing negative samples is , the total interaction set of enterprise-scientific and technological achievements The i-th enterprise For Scientific and technological achievements The interaction triplet of ,in, represents the i-th enterprise For Scientific and technological achievements The interaction result is , indicating the i-th enterprise To truly complete Scientific and technological achievements ,like , indicating the i-th enterprise For unfinished Scientific and technological achievements , the total interaction set of enterprise-scientific and technological achievements The i-th enterprise The interaction set of all scientific and technological achievements is recorded as ;

[0024] make Middle Scientific and technological achievements Belongs to the pth field The matrix of scientific and technological achievements is recorded as .

[0025] Further, the step 2 is performed according to the following steps:

[0026] Step 2.1: The initial embedding representation layer uses equations (1) and (2) to obtain the i-th enterprise Enterprise Capability Mastery Vector and Scientific and technological achievements Belongs to the pth field The result field matrix vector ;

[0027] (1)

[0028] (2)

[0029] In formula (1) and formula (2), represents the i-th enterprise The one-hot encoded vector of Indicates Scientific and technological achievements The one-hot encoded vector of ; Represents the matrix to be trained;

[0030] Step 2.2: The initial embedding representation layer constructs Scientific and technological achievements Technology metrics , the difficulty vector of scientific and technological achievements completion and Scientific and technological achievements The discrimination vector ;

[0031] Step 2.3: The enterprise capability diagnosis layer includes an interaction layer and two fully connected layers, and , as well as Processing, get the i-th enterprise The pth field completed No. Scientific and technological achievements The feature representation And dimension reduction feature representation ;

[0032] Step 2.4: The prediction layer uses formula (9) to obtain the i-th enterprise For the p-th field No. Scientific and technological achievements The predicted interaction results :

[0033] (9)

[0034] In formula (9), is the weight matrix to be trained for the prediction layer, is the bias vector of the prediction layer;

[0035] Step 2.5: Use formula (10) to calculate the i-th enterprise Loss of all scientific and technological achievements : (10)

[0036] In formula (10), represents the i-th enterprise For The completion status of scientific and technological achievements, represents the i-th enterprise For the p-th field No. Scientific and technological achievements prediction results.

[0037] Further, the step 2.2 is performed according to the following steps:

[0038] Step 2.2.1. Scientific and technological achievements The level of scientific and technological achievements is recorded as , will Scientific and technological achievements exist The corresponding quantity is recorded as , and then use formula (3) to calculate the Scientific and technological achievements Technology metrics :

[0039] (3)

[0040] In formula (3), min-max represents normalization processing;

[0041] Step 2.2.2: Use equations (4) and (5) to calculate Scientific and technological achievements The completion difficulty vector and Scientific and technological achievements The discrimination vector :

[0042] (4)

[0043] (5)

[0044] In formula (4) and formula (5), Indicates Scientific and technological achievements The one-hot encoded vector of , Respectively represent two matrices to be trained.

[0045] Further, the step 2.3 is performed according to the following steps:

[0046] Step 2.3.1: The interaction layer uses formula (6) to establish the i-th enterprise The pth field completed No. Scientific and technological achievements The interaction function :

[0047] (6)

[0048] In formula (6), is the element-wise product;

[0049] Step 2.3.2: The fully connected layer uses equations (7) and (8) to obtain the i-th enterprise The pth field completed No. Scientific and technological achievements The feature representation And dimension reduction feature representation :

[0050] (7)

[0051] (8)

[0052] In formula (7)-formula (8), represents the i-th enterprise The pth field completed No. Scientific and technological achievements The characteristic representation of The i-th enterprise The pth field completed No. Scientific and technological achievements The characteristic representation of is the sigmoid activation function, , are the weight matrices to be trained for the two fully connected layers, , are the bias vectors of the two fully connected layers respectively.

[0053] Further, the step 3.1 is performed according to the following steps:

[0054] Step 3.1.1: From the enterprise-scientific and technological achievements interaction set after constructing negative samples Get the i-th enterprise For Scientific and technological achievements The interaction triplet , and use formula (11) to extract Scientific and technological achievements Similar results set :

[0055] (11)

[0056] In formula (11), It is a collection of scientific and technological achievements that have not been exchanged among enterprises. is the uth non-interactive scientific and technological achievement, For the non-interactive scientific and technological achievements of item u Related domain sets, For the Technical achievements Related set of fields;

[0057] Step 3.1.2: Use formula (12) to construct the i-th enterprise For item u of non-interactive scientific and technological achievements The interaction triplet :

[0058] (12)

[0059] In formula (12), For the i-th enterprise For item u of non-interactive scientific and technological achievements The interaction results;

[0060] Step 3.1.3: Use formula (14) to get the i-th enterprise A reliable candidate interaction set :

[0061] (14)

[0062] In formula (14), for The number of uninteracted samples in , represents the i-th enterprise For item u of non-interactive scientific and technological achievements The interaction triplet of .

[0063] Further, the step 3.2 is performed according to the following steps:

[0064] Step 3.2.1: Use formula (15) to get the i-th enterprise In the Uninteractive scientific and technological achievements Expected capacity change :

[0065] (15)

[0066] In formula (15), represents the i-th enterprise In the Uninteractive scientific and technological achievements The expected change in capability of is given by using the diagnostic interaction function, represents the i-th enterprise exist The level of mastery of various fields, represents the i-th enterprise exist Plus the uth non-interactive scientific and technological achievements The subsequent mastery of each field;

[0067] Step 3.2.2: Calculate the uth non-interactive scientific and technological achievement using formula (16): The normalized expected capacity change weight :

[0068] (16)

[0069] In formula (16), Represents a reliable candidate interaction set The number of interactive scientific and technological achievements in the

[0070] Furthermore, the step 3..3 is performed according to the following steps:

[0071] Step 3.3.1: For the i-th enterprise , according to the different interaction results, the interaction set after constructing negative samples will be Divided into positive interaction result sets after constructing negative samples And the negative interaction result set after constructing negative samples ;

[0072] Reliable candidate interaction set The positive interaction result set is divided into reliable candidate sets and the negative interaction result set in the reliable candidate set ;

[0073] from , , , Take any positive interaction scientific and technological achievements , any candidate positive interactive scientific and technological achievements , any negative interactive scientific and technological achievements , any candidate negative interaction scientific and technological achievements , and thus use formula (17) to construct the enterprise-scientific and technological achievement interaction result constraint:

[0074] (17)

[0075] In formula (17), represents the i-th enterprise Align interactive scientific and technological achievements The interaction results, represents the i-th enterprise Negative interactive scientific and technological achievements The interactive results represents the i-th enterprise right The interaction results, represents the i-th enterprise right The interaction results;

[0076] Step 3.3.2: Use formula (18) to construct the i-th enterprise BPR loss for all scientific and technological achievements : (18).

[0077] An electronic device of the present invention includes a memory and a processor, wherein the memory is used to store a program that supports the processor to execute the enterprise capability assessment method, and the processor is configured to execute the program stored in the memory.

[0078] The present invention provides a computer-readable storage medium, wherein a computer program is stored on the computer-readable storage medium, and the computer program executes the steps of the enterprise capability assessment method when executed by a processor.

[0079] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0080] 1. The present invention proposes a method for evaluating enterprise capabilities based on neurocognitive diagnosis technology. For the first time, factors such as enterprises, achievements, fields, and achievement levels are modeled in a neurocognitive diagnosis model to predict the completion of different achievements by different enterprises, thereby realizing the diagnosis of the degree of scientific and technological capabilities of different enterprises in different fields.

[0081] 2. The present invention constructs negative sample data in diagnostic tasks based on a rule-based approach; at the same time, it defines the partial order relationship between enterprise, outcome, and field, and proposes a partial order response sampling framework, which provides reliable auxiliary training signals for subsequent diagnostic models, thereby alleviating the negative impact of the long-tail problem and improving the accuracy of cognitive diagnosis of enterprise capabilities.

[0082] 3. The present invention also introduces an expected capability change weighted information sampling module, which adaptively selects samples from the perspective of enterprise capability changes, ensures the reliability of the training signal and enables the model to converge more quickly, and can more quickly diagnose the mastery of different enterprises in different fields. BRIEF DESCRIPTION OF THE DRAWINGS

[0083] Figure 1 is a flow chart of the overall method of the present invention;

[0084] Figure 2 shows the running results of the three algorithms on the cstad dataset divided into different groups;

[0085] Figure 3 shows the running results of the three algorithms on the magazine dataset divided into different groups. DETAILED DESCRIPTION

[0086] In this embodiment, Figure 1 As shown, a method for evaluating enterprise capabilities based on neurocognitive diagnosis technology is to use supervised deep learning methods to model various fine-grained influencing factors in the interaction between enterprises and scientific and technological achievements, that is, to use the existing enterprise-scientific and technological achievements interaction data and scientific and technological achievements level factors to model in a neurocognitive diagnosis model, and to use basic scientific and technological achievements interaction data for data expansion to construct interaction data for adaptive diagnosis; thereby, the ability of enterprises in different scientific and technological fields can be evaluated. Specifically, the method is carried out in the following steps:

[0087] Step 1: Define enterprise technology data and use the FastText model to divide the enterprise technology data into different fields to obtain the technology achievement field matrix. , and construct the total interaction set of enterprises and scientific and technological achievements after negative samples ;

[0088] Step 1.1: Define the number of enterprises in the technology data as The number of scientific and technological achievements is , the number of fields is , let the i-th enterprise, the m-th scientific and technological achievement and the p-th field be , and , , , ;

[0089] Let the i-th enterprise For the mth scientific and technological achievement The interaction result is recorded as ,and , Represents the total set of enterprise-scientific and technological achievements interactions.

[0090] Step 1.2: Use FastText classification model to Divide the fields and obtain the scientific and technological achievements field matrix ,make The mth scientific and technological achievement Belongs to the pth field The matrix of scientific and technological achievements is recorded as .

[0091] Step 1.3: Based on the total interaction set of enterprise-scientific and technological achievements Construct negative sample data for all enterprises;

[0092] Define the level of scientific and technological achievements as k, , C represents the maximum level;

[0093] Search the i-th enterprise according to the level k of scientific and technological achievements from high to low For all the scientific and technological achievements made, if there are corresponding scientific and technological achievements at the current level of scientific and technological achievements, no negative samples will be constructed; otherwise, Randomly select The negative samples of each non-interacting field are constructed respectively, so as to obtain the i-th enterprise Constructed negative samples, and ;in, The reverse order of the scientific and technological achievement level of the mth scientific and technological achievement, For the i-th enterprise The total number of all scientific and technological achievements made, To control the threshold of the number of negative samples for each enterprise, is the threshold for controlling the total number of negative samples for each enterprise.

[0094] Let the total number of scientific and technological achievements after constructing negative samples be , Scientific and technological achievements are recorded as , , the total interaction set of enterprise-scientific and technological achievements after constructing negative samples is , the total interaction set of enterprise-scientific and technological achievements The i-th enterprise For Scientific and technological achievements The interaction triplet of ,in, represents the i-th enterprise For Scientific and technological achievements The interaction result is , indicating the i-th enterprise To truly complete Scientific and technological achievements ,like , indicating the i-th enterprise For unfinished Scientific and technological achievements , the total interaction set of enterprise-scientific and technological achievements The i-th enterprise The interaction set of all scientific and technological achievements is recorded as ;

[0095] make Middle Scientific and technological achievements Belongs to the pth field The matrix of scientific and technological achievements is recorded as .

[0096] Step 2: Construct a basic neurocognitive diagnosis enterprise capability assessment learning module, including: initial embedding representation layer, enterprise capability diagnosis layer, and prediction layer, and conduct a matrix analysis of the scientific and technological achievements domain. The Scientific and technological achievements Belongs to the pth field Matrix of scientific and technological achievements Processing, get the i-th enterprise For the pth field No. Scientific and technological achievements The predicted interaction results , thereby constructing a basic neurocognitive diagnostic enterprise capability assessment network for loss .

[0097] Step 2.1: The initial embedding representation layer uses equations (1) and (2) to obtain the i-th enterprise Enterprise Capability Mastery Vector and Scientific and technological achievements Belongs to the pth field The result field matrix vector ;

[0098] (1)

[0099] (2)

[0100] In formula (1) and formula (2), represents the i-th enterprise The one-hot encoded vector of Indicates Scientific and technological achievements The one-hot encoded vector of ; Represents the matrix to be trained.

[0101] Step 2.2: The initial embedding representation layer constructs Scientific and technological achievements Technology metrics , the difficulty vector of scientific and technological achievements completion and Scientific and technological achievements The discrimination vector ;

[0102] Step 2.2.1. Scientific and technological achievements The level of scientific and technological achievements is recorded as , will Scientific and technological achievements exist The corresponding quantity is recorded as , and then use formula (3) to calculate the Scientific and technological achievements Technology metrics :

[0103] (3)

[0104] In formula (3), min-max represents normalization processing.

[0105] Step 2.2.2: Use equations (4) and (5) to calculate Scientific and technological achievements The completion difficulty vector and Scientific and technological achievements The discrimination vector :

[0106] (4)

[0107] (5)

[0108] In formula (4) and formula (5), Indicates Scientific and technological achievements The one-hot encoded vector of , Respectively represent two matrices to be trained.

[0109] Step 2.3, the enterprise capability diagnosis layer includes an interaction layer and two fully connected layers;

[0110] Step 2.3.1: The interaction layer uses formula (6) to establish the i-th enterprise The pth field completed No. Scientific and technological achievements The interaction function :

[0111] (6)

[0112] In formula (6), is the element-wise product.

[0113] Step 2.3.2: The fully connected layer uses equations (7) and (8) to obtain the i-th enterprise The pth field completed No. Scientific and technological achievements The feature representation And dimension reduction feature representation :

[0114] (7)

[0115] (8)

[0116] In formula (7)-formula (8), represents the i-th enterprise The pth field completed No. Scientific and technological achievements The characteristic representation of The i-th enterprise The pth field completed No. Scientific and technological achievements The characteristic representation of is the sigmoid activation function, , are the weight matrices to be trained for the two fully connected layers, , are the bias vectors of the two fully connected layers respectively.

[0117] Step 2.4: The prediction layer uses formula (9) to obtain the i-th enterprise For the p-th field No. Scientific and technological achievements The predicted interaction results :

[0118] (9)

[0119] In formula (9), is the weight matrix to be trained for the prediction layer, is the bias vector of the prediction layer;

[0120] Step 2.5: Use formula (10) to calculate the i-th enterprise Loss of all scientific and technological achievements :

[0121] (10)

[0122] In formula (10), represents the i-th enterprise For The completion status of scientific and technological achievements, represents the i-th enterprise For the p-th field No. Scientific and technological achievements prediction results.

[0123] Step 3: Construct a partial order sorting learning module, including: partial order learning module, expected capability change module, enterprise-scientific and technological achievement interaction result constraint design module, and BPR loss construction;

[0124] Step 3.1: Construct a partial order learning module and use partial order response sampling to sample the total interaction set of enterprises and scientific and technological achievements after negative samples. Perform partial order sampling to obtain the i-th enterprise A reliable set of candidate interactions ;

[0125] Step 3.1.1: From the enterprise-scientific and technological achievements interaction set after constructing negative samples Get the i-th enterprise For Scientific and technological achievements The interaction triplet , and use formula (11) to extract Scientific and technological achievements Similar results set :

[0126] (11)

[0127] In formula (11), It is a collection of scientific and technological achievements that have not been exchanged among enterprises. is the uth non-interactive scientific and technological achievement, For the non-interactive scientific and technological achievements of item u Related domain sets, For the Technical achievements A set of related fields.

[0128] Step 3.1.2: Use formula (12) to construct the i-th enterprise For item u of non-interactive scientific and technological achievements The interaction triplet :

[0129] (12)

[0130] In formula (12), For the i-th enterprise For item u of non-interactive scientific and technological achievements The interaction results;

[0131] Step 3.1.3: Use formula (13) to calculate the i-th enterprise For Interactive scientific and technological achievements Non-interactive scientific and technological achievements with item u and as a basis for constructing the i-th enterprise The sampling probability of non-interactive scientific and technological achievements :

[0132] (13)

[0133] In formula (13), represents the i-th enterprise output by the neurocognitive diagnosis enterprise capability assessment network For Scientific and technological achievements The characteristic factor of Indicates the uth non-interactive scientific and technological achievement The characteristic factor of Indicates that it belongs to a similar results set The jth scientific and technological achievement in Characteristic factor of , T represents transposition;

[0134] from Select the topk non-interactive scientific and technological achievements with higher sampling probability as the i-th enterprise The sampling scientific and technological achievements are recorded as .

[0135] Step 3.1.4: Use formula (14) to get the i-th enterprise A reliable candidate interaction set :

[0136] (14)

[0137] In formula (14), represents the i-th enterprise For item u of non-interactive scientific and technological achievements The interaction triplet of .

[0138] Step 3.2: Construct an expected capability change module to calculate reliable candidate interaction sets The i-th enterprise For item u of non-interactive scientific and technological achievements The expected capability change of the interaction triplet ,and After normalization, we get the i-th enterprise For item u of non-interactive scientific and technological achievements The expected capacity change weight .

[0139] Step 3.2.1: Use formula (15) to get the i-th enterprise In the Uninteractive scientific and technological achievements Expected capacity change :

[0140] (15)

[0141] In formula (15), represents the i-th enterprise In the Uninteractive scientific and technological achievements The expected change in capability of is given by using the diagnostic interaction function, represents the i-th enterprise exist The level of mastery of various fields, represents the i-th enterprise exist Plus the uth non-interactive scientific and technological achievements Then the level of mastery in each field.

[0142] Step 3.2.2: Calculate the uth non-interactive scientific and technological achievement using formula (16): The normalized expected capacity change weight :

[0143] (16)

[0144] In formula (16), Represents a reliable candidate interaction set The number of interactive scientific and technological achievements in the

[0145] Step 3.3, design the constraints of enterprise-scientific and technological achievements interaction results and construct the BPR loss function;

[0146] For the i-th enterprise , according to the different interaction results, the interaction set after constructing negative samples will be Divided into positive interaction result sets after constructing negative samples And the negative interaction result set after constructing negative samples ;

[0147] Reliable candidate interaction set The positive interaction result set is divided into reliable candidate sets and the negative interaction result set in the reliable candidate set .

[0148] Step 3.3.1. From , , , Take any positive interaction scientific and technological achievements , any candidate positive interactive scientific and technological achievements , any negative interactive scientific and technological achievements , any candidate negative interaction scientific and technological achievements , and thus use formula (17) to construct the enterprise-scientific and technological achievement interaction result constraint:

[0149] (17)

[0150] In formula (17), represents the i-th enterprise Align interactive scientific and technological achievements The interaction results, represents the i-th enterprise Negative interactive scientific and technological achievements The interactive results represents the i-th enterprise right The interaction results, represents the i-th enterprise right The interaction results.

[0151] Step 3.3.2: Use formula (18) to construct the i-th enterprise BPR loss for all scientific and technological achievements :

[0152] (18)

[0153] Step 4: Training of the neurocognitive diagnostic enterprise capability assessment network;

[0154] Step 4.1: Use formula (19) to construct the i-th enterprise The total loss function for all scientific and technological achievements :

[0155] (19)

[0156] In formula (19), is a hyperparameter to balance the two losses.

[0157] Step 4.2: Use the Adam optimizer to train the neurocognitive diagnosis enterprise capability assessment network composed of the basic neurocognitive diagnosis enterprise capability assessment learning module and the partial order sorting learning module, and minimize until Until convergence, the optimal neurocognitive diagnosis enterprise capability assessment network is obtained, which is used to diagnose the enterprise's mastery of different fields, thereby obtaining the enterprise's capability assessment results in different fields.

[0158] In this embodiment, an electronic device includes a memory and a processor, wherein the memory is used to store a program that supports the processor to execute the above method, and the processor is configured to execute the program stored in the memory.

[0159] In this embodiment, a computer-readable storage medium stores a computer program on the computer-readable storage medium, and the computer program executes the steps of the above method when executed by a processor.

[0160] Example:

[0161] In order to verify the effectiveness of this method, the experimental data sets of the present invention come from four types of scientific and technological data provided by the Qinghai Provincial Science and Technology Department, namely project (project data), magazine (journal data), patent (patent data), and cstad (achievement data).

[0162] Since the real capability level of an enterprise is unknown, it is extremely difficult to directly evaluate the capability performance of a CDM. In cognitive diagnosis in the field of education, a reasonable solution is proposed, that is, to obtain diagnostic results through the learner's academic performance prediction task, so as to measure the student's academic performance by using the prediction score in the diagnostic model. Here, the evaluation method in the field of education is used, and AUC (area under the curve), ACC (accuracy) and RMSE (root mean square error) are used as evaluation criteria. The present invention selected 7 methods for comparison, namely SVM (Support Vector Machine, SVM), KNN (K-Nearest Neighbors, KNN), DT (Decision Tree, DT), RF (Random Forest, RF), NCDM (Wang, Fei, et al. "Neural cognitive diagnosis for intelligent educationsystems." Proceedings of the AAAI conference on artificial intelligence. Vol.34. No. 04. 2020), EIRS (Yao, Fangzhou, et al. "Exploiting non-interactiveexercises in cognitive diagnosis." Interaction 100.200 (2023):300), TNCDM (an enterprise capability diagnosis method based on NCDM), and compared the performance of long-tail data and overall data with the baseline method. The long-tail data extracted from the test data includes the response records of enterprises with less than 10 enterprise-outcome interactions. Specifically, Table 1 shows the experimental results in the above data sets, and it can be observed that the present invention is superior to the comparison method in indicators such as ACC, AUC and RMSE.

[0163] Table 1. Results of multiple algorithms on patent and project datasets

[0164]

[0165] Table 2. Running results of multiple algorithms on magazine and cstad datasets

[0166]

[0167] In order to further verify the diagnostic effect of this method on long-tail data, it is further divided into four groups according to the number of interactions to verify the prediction of different groups. The results are as follows Figure 2 , Figure 3 shown.

[0168] It can be seen that the model has achieved good results in all groups, especially in the enterprises with fewer interactions (5-15, 15-25), the prediction results are significantly improved compared with the baseline model. Therefore, it can be concluded that the contribution of the present invention in diagnosing data with fewer interactions in enterprises is relatively outstanding.

[0169] Here, this model is used to predict the completion of scientific and technological achievements of enterprises, and the enterprise's ability mastery vector can be obtained , is an L-dimensional vector, and the Lth component of the vector represents the capability value of the enterprise in the lth field. Since experiments were conducted in four scientific and technological achievements, four capability vectors can be obtained. According to the weight division of Qinghai's scientific and technological achievements and the different importance of different scientific research achievements to enterprises, the weighted sum is calculated according to the weight of project: achievement: patent: paper = 4:2:1:1 to obtain the mastery of the enterprise in all fields. According to the field division, the TOPK enterprises with the highest mastery in the field are recommended as the field prediction results.

[0170] In order to verify the accuracy of this model in evaluating and recommending science and technology enterprises, we compare the data of all Qinghai units and scientific research results with the actual recommendation results fed back by the Qinghai Provincial Department of Science and Technology. The evaluation indicators used are Precision@K (precision), Recall@K (recall rate) and F-score@K (F score).

[0171] The comparison results are shown in Table 3, where TNCD is a recent talent evaluation model and EIRS is a diagnostic model in the field of education. The results of the three indicators are compared in two cases where the field recommendation length is 5 and 8. It can be found that this model outperforms the recent talent evaluation model TNCD and the EIRS model in the field of education in terms of overall performance. This is because this model considers the quality and quantity factors of the results in the modeling of the difficulty value and the discrimination value. At the same time, we also fully consider the long-tail problem of enterprise interaction in the modeling process, that is, we adopt a partial order response sampling scheme, which makes the overall model more robust and closer to the real recommendation needs of the field.

[0172] Table 3. Comparison of recommendation results in multiple algorithm fields with expert annotations

[0173]

[0174] In summary, the methods proposed in the present invention are significantly better than many of the compared methods, thus proving the feasibility of the methods proposed in the present invention. By considering the quality and quantity factors of scientific and technological achievements, they are embedded into the model as the initial learning parameters of the model, and the achievement-aware sampling method is used to alleviate the long-tail problem of the data, making the model more effective and robust. Experiments were conducted on 4 scientific and technological data sets in Qinghai. The effectiveness of this model was verified by comparing it with the baseline model, and the robustness of this model to long-tail data was verified by long-tail experiments. At the same time, the results obtained based on capability diagnosis also verified the effectiveness of this model in capability diagnosis through example results.

Claims

1. A method for evaluating enterprise capabilities based on neurocognitive diagnostic technology, characterized in that: Follow these steps: Step 1: Define enterprise technology data and use the FastText model to divide the enterprise technology data into different fields to obtain the technology achievement field matrix. , and construct the total interaction set of enterprises and scientific and technological achievements after negative samples ; Step 2: Construct a basic neurocognitive diagnosis enterprise capability assessment learning module, including: initial embedding representation layer, enterprise capability diagnosis layer, and prediction layer, and then perform a matrix of scientific and technological achievements on the matrix. The Scientific and technological achievements Belongs to the pth field Matrix of scientific and technological achievements Processing, get the i-th enterprise For the pth field No. Scientific and technological achievements The predicted interaction results , thereby constructing a basic neurocognitive diagnostic enterprise capability assessment network for loss ; Step 3: Construct a partial order sorting learning module, including: partial order learning module, expected capability change module, and enterprise-scientific and technological achievement interaction result constraint design module; Step 3.1: Construct a partial order learning module and use partial order response sampling to sample the total interaction set of enterprises and scientific and technological achievements after negative samples. Perform partial order sampling to obtain the i-th enterprise A reliable set of candidate interactions ; Step 3.2: Construct an expected capability change module to calculate reliable candidate interaction sets The i-th enterprise For item u of non-interactive scientific and technological achievements The expected capability change of the interaction triplet ,and After normalization, we get the i-th enterprise For item u of non-interactive scientific and technological achievements The expected capacity change weight ; Step 3.3: Design the constraints of enterprise-scientific and technological achievements interaction results and construct the i-th enterprise BPR loss function for all scientific and technological achievements ; Step 4: Use formula (19) to construct the i-th enterprise The total loss function for all scientific and technological achievements : (19) In formula (19), Hyperparameters to balance the two losses; Step 5: Use the Adam optimizer to train the neurocognitive diagnosis enterprise capability assessment network composed of the basic neurocognitive diagnosis enterprise capability assessment learning module and the partial order sorting learning module, and minimize until Until convergence, the optimal neurocognitive diagnosis enterprise capability assessment network is obtained, which is used to diagnose the enterprise's mastery of different fields, thereby obtaining the enterprise's capability assessment results in different fields.

2. According to claim 1, a method for enterprise capability assessment based on neurocognitive diagnosis technology is characterized in that: The step 1 is performed according to the following steps: Step 1.1: Define the number of enterprises in the technology data as The number of scientific and technological achievements is , the number of fields is , let the i-th enterprise, the m-th scientific and technological achievement and the p-th field be , and , , , ; Let the i-th enterprise For the mth scientific and technological achievement The interaction result is recorded as ,and , represents the total set of enterprise-scientific and technological achievements interactions; Step 1.2: Use FastText classification model to Divide the fields and obtain the scientific and technological achievements field matrix ,make The mth scientific and technological achievement Belongs to the pth field The matrix of scientific and technological achievements is recorded as ; Step 1.3: Based on the total interaction set of enterprise-scientific and technological achievements Construct negative sample data for all enterprises; Define the level of scientific and technological achievements as k, , C represents the maximum level; Search the i-th enterprise according to the level k of scientific and technological achievements from high to low For all the scientific and technological achievements made, if there are corresponding scientific and technological achievements at the current level of scientific and technological achievements, no negative samples will be constructed; otherwise, Randomly select The negative samples of each non-interacting field are constructed respectively, so as to obtain the i-th enterprise Constructed negative samples, and ;in, The reverse order of the scientific and technological achievement level of the mth scientific and technological achievement, For the i-th enterprise The total number of all scientific and technological achievements made, To control the threshold of the number of negative samples for each enterprise, To control the threshold of the total number of negative samples for each enterprise; Let the total number of scientific and technological achievements after constructing negative samples be , Scientific and technological achievements are recorded as , , the total interaction set of enterprise-scientific and technological achievements after constructing negative samples is , the total interaction set of enterprise-scientific and technological achievements The i-th enterprise For Scientific and technological achievements The interaction triplet of ,in, represents the i-th enterprise For Scientific and technological achievements The interaction result is , indicating the i-th enterprise To truly complete Scientific and technological achievements ,like , indicating the i-th enterprise For unfinished Scientific and technological achievements , the total interaction set of enterprise-scientific and technological achievements The i-th enterprise The interaction set of all scientific and technological achievements is recorded as ; make Middle Scientific and technological achievements Belongs to the pth field The matrix of scientific and technological achievements is recorded as .

3. The enterprise capability assessment method based on neurocognitive diagnosis technology according to claim 2 is characterized in that: The step 2 is performed according to the following steps: Step 2.1: The initial embedding representation layer uses equations (1) and (2) to obtain the i-th enterprise Enterprise Capability Mastery Vector and Scientific and technological achievements Belongs to the pth field The result field matrix vector ; (1) (2) In formula (1) and formula (2), represents the i-th enterprise The one-hot encoded vector of Indicates Scientific and technological achievements The one-hot encoded vector of ; Represents the matrix to be trained; Step 2.2: The initial embedding representation layer constructs Scientific and technological achievements Technology metrics , the difficulty vector of scientific and technological achievements completion and Scientific and technological achievements The discrimination vector ; Step 2.3: The enterprise capability diagnosis layer includes an interaction layer and two fully connected layers, and , as well as Processing, get the i-th enterprise The pth field completed No. Scientific and technological achievements The feature representation And dimension reduction feature representation ; Step 2.4: The prediction layer uses formula (9) to obtain the i-th enterprise For the p-th field No. Scientific and technological achievements The predicted interaction results : (9) In formula (9), is the weight matrix to be trained for the prediction layer, is the bias vector of the prediction layer; Step 2.5: Use formula (10) to calculate the i-th enterprise Loss of all scientific and technological achievements : (10) In formula (10), represents the i-th enterprise For The completion status of scientific and technological achievements, represents the i-th enterprise For the p-th field No. Scientific and technological achievements prediction results.

4. The enterprise capability assessment method based on neurocognitive diagnosis technology according to claim 3 is characterized in that: The step 2.2 is performed according to the following steps: Step 2.2.

1. Scientific and technological achievements The level of scientific and technological achievements is recorded as , will Scientific and technological achievements exist The corresponding quantity is recorded as , and then use formula (3) to calculate the Scientific and technological achievements Technology metrics : (3) In formula (3), min-max represents normalization processing; Step 2.2.2: Use equations (4) and (5) to calculate Scientific and technological achievements The completion difficulty vector and Scientific and technological achievements The discrimination vector : (4) (5) In formula (4) and formula (5), Indicates Scientific and technological achievements The one-hot encoded vector of , Respectively represent two matrices to be trained.

5. The enterprise capability assessment method based on neurocognitive diagnosis technology according to claim 4 is characterized in that: The step 2.3 is performed according to the following steps: Step 2.3.1: The interaction layer uses formula (6) to establish the i-th enterprise The pth field completed No. Scientific and technological achievements The interaction function : (6) In formula (6), is the element-wise product; Step 2.3.2: The fully connected layer uses equations (7) and (8) to obtain the i-th enterprise The pth field completed No. Scientific and technological achievements The feature representation And dimension reduction feature representation : (7) (8) In formula (7)-formula (8), represents the i-th enterprise The pth field completed No. Scientific and technological achievements The characteristic representation of The i-th enterprise The pth field completed No. Scientific and technological achievements The characteristic representation of is the sigmoid activation function, , are the weight matrices to be trained for the two fully connected layers, , are the bias vectors of the two fully connected layers respectively.

6. The enterprise capability assessment method based on neurocognitive diagnosis technology according to claim 5 is characterized in that: The step 3.1 is performed according to the following steps: Step 3.1.1: From the enterprise-scientific and technological achievements interaction set after constructing negative samples Get the i-th enterprise For Scientific and technological achievements The interaction triplet , and use formula (11) to extract Scientific and technological achievements Similar results set : (11) In formula (11), It is a collection of scientific and technological achievements that have not been exchanged among enterprises. is the uth non-interactive scientific and technological achievement, For the non-interactive scientific and technological achievements of item u Related domain sets, For the Technical achievements Related set of fields; Step 3.1.2: Use formula (12) to construct the i-th enterprise For item u of non-interactive scientific and technological achievements The interaction triplet : (12) In formula (12), For the i-th enterprise For item u of non-interactive scientific and technological achievements The interaction results; Step 3.1.3: Use formula (14) to get the i-th enterprise A reliable candidate interaction set : (14) In formula (14), for The number of uninteracted samples in , represents the i-th enterprise For item u of non-interactive scientific and technological achievements The interaction triplet of .

7. The enterprise capability assessment method based on neurocognitive diagnosis technology according to claim 6 is characterized in that: The step 3.2 is performed according to the following steps: Step 3.2.1: Use formula (15) to get the i-th enterprise In the Uninteractive scientific and technological achievements Expected capacity change : (15) In formula (15), represents the i-th enterprise In the Uninteractive scientific and technological achievements The expected change in capability of is given by using the diagnostic interaction function, represents the i-th enterprise exist The level of mastery of various fields, represents the i-th enterprise exist Plus the uth non-interactive scientific and technological achievements The subsequent mastery of each field; Step 3.2.2: Calculate the uth non-interactive scientific and technological achievement using formula (16): The normalized expected capacity change weight : (16) In formula (16), Represents a reliable candidate interaction set The number of interactive scientific and technological achievements in the 8. The enterprise capability assessment method based on neurocognitive diagnosis technology according to claim 7 is characterized in that: The step 3..3 is performed according to the following steps: Step 3.3.1: For the i-th enterprise , according to the different interaction results, the interaction set after constructing negative samples will be Divided into positive interaction result sets after constructing negative samples And the negative interaction result set after constructing negative samples ; Reliable candidate interaction set The positive interaction result set is divided into reliable candidate sets and the negative interaction result set in the reliable candidate set ; from , , , Take any positive interaction scientific and technological achievements , any candidate positive interactive scientific and technological achievements , any negative interactive scientific and technological achievements , any candidate negative interaction scientific and technological achievements , and thus use formula (17) to construct the enterprise-scientific and technological achievement interaction result constraint: (17) In formula (17), represents the i-th enterprise Align interactive scientific and technological achievements The interaction results, represents the i-th enterprise Negative interactive scientific and technological achievements The interactive results represents the i-th enterprise right The interaction results, represents the i-th enterprise right The interaction results; Step 3.3.2: Use formula (18) to construct the i-th enterprise BPR loss for all scientific and technological achievements : (18)。 9. An electronic device, comprising a memory and a processor, characterized in that: The memory is used to store a program that supports the processor to execute the enterprise capability assessment method described in any one of claims 1 to 8, and the processor is configured to execute the program stored in the memory.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the enterprise capability assessment method according to any one of claims 1 to 8 are executed.

Images