Personnel ability automatic evaluation method and system based on attention neural network

Abstract

The invention discloses an attention neural network-based personnel ability automatic evaluation method and system. The method comprises the following steps of: dividing an ability evaluation data set # imgabs0 # into a training set # imgabs1 #, a verification set # imgabs2 # and a test set # imgabs3 #; setting a plurality of different neural network configurations to form a configuration set # imgabs4 #; the obtained hyper-parameter # imgabs5 # is selected, and a neural network model # imgabs6 # is initialized; using a training set # imgabs7 # to train a neural network model # imgabs8 # obtained through initialization; after a # imgabs9 # round is carried out, the neural network model trained in the # imgabs10 # round is recorded as # imgabs11 #, a verification set # imgabs12 # is used for evaluating the trained model, and the neural network model and the evaluation result after each round of training are stored; the model obtaining the highest evaluation index on the verification set in N rounds is selected to be tested and stored on a test set # imgabs13 #; and until all hyper-parameters are used for initializing the neural network models and tested by a test set # imgabs14 #, deploying the neural network model with the highest model evaluation result for personnel ability grade evaluation work. And the requirement on multi-dimensional personnel ability portraits in a personnel evaluation scene is met.

Description

Background Art

[0002] In personnel management and task allocation, it is necessary to reasonably manage personnel and allocate work tasks according to the different ability levels and status of personnel, which requires the characterization of various abilities of personnel. Since the various abilities of personnel change dynamically over time, it is often necessary to re-evaluate the various abilities of personnel at regular intervals. In the existing personnel data information management system, the evaluation of personnel abilities requires managers to re-analyze the static information and recent performance of personnel, which brings considerable manpower costs.

[0003] In order to solve the above problems faced by the personnel information management system, it is necessary to design a model that can mine the internal knowledge of huge personnel information and summarize the internal rules of evaluation information data. It can summarize personnel information and recent performance according to the mined rules, and evaluate the ability level of various personnel capabilities, so as to achieve the effect of automatically updating the personnel capability portrait. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a method and system for automatic personnel capability assessment based on an attention neural network in response to the deficiencies in the above-mentioned prior art, which is used to solve the technical problem that the current personnel data management system cannot utilize historical personnel evaluation information and requires frequent re-evaluation of personnel capabilities, which brings heavy manpower costs. It improves the utilization rate of various personnel information in the personnel data system, can use historical personnel capability evaluation information that has not been utilized by the original system, summarize evaluation rules, and generate personnel capability assessments based on personnel static information and recent performance, thereby saving management personnel costs.

[0005] The present invention adopts the following technical solutions: A method for automatically evaluating personnel capabilities based on an attention neural network comprises the following steps: S1. Collect static feature sets of personnel , Dynamic feature type collection , dynamic eigenvalue set , and the capability type set and capability level set ; S2. Collect historical personnel characteristics and corresponding capability evaluation data sets , each data in the data set is Organization in the form of are static competence-related characteristics of personnel whose importance does not change over time, and is the personnel dynamic feature associated with importance and time, The rating results for each capability type of the person; the capability evaluation data set Split into training set , validation set , test set ; S3. Set up multiple different neural network configurations Composition configuration collection , each configuration can initialize different neural networks and training methods; S4. Select the hyperparameters obtained in step S3 , initialize the neural network model ; S5. Use the training set obtained in step S2 The neural network model obtained by initialization in training step S4 ; S6, step S5 is carried out Wheel, remember The neural network model after rounds of training is , and use the validation set Evaluate the trained model and save the neural network model and evaluation results after each round of training. After N rounds of training, select the model with the highest evaluation index on the validation set in the round and save it on the test set. Conduct tests on the system and save the test evaluation results; S7, repeat steps S4 to S6 until all hyperparameters are used to initialize the neural network model and the test set obtained in step S2 Test and deploy the neural network model with the highest f1 score in the model evaluation results for personnel capability level assessment.

[0006] Preferably, in step S3, for any , contains two parts of configuration , Configuration for model hyperparameters, including embedding size , capability type mapping network layer number , feature type mapping network layers , the number of eigenvalue mapping network layers , the number of network layers for extracting features of personnel capability classification ; Configuration of model training parameters, including learning rate , weight decay regularization weight , type vector dot product regularization term Weight , embedding vector distance regularization term Weight .

[0007] Preferably, in step S4, the neural network model Contains embedded modules , Capability Embedding Mapping Module , cross-weight feature fusion module and classification module , and initialize them respectively; Xavier initialization method is used for The activation function is performed, the elements in the parameters are sampled from a normal distribution with a fixed standard deviation, and the He initialization method is used for Activation function.

[0008] Preferably, the embedded module is initialized When, according to the collected static feature set , Dynamic feature type collection , dynamic eigenvalue set , Capability Type Set and embedding size , use the He initialization method to randomly initialize the static feature type embedding matrix , dynamic feature type embedding matrix , dynamic feature optional value embedding matrix , and randomly initialize the dynamic feature type encoding vector , dynamic eigenvalue encoding vector , static feature type encoding vector , static eigenvalue encoding vector , use Xavier initialization method to initialize the capability type embedding matrix ; Initialize capability mapping module When mapping the number of network layers according to the capability type , using Xavier initialization method to randomly initialize the composition The linear layer matrix parameters in each layer of the bit-wise feed-forward network are With offset parameters , where the parameters of each layer are , , ; Initialize the cross-weight feature fusion module When using the He initialization method, according to the hyperparameters and Initialize the feature type mapping network contained therein respectively and eigenvalue mapping network The matrix parameters of the linear layer in each layer of the bit-wise feed-forward network and With offset parameters and , the parameters of each layer in the two networks , , , ,in , , and initialization Medium time decay layer Parameters is an experience value; Initialize the classification module When, according to the parameters Using the He initialization method, the linear layer matrix parameters of each layer of the bit-by-bit feedforward network are randomly initialized. and the linear layer offset parameter , where for any , , , and initialized randomly using the He initialization method; according to the level set initialization The final fully connected classification layer parameters , .

[0009] Preferably, the Xavier initialization method is used to initialize The activation function is as follows:

[0010] Use He initialization method for The activation function is as follows:

[0011] in, is a Gaussian random distribution, is the standard deviation of the Gaussian random distribution, Enter the eigenvector dimensions for the matrix, Output eigenvector dimensions for the matrix.

[0012] Preferably, step S5 is specifically: The randomly selected samples Input neural network model , the embedding layer will , The feature types in the sample are processed to obtain the static feature type embedding vector sequence of the sample personnel. and dynamic feature type embedding vector sequence , and As the target capability input to the embedding layer to obtain the target capability embedding sequence ; The capability embedding sequence is input into the capability type mapping network Get the capability embedding sequence after feature extraction ; For the sequence and The embedding layer uses vector addition to add the encoding vector to the type and value variables respectively , , , , then complete The length of historical events for each feature in ; Then the two are spliced ​​and merged into a unified personnel capability feature sequence, and the content in the split tuple is divided into three sequences - feature type embedding sequence , the eigenvalue embedding sequence and characteristic time series Input to submodule Processing and aggregation in Feature type embedding sequence Through the module Perform feature transformation to obtain a new feature type sequence ; Personnel characteristic value sequence Through the module Perform feature transformation to obtain a new feature value sequence ; Characteristic time value series After the time decay layer, the time weight sequence is obtained The time decay layer uses a time parameter The time decay function Perform calculations; What you get , , pass Module according to Each capability type in is aggregated separately; Middle Vector , using scaled click attention to calculate each feature in the person feature sequence relative to The attention weight sequence; By crossing the time weight sequence With attention weight The sequence gets the final eigenvalue weight sequence , No. The weight of the feature value ;use Embed the input feature value into a sequence Perform weighted average aggregation to obtain the personnel's ability type Temporal and attentional convergence representation under ; Pair Sequence After each capability type is embedded and aggregated, the personnel capability representation sequence under each target type is obtained. ; The final classification results of each personnel's ability are obtained through sub-modules right Perform feature extraction and classification to obtain the ability level sequence of different ability types of personnel ; Submodule First, the input sequence Each embedding is passed through the feature extraction network to obtain a linearly separable feature representation ; Eventually Each element in The final fully connected layer performs linear classification to obtain the classification result sequence ; During training, the neural network model The fully connected layer output result The prediction vectors are calculated one by one, and the neural network model is optimized using back-propagation gradient descent. Parameters; With two biased regularization terms and The loss function After the training set is sampled, use the validation set and validation indicator evaluation performance, and save it.

[0013] Preferably, Contains multiple layers of the same structure using residual connections and As the activation function, the bitwise feed-forward network embeds the input capability type , No. The operation of the layer network is equivalent to the function as follows:

[0014] Multiple layers use residual connections and The bit-by-bit feed-forward network of the activation function is composed of Layer network for feature type sequence Mean vector The operation process as follows:

[0015] Using residual connections and The bit-by-bit feed-forward network of the activation function is composed of Layer network sequence Mean vector Feature extraction operation as follows:

[0016] in, is the capability type feature mapping matrix, is the capability type feature mapping offset, is the capability type feature vector, is the personnel feature mapping matrix, Map offsets for person feature types, is the characteristic vector of personnel feature type, is the comprehensive feature mapping matrix of personnel, Offset for the comprehensive feature map of personnel, is the comprehensive feature vector of personnel.

[0017] Preferably, two biased regularization terms and The loss function as follows:

[0018] in, represents the correct predicted rank subscript, is the model’s predicted probability for different levels, is the ability type regularization loss, is the regularization loss of personnel feature type, is the capability type regularization term ratio hyperparameter, is the regularization term ratio hyperparameter of the personnel feature type.

[0019] Preferably, in step S6, from each round of training Select the one with the highest f1 score among the evaluation indicators As a hyperparameter The best neural network model obtained , and put it in the test set The same evaluation index is used to test and save the neural network model and model evaluation results.

[0020] In a second aspect, an embodiment of the present invention provides a personnel capability automatic assessment system based on an attention neural network, characterized in that it includes: Collection module, collects static feature sets of personnel , Dynamic feature type collection , dynamic eigenvalue set , and the capability type set and capability level set ; Collect historical personnel characteristics and corresponding capability evaluation data sets , each data in the data set is Organization in the form of are static competence-related characteristics of personnel whose importance does not change over time, and is the personnel dynamic feature associated with importance and time, The rating results for each capability type of the person; the capability evaluation data set Split into training set , validation set , test set ; Setting module, setting multiple different neural network configurations Composition configuration collection , each configuration can initialize different neural networks and training methods; the hyperparameters selected , initialize the neural network model ; Training module, using the obtained training set The neural network model obtained by training initialization ; Test module, the training module Wheel, remember The neural network model after rounds of training is , and use the validation set Evaluate the trained model and save the neural network model and evaluation results after each round of training. After N rounds of training, select the model with the highest evaluation index on the validation set in the round and save it on the test set. Conduct tests on the system and save the test evaluation results; Evaluation module, repeatedly set the module to the test module until all hyperparameters are used to initialize the neural network model and the test set is obtained Test and deploy the neural network model with the highest f1 score in the model evaluation results for personnel capability level assessment.

[0021] In a third aspect, a computer device comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned method for automatic assessment of personnel capabilities based on an attention neural network when executing the computer program.

[0022] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, including a computer program, which, when executed by a processor, implements the steps of the above-mentioned method for automatic assessment of personnel capabilities based on an attention neural network.

[0023] Compared with the prior art, the present invention has at least the following beneficial effects: An automatic personnel capability assessment method based on attention neural network automatically summarizes personnel capability ratings according to personnel data information, reducing the burden of personnel assessment in personnel management; the designed neural network fully considers the attributes of different personnel information, and adopts different feature representation methods for personnel static information and dynamic information respectively; considering that personnel features have different importance according to their types when evaluating personnel capabilities, a reasonable personnel capability feature fusion mechanism is designed to ensure the validity and interpretability of the obtained personnel representation; the designed training method takes into account that type representations should be sparsely distributed in space to avoid attention confusion problems to prevent overfitting of model training; the designed neural network and training method can fully explore the laws of historical personnel information, realize automatic determination of the capability levels corresponding to different capability types of personnel, and reduce labor costs.

[0024] Furthermore, the static features of personnel collected in step S1 refer to features that are not related to time, such as the certificates possessed by personnel, which do not change over time; while the dynamic features of personnel refer to features whose importance becomes less important as time goes by, such as personnel's work records. The collected competence levels range from low to high, and the lowest level is used to indicate that there is no such type of competence. By considering static features and dynamic features separately, and setting competence levels, the model can more reasonably consider the characteristics of personnel and judge whether personnel have competence and proficiency.

[0025] Furthermore, in step S4, the submodules of the model use Activation function. This is because the ability type embeddings are all input to the network during training. Each ability type embedding will be optimized during each gradient descent optimization process. There is a high optimization frequency, so there is no need to worry about submodules. The parameter convergence problem of The activation function can also prevent the ability type embedding from being optimized too much during the optimization process, resulting in gradient explosion. Use in The activation function is because people often only have a part of the feature types, which means that the optimization frequency of this part of the embedding is low. The activation function has a faster convergence to ensure that this part of the embedding can also fully converge. The Xavier initialization method can ensure that the The activation function has a good propagation effect at the beginning of training, and the He initialization method has a good propagation effect at the beginning of training. The activation function part has the same effect.

[0026] Furthermore, the two static and dynamic feature type codes initialized in step S4 are , and two static dynamic eigenvalue codes , It is used to uniformly identify the differences between static and dynamic feature types and feature values, so that the model can obtain information from both for consideration when performing feature transformation. At the same time, the use of type coding instead of splicing identification can avoid the expansion of model parameters and overfitting of type coding information.

[0027] Furthermore, the attention weight sequence calculated by the attention mechanism adopted by the model in step S5 is The attention weights in the attention mechanism represent the different degrees of importance that the model attaches to the characteristics of various types of personnel under this ability type. Compared with the long short-term memory network LSTM and the convolutional network GCN, the attention mechanism directly considers the influence of the ability type when fusion, which is relatively more reasonable. The attention weights calculated by the attention mechanism can explain the emphasis of the model when evaluating personnel capabilities, and have better interpretability. In addition, the calculation process can be performed in parallel, and the time complexity of training and reasoning is better than that of the LSTM network and the GCN network; and the time decay layer Parameters The empirical value used in the initialization allows the model to calculate a more reasonable time weight from the beginning, and gradually optimize the parameters according to the data performance during the gradient descent process. to make it perform better.

[0028] Furthermore, the regularization term of the loss function used in step S5 is Used to constrain the model calculation The representation vectors and The representation vectors of different feature types in . When training and optimizing the gradient calculation, for The calculation part of the term, when hour, Regularity Constraints and have smaller dot product results between them to prevent them from having similar dot product properties in attention calculation to prevent the model from confusing different ability types; when hour, , which is the bi-norm of the capability type representation, can constrain the size of the capability type representation to prevent the capability type vector from being too long. For item The effect of the regularization term is similar.

[0029] Furthermore, the regularization term of the loss function used in step S5 is It is used to control the distance between the embedding layer type vectors to ensure the distance between different capability types and feature types, so as to prevent the model from confusing different capability types and feature types during the embedding stage. In the regular term Part, when hour, for and The previous distance is negative, and the back propagation optimization will expand the distance between the two embeddings so that the representations of the two ability types are different at the initial stage. The above effect is for and Partly the same.

[0030] It can be understood that the beneficial effects of the second aspect mentioned above can be found in the relevant description of the first aspect mentioned above, and will not be repeated here.

[0031] In summary, the personnel capability assessment model designed by the present invention fully considers the characteristics related to personnel capability assessment, and reasonably integrates personnel characteristics through the designed neural network adopting attention and time decay mechanism to obtain the capability level of each capability type of personnel.

[0032] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 The overall flow chart for the application and deployment of the model of the present invention; Figure 2 A structural diagram of a neural network model designed for the present invention; Figure 3 A schematic diagram of a computer device provided by an embodiment of the present invention; Figure 4 The present invention is a block diagram of an electronic device provided according to an embodiment. DETAILED DESCRIPTION

[0034] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0035] In the description of the present invention, it should be understood that the terms “include” and “comprises” indicate the presence of described features, wholes, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or collections thereof.

[0036] It should also be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0037] It should be further understood that the term "and / or" used in the present specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes these combinations. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in the present invention generally indicates that the associated objects are in an "or" relationship.

[0038] It should be understood that, although the terms first, second, third, etc. may be used to describe preset ranges, etc. in the embodiments of the present invention, these preset ranges should not be limited to these terms. These terms are only used to distinguish preset ranges from each other. For example, without departing from the scope of the embodiments of the present invention, the first preset range may also be referred to as the second preset range, and similarly, the second preset range may also be referred to as the first preset range.

[0039] The word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining" or "in response to detecting", depending on the context. Similarly, the phrases "if it is determined" or "if (stated condition or event) is detected" may be interpreted as "when it is determined" or "in response to determining" or "when detecting (stated condition or event)" or "in response to detecting (stated condition or event)", depending on the context.

[0040] Various structural schematic diagrams of the embodiments disclosed in the present invention are shown in the accompanying drawings. These figures are not drawn to scale, and some details are magnified and some details may be omitted for the purpose of clear expression. The shapes of various regions and layers shown in the figures and the relative sizes and positional relationships therebetween are only exemplary, and may deviate in practice due to manufacturing tolerances or technical limitations, and those skilled in the art may additionally design regions / layers with different shapes, sizes, and relative positions according to actual needs.

[0041] The present invention provides an automatic personnel capability assessment method based on an attention neural network, which can realize automatic personnel capability portrait generation by integrating personnel static features and dynamic features, avoiding the problems of slow update, outdated analysis results and large manpower loss in traditional personnel personnel processing methods, and meeting the needs of multi-dimensional personnel capability portraits in scenarios such as enterprise personnel assessment and personnel task allocation. Aiming at application scenarios, the present invention introduces multiple mechanisms to improve the reasoning accuracy on the attention neural network model, and designs multiple adjustable neural network modules to meet actual needs. Under the same parameter amount, the training reasoning speed and reasoning accuracy are better than the traditional RNN network and convolutional neural network of direct migration.

[0042] See also Figure 1 The present invention provides a method for automatically evaluating personnel capabilities based on an attention neural network, comprising the following steps: S1. Collect static feature sets from the personnel database , Dynamic feature type collection , dynamic eigenvalue set , and the capability type set and capability level set ; Among them, static features refer to features of personnel whose importance does not change over time, such as personnel's certificates, while dynamic features refer to features whose importance decreases over time, such as personnel's work records. Considering that the impact of static features is often different from each other, but dynamic features often have the same impact, a dynamic feature value set is set And complete the build by collecting data.

[0043] The capabilities of a person are represented by a sequence of two tuples. Description, where ,and .

[0044] Similarly, static characteristics related to personnel and capabilities can be sequenced Description, where Indicates the number of static features of a person. ; Similarly, all dynamic features of a person are represented by a triple sequence Description, where Indicates the number of personnel with dynamic characteristics; , , Indicates the historical time length of the dynamic feature.

[0045] S2. Collect historical personnel characteristics and corresponding capability evaluation data sets , each data in the data set is Organized in the form of The proportion will Split into training set , validation set , test set ; S3. Set up multiple different neural network configurations Composition configuration collection , each configuration can initialize different neural networks and training methods; For any , contains two parts of configuration ,in, Configuration for model hyperparameters, including embedding size , capability type mapping network layer number , feature type mapping network layers , the number of eigenvalue mapping network layers , the number of network layers for extracting features of personnel capability classification ; Configuration of model training parameters, including learning rate , weight decay regularization weight , type vector dot product regularization term Weight , embedding vector distance regularization term Weight .

[0046] S4. Select hyperparameters , initialize the neural network model ; See also Figure 2 , the designed neural network model Contains embedded modules , Capability Embedding Mapping Module , cross-weight feature fusion module and classification module , initialize them respectively.

[0047] Initialize the embedded module When, according to the collected static feature set , Dynamic feature type collection , dynamic eigenvalue set , Capability Type Set and embedding size , use the He initialization method to randomly initialize the static feature type embedding matrix , dynamic feature type embedding matrix , dynamic feature optional value embedding matrix , and randomly initialize the dynamic feature type encoding vector , dynamic eigenvalue encoding vector , static feature type encoding vector , static eigenvalue encoding vector , use Xavier initialization method to initialize the capability type embedding matrix .

[0048] Initialize capability mapping module When mapping the number of network layers according to the capability type , using Xavier initialization method to randomly initialize the composition The linear layer matrix parameters in each layer of the bit-wise feed-forward network are With offset parameters , where the parameters of each layer are , , .

[0049] Initialize the cross-weight feature fusion module When using the He initialization method, according to the hyperparameters and Initialize the feature type mapping network contained therein respectively and eigenvalue mapping network The matrix parameters of the linear layer in each layer of the bit-wise feed-forward network and With offset parameters and , the parameters of each layer in the two networks , , , ,in , , and initialization Medium time decay layer Parameters An experience value.

[0050] Initialize the classification module When, according to the parameters Using the He initialization method, the linear layer matrix parameters of each layer of the bit-by-bit feedforward network are randomly initialized. and the linear layer offset parameter , where for any , , And initialized randomly using the He initialization method. According to the level set initialization The final fully connected classification layer parameters ,in .

[0051] For parameters to be initialized The Xavier initialization method used in the initialization process is for The activation function is performed, and the elements in the parameters are sampled from a normal distribution with a fixed standard deviation, as described in Formula 1: (1) The He initialization method used is for The activation function, as described in Formula 2: (2) S5. Use training set The neural network model obtained by initializing the sample training step S4 in ; For randomly selected samples , which is input to the neural network model After that, the embedding layer will , The feature types in the sample are processed to obtain the static feature type embedding vector sequence of the sample personnel. and dynamic feature type embedding vector sequence , and As the target capability input to the embedding layer to obtain the target capability embedding sequence ; After that, the capability embedding sequence is input into the capability type mapping network Get the capability embedding sequence after feature extraction , Contains multiple layers of the same structure using residual connections and As the activation function, the bitwise feed-forward network embeds the input capability type , No. The operation of the layer network is equivalent to the function , as shown in Formula 3: (3) For the sequence and , the embedding layer performs vector addition on the type and value variables plus the encoding vector , , , , then complete The length of historical events for each feature in ,Then the two are spliced ​​and merged into a unified personnel capability feature sequence, and the content in the split tuple is divided into three sequences: feature type embedding sequence , the eigenvalue embedding sequence and characteristic time series Input to submodule for further processing and aggregation, , , The length of .

[0052] sequence Through the module Perform feature transformation to obtain a new feature type sequence . Multiple layers use residual connections and The activation function is a bitwise feedforward network, whose first Layer network for feature type sequence Mean vector The operation process As shown in Formula 4: (4) Personnel characteristic value sequence Through the module Perform feature transformation to obtain a new feature value sequence . Using residual connections and The activation function is composed of a bit-by-bit feed-forward network, the first Layer network sequence Mean vector Feature extraction operation As shown in Formula 5: (5) Characteristic time value series After the time decay layer, the time weight sequence is obtained The time decay layer uses a time parameter The time decay function Calculate as shown in Formula 6: (6) What you get , , pass Module according to Each capability type in is aggregated separately. Middle Vector , using scaled click attention to calculate each feature in the person feature sequence relative to The attention weight Sequence, for the Feature type embedding , the corresponding attention weight The calculation method is shown in Formula 7: (7) By crossing the time weight sequence With attention weight The sequence gets the final eigenvalue weight sequence , No. The weight of the feature value ;use Embed the input feature value into a sequence Perform weighted average aggregation to obtain the personnel's ability type Temporal and attentional convergence representation under ; for sequence After each capability type is embedded and aggregated, the personnel capability representation sequence under each target type is obtained. ,for , Corresponding ability type sequence In .

[0053] The final classification results of each personnel's ability are obtained through sub-modules right Perform feature extraction and classification to obtain the ability level sequence of different ability types of personnel ; Submodule First, the input sequence Each embedding is passed through the feature extraction network to obtain a linearly separable feature representation , the feature extraction network The operation of the layer network is equivalent to the function in formula 8 : (8) Eventually Each element in The final fully connected layer performs linear classification to obtain the classification result sequence , Formula 9 describes the sequence Middle Feature representation get The classification process, Represents a set of levels Take out elements, Representation vector No. The element corresponding to the subscript.

[0054] (9) During training, the neural network model The fully connected layer output result , The prediction vectors are calculated one by one, and the neural network model is optimized using back-propagation gradient descent. Parameters.

[0055] With two biased regularization terms and The loss function , Formula 10 describes The calculation method of Indicates the correct predicted level subscript, regular term For feature sequences and Calculation, regularization term For the embedded sequence , and Calculation, Formulas 11 and 12 describe the regularization term and The calculation method of .

[0056] (10) (11) (12) After the training set is sampled, use the validation set and validation indicator evaluation The performance of the network is saved; the evaluation indicator uses accuracy , recall rate , f1 score Take measurement.

[0057] S6, step S5 is carried out Wheel, remember The neural network model after rounds of training is , save the neural network model and evaluation results after each round of training; From each round of training Select the one with the highest f1 score among the evaluation indicators As a hyperparameter The best neural network model obtained , and put it in the test set The same evaluation index is used to test and save the neural network model and model evaluation results.

[0058] S7. Repeat steps S4 to S6 until all hyperparameters are used to initialize the neural network model and are tested, and deploy the model with the highest f1 score in the model evaluation results on the test set for subsequent personnel capability level assessment.

[0059] It will be appreciated by those skilled in the art that various aspects of the present invention may be implemented as systems, methods or program products. Therefore, various aspects of the present invention may be specifically implemented in the following forms, namely: complete hardware implementation, complete software implementation (including firmware, microcode, etc.), or a combination of hardware and software implementations, which may be collectively referred to herein as "circuits", "modules" or "platforms".

[0060] In another embodiment of the present invention, a system for automatic assessment of personnel capabilities based on an attention neural network is provided. The system can be used to implement the above-mentioned method for automatic assessment of personnel capabilities based on an attention neural network. Specifically, the system for automatic assessment of personnel capabilities based on an attention neural network includes a collection module, a setting module, a training module, a testing module and an assessment module.

[0061] Among them, the collection module collects the static feature set of personnel , Dynamic feature type collection , dynamic eigenvalue set , and the capability type set and capability level set ; Collect historical personnel characteristics and corresponding capability evaluation data sets , each data in the data set is Organization in the form of are static competence-related characteristics of personnel whose importance does not change over time, and is the personnel dynamic feature associated with importance and time, The rating results for each capability type of the person; the capability evaluation data set Split into training set , validation set , test set ; Setting module, setting multiple different neural network configurations Composition configuration collection , each configuration can initialize different neural networks and training methods; the hyperparameters selected , initialize the neural network model ; Training module, using the obtained training set The neural network model obtained by training initialization ; Test module, the training module Wheel, remember The neural network model after rounds of training is , and use the validation set Evaluate the trained model and save the neural network model and evaluation results after each round of training. After N rounds of training, select the model with the highest evaluation index on the validation set in the round and save it on the test set. Conduct tests on the system and save the test evaluation results; Evaluation module, repeatedly set the module to the test module until all hyperparameters are used to initialize the neural network model and the test set is obtained Test and deploy the neural network model with the highest f1 score in the model evaluation results for personnel capability level assessment.

[0062] In another embodiment of the present invention, a terminal device is provided, which includes a processor and a memory, wherein the memory is used to store a computer program, the computer program includes program instructions, and the processor is used to execute the program instructions stored in the computer storage medium. The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, which is suitable for implementing one or more instructions, and is specifically suitable for loading and executing one or more instructions to implement the corresponding method flow or corresponding function; the processor described in the embodiment of the present invention can be used for the operation of the personnel ability automatic assessment method based on the attention neural network, including: Collect static characteristics of personnel , Dynamic feature type collection , dynamic eigenvalue set , and the capability type set and capability level set ; Collect historical personnel characteristics and corresponding capability evaluation data sets , each data in the data set is Organization in the form of are static competence-related characteristics of personnel whose importance does not change over time, and is the personnel dynamic feature associated with importance and time, The rating results for each capability type of the person; the capability evaluation data set Split into training set , validation set , test set ; Set up multiple different neural network configurations Composition configuration collection , each configuration can initialize different neural networks and training methods; the hyperparameters selected , initialize the neural network model ; Use the obtained training set The neural network model obtained by training initialization ;conduct After the round, remember The neural network model after rounds of training is , and use the validation set Evaluate the trained model and save the neural network model and evaluation results after each round of training. After N rounds of training, select the model with the highest evaluation index on the validation set in the round and save it on the test set. Test on the test set and save the test evaluation results; repeat the above steps until all hyperparameters are used to initialize the neural network model and are tested on the test set. Test and deploy the neural network model with the highest f1 score in the model evaluation results for personnel capability level assessment.

[0063] In another embodiment of the present invention, the present invention also provides a storage medium, specifically a computer-readable storage medium (Memory), which is a memory device in a terminal device for storing programs and data. It is understandable that the computer-readable storage medium here can include both the built-in storage medium in the terminal device and the extended storage medium supported by the terminal device, and can be any tangible medium containing or storing a program, which can be used by an instruction execution system, device or device or used in combination with it. The computer-readable storage medium provides a storage space, which stores the operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by the processor are also stored in the storage space, and these instructions can be one or more computer programs (including program codes). It should be noted that more specific examples (non-exhaustive list) of the computer-readable storage medium here include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0064] Computer readable storage media also include data signals propagated in baseband or as part of a carrier wave, which carry readable program codes. Such propagated data signals can take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. The readable storage medium can also be any readable medium other than a readable storage medium, which can send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, device, or device. The program code contained on the readable storage medium can be transmitted using any appropriate medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination of the above.

[0065] Program code for performing the operations of the present invention may be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device may be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., through the Internet using an Internet service provider).

[0066] The processor may load and execute one or more instructions stored in the computer-readable storage medium to implement the corresponding steps of the method for automatically evaluating personnel capabilities based on the attention neural network in the above embodiment; the processor may load and execute the following steps: Collect static characteristics of personnel , Dynamic feature type collection , dynamic eigenvalue set , and the capability type set and capability level set ; Collect historical personnel characteristics and corresponding capability evaluation data sets , each data in the data set is Organization in the form of are static competence-related characteristics of personnel whose importance does not change over time, and is the personnel dynamic feature associated with importance and time, The rating results for each capability type of the person; the capability evaluation data set Split into training set , validation set , test set ; Set up multiple different neural network configurations Composition configuration collection , each configuration can initialize different neural networks and training methods; the hyperparameters selected , initialize the neural network model ; Use the obtained training set The neural network model obtained by training initialization ;conduct After the round, remember The neural network model after rounds of training is , and use the validation set Evaluate the trained model and save the neural network model and evaluation results after each round of training. After N rounds of training, select the model with the highest evaluation index on the validation set in the round and save it on the test set. Test on the test set and save the test evaluation results; repeat the above steps until all hyperparameters are used to initialize the neural network model and are tested on the test set. Test and deploy the neural network model with the highest f1 score in the model evaluation results for personnel capability level assessment.

[0067] See also Figure 3 , the terminal device is a computer device, and the computer device 60 of this embodiment includes: a processor 61, a memory 62, and a computer program 63 stored in the memory 62 and executable on the processor 61. When the computer program 63 is executed by the processor 61, the method for automatically evaluating personnel capabilities based on the attention neural network in the embodiment is implemented. To avoid repetition, it is not described one by one here. Alternatively, when the computer program 63 is executed by the processor 61, the functions of each model / unit in the system for automatically evaluating personnel capabilities based on the attention neural network in the embodiment are implemented. To avoid repetition, it is not described one by one here.

[0068] The computer device 60 may be a computing device such as a desktop computer, a notebook, a PDA, or a cloud server. The computer device 60 may include, but is not limited to, a processor 61 and a memory 62. Those skilled in the art will appreciate that Figure 3 This is only an example of the computer device 60 and does not constitute a limitation of the computer device 60. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, the computer device may also include input and output devices, network access devices, buses, etc.

[0069] The processor 61 may be a central processing unit (CPU), or other general-purpose processors, central processing units, graphics processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, data processing logic based on quantum computing, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may be any conventional processor, etc.

[0070] The memory 62 may be an internal storage unit of the computer device 60, such as a hard disk or memory of the computer device 60. The memory 62 may also be an external storage device of the computer device 60, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc., equipped on the computer device 60.

[0071] Furthermore, the memory 62 may include both an internal storage unit of the computer device 60 and an external storage device. The memory 62 is used to store computer programs and other programs and data required by the computer device. The memory 62 may also be used to temporarily store data that has been output or is to be output.

[0072] Any reference to a memory, database or other medium used in the embodiments provided in the present application may include at least one of a non-volatile and a volatile memory. Non-volatile memory may include a read-only memory (ROM), a magnetic tape, a floppy disk, a flash memory, an optical memory, a high-density embedded non-volatile memory, a resistive random access memory (ReRAM), a magnetic random access memory (MRAM), a ferroelectric random access memory (FRAM), a phase change memory (PCM), a graphene memory, etc. Volatile memory may include a random access memory (RAM) or an external cache memory, etc. As an illustration and not limitation, RAM may be in various forms, such as a static random access memory (SRAM) or a dynamic random access memory (DRAM), etc.

[0073] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.

[0074] See also Figure 4 The terminal device 600 is an electronic device, and the electronic device is in the form of a general computing device. The components of the electronic device may include but are not limited to: at least one processing unit 610, at least one storage unit 620, a bus 630 connecting different platform components (including the storage unit 620 and the processing unit 610), a display unit 640, etc.

[0075] The storage unit stores program codes, which can be executed by the processing unit 610, so that the processing unit 610 performs the steps according to various exemplary embodiments of the present invention described in the above method section of this specification. For example, the processing unit 610 can perform the following steps: Figure 1 Follow the steps shown in .

[0076] The storage unit 620 may include a readable medium in the form of a volatile storage unit, such as a random access memory unit (RAM) 6201 and / or a cache memory unit 6202 , and may further include a read-only memory unit (ROM) 6203 .

[0077] The storage unit 620 may also include a program / utility 6204 having a set (at least one) of program modules 6205, such program modules 6205 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0078] Bus 630 may represent one or more of several types of bus structures, including a memory unit bus or memory unit controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.

[0079] The electronic device 600 may also communicate with one or more external devices 700 (e.g., keyboards, pointing devices, Bluetooth devices, etc.), may also communicate with one or more devices that enable a user to interact with the electronic device 600, and / or communicate with any device that enables the electronic device 600 to communicate with one or more other computing devices (e.g., routers, modems, etc.). Such communication may be performed via an input / output (I / O) interface 650. Furthermore, the electronic device 600 may also communicate with one or more networks (e.g., a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) via a network adapter 660. The network adapter 660 may communicate with other modules of the electronic device 600 via a bus 630. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the electronic device 600, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage platforms, etc.

[0080] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. The components of the embodiments of the present invention described and shown in the drawings here can usually be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0081] In the simulation data of subway maintenance personnel's ability portrait, the model can integrate personnel characteristics including personnel's education information, personnel's ability certificates and personnel's work history, and generate maintenance personnel's maintenance ability levels on different types and different equipment. The accuracy and f1 score of the final test results on this simulation data set can reach above 0.85, which is more than 20% higher than that of traditional convolutional networks and RNN networks; under the same number of parameters, the training time and inference time on parallel computing devices (GPU, etc.) are also significantly lower than those of RNN networks and convolutional networks.

[0082] In summary, the present invention provides an automatic personnel capability assessment method and system based on an attention neural network, which can automatically mine the association rules between personnel characteristics and personnel capability portraits, and based on attention neural networks, time attenuation, feature intersection and other technologies, reasonably fit personnel characteristics for reasoning and judgment, and automatically generate dynamic personnel capability portraits to meet the needs of enterprises for personnel capability portraits in regular personnel assessments and personnel task scheduling, thereby achieving the purpose of improving execution efficiency while reducing personnel manpower costs.

[0083] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0084] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0085] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in the present invention can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0086] In the embodiments provided by the present invention, it should be understood that the disclosed devices / terminals and methods can be implemented in other ways. For example, the device / terminal embodiments described above are only schematic. For example, the division of the modules or units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0087] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0088] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0089] If the integrated module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electric carrier signals and telecommunication signals.

[0090] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0091] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0092] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0093] The above contents are only for explaining the technical idea of ​​the present invention and cannot be used to limit the protection scope of the present invention. Any changes made on the basis of the technical solution in accordance with the technical idea proposed by the present invention shall fall within the protection scope of the claims of the present invention.

Claims

1. A method for automatic assessment of personnel capabilities based on attention neural network, characterized in that: The following steps are involved: S1. Collect static feature sets of personnel , Dynamic feature type collection , dynamic eigenvalue set , and the capability type set and capability level set ; S2. Collect historical personnel characteristics and corresponding capability evaluation data sets , each data in the data set is Organization in the form of are static competence-related characteristics of personnel whose importance does not change over time, and is the personnel dynamic feature associated with importance and time, The rating results for each capability type of the person; the capability evaluation data set Split into training set , validation set , test set ; S3. Set up multiple different neural network configurations Composition configuration collection , each configuration can initialize different neural networks and training methods; S4. Select the hyperparameters obtained in step S3 , initialize the neural network model ; S5. Use the training set obtained in step S2 The neural network model obtained by initialization in training step S4 ; S6, step S5 is carried out Wheel, remember The neural network model after rounds of training is , and use the validation set Evaluate the trained model and save the neural network model and evaluation results after each round of training. After N rounds of training, select the model with the highest evaluation index on the validation set in the round and save it on the test set. Conduct tests on the system and save the test evaluation results; S7, repeat steps S4 to S6 until all hyperparameters are used to initialize the neural network model and the test set obtained in step S2 Test and deploy the neural network model with the highest f1 score in the model evaluation results for personnel capability level assessment.

2. The method for automatic personnel ability assessment based on attention neural network according to claim 1 is characterized in that: In step S3, for any , contains two parts of configuration , Configuration for model hyperparameters, including embedding size , capability type mapping network layer number , feature type mapping network layers , the number of eigenvalue mapping network layers , the number of network layers for extracting features of personnel capability classification ; Configuration of model training parameters, including learning rate , weight decay regularization weight , type vector dot product regularization term Weight , embedding vector distance regularization term Weight .

3. The method for automatic assessment of personnel capabilities based on attention neural network according to claim 1 is characterized in that: In step S4, the neural network model Contains embedded modules , Capability Embedding Mapping Module , cross-weight feature fusion module and classification module , and initialize them respectively; Xavier initialization method is used for The activation function is performed, the elements in the parameters are sampled from a normal distribution with a fixed standard deviation, and the He initialization method is used for Activation function.

4. The method for automatic personnel ability assessment based on attention neural network according to claim 3 is characterized in that: Initialize the embedded module When, according to the collected static feature set , Dynamic feature type collection , dynamic eigenvalue set , Capability Type Set and embedding size , use the He initialization method to randomly initialize the static feature type embedding matrix , dynamic feature type embedding matrix , dynamic feature optional value embedding matrix , and randomly initialize the dynamic feature type encoding vector , dynamic eigenvalue encoding vector , static feature type encoding vector , static eigenvalue encoding vector , use Xavier initialization method to initialize the capability type embedding matrix ; Initialize capability mapping module When mapping the number of network layers according to the capability type , using Xavier initialization method to randomly initialize the composition The linear layer matrix parameters in each layer of the bit-wise feed-forward network are With offset parameters , where the parameters of each layer are , , ; Initialize the cross-weight feature fusion module When using the He initialization method, according to the hyperparameters and Initialize the feature type mapping network contained therein respectively and eigenvalue mapping network The matrix parameters of the linear layer in each layer of the bit-wise feed-forward network and With offset parameters and , the parameters of each layer in the two networks , , , ,in , , and initialization Medium time decay layer Parameters is an experience value; Initialize the classification module When, according to the parameters Using the He initialization method, the linear layer matrix parameters of each layer of the bit-by-bit feedforward network are randomly initialized. and the linear layer offset parameter , where for any , , , and initialized randomly using the He initialization method; according to the level set initialization The final fully connected classification layer parameters , .

5. The method for automatic personnel ability assessment based on attention neural network according to claim 3 is characterized in that: Using Xavier initialization method The activation function is as follows: Use He initialization method for The activation function is as follows: in, is a Gaussian random distribution, is the standard deviation of the Gaussian random distribution, Enter the eigenvector dimensions for the matrix, Output eigenvector dimensions for the matrix.

6. The method for automatic assessment of personnel capabilities based on attention neural network according to claim 1 is characterized in that: Step S5 is specifically as follows: The randomly selected samples Input neural network model , the embedding layer will , The feature types in the sample are processed to obtain the static feature type embedding vector sequence of the sample personnel. and dynamic feature type embedding vector sequence , and As the target capability input to the embedding layer to obtain the target capability embedding sequence ; The capability embedding sequence is input into the capability type mapping network Get the capability embedding sequence after feature extraction ; For the sequence and The embedding layer uses vector addition to add the encoding vector to the type and value variables respectively , , , , then complete The length of historical events for each feature in ; Then the two are spliced ​​and merged into a unified personnel capability feature sequence, and the content in the split tuple is divided into three sequences - feature type embedding sequence , the eigenvalue embedding sequence and characteristic time series Input to submodule Processing and aggregation; Feature type embedding sequence Through the module Perform feature transformation to obtain a new feature type sequence ; Personnel characteristic value sequence Through the module Perform feature transformation to obtain a new feature value sequence ; Characteristic time value series After the time decay layer, the time weight sequence is obtained The time decay layer uses a time parameter The time decay function Perform calculations; What you get , , pass Module according to Each capability type in is aggregated separately; Middle Vector , using scaled click attention to calculate each feature in the person feature sequence relative to The attention weight sequence; By crossing the time weight sequence With attention weight The sequence gets the final eigenvalue weight sequence , No. The weight of the feature value ;use Embed the input feature value into a sequence Perform weighted average aggregation to obtain the personnel's ability type Temporal and attentional convergence representation under ; Pair Sequence After each capability type is embedded and aggregated, the personnel capability representation sequence under each target type is obtained. ; The final classification results of each personnel's ability are obtained through sub-modules right Perform feature extraction and classification to obtain the ability level sequence of different ability types of personnel ; Submodule First, the input sequence Each embedding is passed through the feature extraction network to obtain a linearly separable feature representation ; Eventually Each element in The final fully connected layer performs linear classification to obtain the classification result sequence ; During training, the neural network model The fully connected layer output result The prediction vectors are calculated one by one, and the neural network model is optimized using back-propagation gradient descent. Parameters; With two biased regularization terms and The loss function After the training set is sampled, use the validation set and validation indicator evaluation performance, and save it.

7. The method for automatic assessment of personnel capabilities based on attention neural network according to claim 6 is characterized in that: Contains multiple layers of the same structure using residual connections and As the activation function, the bitwise feed-forward network embeds the input capability type , No. The operation of the layer network is equivalent to the function as follows: Multiple layers use residual connections and The bit-by-bit feed-forward network of the activation function is composed of Layer network for feature type sequence Mean vector The operation process as follows: Using residual connections and The bit-by-bit feed-forward network of the activation function is composed of Layer network sequence Mean vector Feature extraction operation as follows: in, is the capability type feature mapping matrix, is the capability type feature mapping offset, is the capability type feature vector, is the personnel feature mapping matrix, Map offsets for person feature types, is the characteristic vector of personnel feature type, is the comprehensive feature mapping matrix of personnel, Offset for the comprehensive feature map of personnel, is the comprehensive feature vector of personnel.

8. The method for automatic personnel ability assessment based on attention neural network according to claim 6 is characterized in that: Two biased regularization terms and The loss function as follows: in, represents the correct predicted rank subscript, is the model’s predicted probability for different levels, is the ability type regularization loss, is the regularization loss of personnel feature type, is the capability type regularization term ratio hyperparameter, is the regularization term ratio hyperparameter of the personnel feature type.

9. The method for automatic assessment of personnel capabilities based on attention neural network according to claim 1, characterized in that: In step S6, from each round of training Select the one with the highest f1 score among the evaluation indicators As a hyperparameter The best neural network model obtained , and put it in the test set The same evaluation index is used to test and save the neural network model and model evaluation results.

10. An automatic personnel capability assessment system based on attention neural network, characterized in that: include: Collection module, collects static feature sets of personnel , Dynamic feature type collection , dynamic eigenvalue set , and the capability type set and capability level set ; Collect historical personnel characteristics and corresponding capability evaluation data sets , each data in the data set is Organization in the form of are static competence-related characteristics of personnel whose importance does not change over time, and is the personnel dynamic feature associated with importance and time, The rating results for each capability type of the person; the capability evaluation data set Split into training set , validation set , test set ; Setting module, setting multiple different neural network configurations Composition configuration collection , each configuration can initialize different neural networks and training methods; the hyperparameters selected , initialize the neural network model ; Training module, using the obtained training set The neural network model obtained by training initialization ; Test module, the training module Wheel, remember The neural network model after rounds of training is , and use the validation set Evaluate the trained model and save the neural network model and evaluation results after each round of training. After N rounds of training, select the model with the highest evaluation index on the validation set in the round and save it on the test set. Conduct tests on the system and save the test evaluation results; Evaluation module, repeatedly set the module to the test module until all hyperparameters are used to initialize the neural network model and the test set is obtained Test and deploy the neural network model with the highest f1 score in the model evaluation results for personnel capability level assessment.

Images