Talent information processing method and device
By preprocessing and fusing multi-source data and using trained talent prediction models to screen talent, the problems of low efficiency and high subjectivity in existing technologies are solved, and efficient and accurate talent screening is achieved.
Patent Information
- Application Number
- CN202510683413.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-26
- Publication Date
- 2025-09-23
AI Technical Summary
The talent information processing in existing technologies is inefficient and highly subjective, resulting in inaccurate talent screening.
By obtaining multi-source data of the target object, performing preprocessing and fusion processing, and using the trained talent prediction model to make predictions, the talent screening results are determined.
It improves the efficiency and accuracy of talent screening, reduces the difficulty and subjectivity of manual screening, and provides fast and automated data processing and screening support.
Smart Images
Figure CN120689017A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of information processing technology, and also to a method and device for processing talent information. Background Art
[0002] With the advent of economic globalization and the information age, the importance of talent information processing has become increasingly prominent. This process not only affects the ability of businesses, organizations, and institutions to effectively attract and retain key talent, but also directly impacts operational efficiency and strategic decision-making. Determining whether collected personnel information qualifies as needed is a crucial step in talent screening. However, existing technologies often rely on manual screening, which is inefficient and highly subjective, significantly impacting talent selection. Summary of the Invention
[0003] The technical problem to be solved by the present invention is to provide a method and device for processing talent information to improve the efficiency and accuracy of talent screening.
[0004] In order to solve the above technical problems, the technical solutions of the present invention are as follows:
[0005] A first aspect of the present invention provides a method for processing talent information, comprising:
[0006] Obtain multi-source data of the target object;
[0007] Preprocessing the multi-source data to obtain preprocessed data;
[0008] Performing fusion processing on the pre-processed data to obtain fused data;
[0009] Obtaining a prediction result based on the fused data and a trained talent prediction model; the talent prediction model is obtained by training a preset network model based on the collected historical multi-source data;
[0010] A talent screening result is determined based on the prediction result.
[0011] Optionally, obtain multi-source data for the target object, including:
[0012] Identify target audiences;
[0013] Acquire multi-source data of the target object; the multi-source data includes basic data, education data, work data, academic achievement data and influence data of the target object.
[0014] Optionally, preprocessing the multi-source data to obtain preprocessed data includes:
[0015] Determining a preprocessing method based on the multi-source data;
[0016] The multi-source data is preprocessed according to the preprocessing method to obtain preprocessed data.
[0017] Optionally, performing fusion processing on the preprocessed data to obtain fused data includes:
[0018] Determining a fusion method according to the dimension of the preprocessed data;
[0019] The pre-processed data is fused according to the fusion method to obtain fused data.
[0020] Optionally, obtaining a prediction result based on the fused data and the trained talent prediction model includes:
[0021] Inputting the fused data into the first processing layer of the talent prediction model to obtain a first output result; wherein the first processing layer includes z1=W1Y+b1, h1=ReLU(z1); wherein z1 is the inactivated output of the first processing layer, W1 is the weight matrix of the first processing layer, Y is the fused data, b1 is the bias term of the first processing layer, h1 is the first output result, and ReLU is the activation function;
[0022] The first output result is input into the second processing layer of the talent prediction model to obtain a second output result; wherein the second processing layer includes in, is the inactivated output of the second processing layer, W2 is the weight matrix of the second processing layer, h1 is the first output result, b2 is the bias term of the second processing layer, is the second output result, ε is the activation function;
[0023] A prediction result is determined according to the second output result and a preset threshold.
[0024] Optionally, the training process of the talent prediction model includes:
[0025] Collect historical multi-source data; the historical multi-source data includes basic data, education data, work data, academic achievement data, and influence data;
[0026] Determining training samples based on the historical multi-source data;
[0027] Train the preset network model according to the training samples to obtain a trained model;
[0028] According to m t =β1m t-1 +(1-β1)g t 、 The trained model is optimized to obtain a talent prediction model; wherein, mt is the gradient mean of the t-th iteration, β1 is the first-order moment decay rate, m t-1 is the mean gradient of the t-1th iteration, g t is the gradient of the tth iteration, v t is the second-order moment estimate of the gradient at the t-th iteration, β2 is the second-order moment decay rate, is the square of the gradient of the tth iteration, v t-1 is the second-order moment estimate of the gradient at the t-1th iteration, is the revised first-order moment estimate, t is the current iteration number, as t increases, Approaching 0, is β1 raised to the power of t, is β2 raised to the power of t, is the modified second-order moment estimate, θ t+1 is the updated model parameter, θ t is the model parameter at the tth iteration, γ is the learning rate, and τ is the preset value.
[0029] Optionally, determining a talent screening result based on the prediction result includes:
[0030] Determine a talent training plan based on the prediction results;
[0031] Determine the talent screening results based on the talent training plan.
[0032] A second aspect of the present invention provides a talent information processing device, comprising:
[0033] Acquisition module, used for multi-source data of target objects;
[0034] The processing module is used to preprocess the multi-source data to obtain preprocessed data; perform fusion processing on the preprocessed data to obtain fused data; obtain prediction results based on the fused data and a trained talent prediction model; the talent prediction model is obtained by training a preset network model based on the collected historical multi-source data; and determine the talent screening results based on the prediction results.
[0035] According to a third aspect of the present invention, a computing device is provided, comprising: a processor and a memory storing a computer program, wherein when the computer program is executed by the processor, the method according to the first aspect is executed.
[0036] A fourth aspect of the present invention provides a computer-readable storage medium storing instructions, which, when executed on a computer, causes the computer to execute the method described in the first aspect.
[0037] The above solution of the present invention includes at least the following beneficial effects:
[0038] The above-mentioned scheme of the present invention obtains multi-source data of the target object, fuses the multi-source data to obtain fused data, pre-processes the fused data to obtain pre-processed data, obtains prediction results based on the pre-processed data and the trained talent prediction model, and finally determines the talent training plan based on the prediction results, thereby reducing the difficulty and subjectivity of manual screening, and can quickly, effectively and automatically process and screen the data, provide support and assistance for determining the talent training plan, and help improve work efficiency and accuracy. BRIEF DESCRIPTION OF THE DRAWINGS
[0039] Figure 1 is a flowchart of a method for processing talent information in an embodiment of the present invention;
[0040] Figure 2 It is a structural diagram of a talent information processing device in an embodiment of the present invention. DETAILED DESCRIPTION
[0041] Exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the accompanying drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present invention and to fully convey the scope of the present invention to those skilled in the art.
[0042] like Figure 1 As shown, an embodiment of the present invention provides a method for processing talent information, comprising the following steps:
[0043] Step 101, obtaining multi-source data of the target object;
[0044] Step 102: preprocessing the multi-source data to obtain preprocessed data;
[0045] Step 103, performing fusion processing on the pre-processed data to obtain fused data;
[0046] Step 104: obtaining a prediction result based on the fused data and the trained talent prediction model;
[0047] Step 105: Determine talent screening results based on the prediction results.
[0048] The talent information processing method of the embodiment of the present invention obtains multi-source data of the target object, fuses the multi-source data to obtain fused data, pre-processes the fused data to obtain pre-processed data, obtains prediction results based on the pre-processed data and a trained talent prediction model, and finally determines the talent training plan based on the prediction results, thereby reducing the difficulty and subjectivity of manual screening, and can quickly, effectively and automatically process and screen data, provide support and assistance for determining talent screening results, and help improve work efficiency and accuracy.
[0049] In an optional embodiment of the present invention, step 101 includes:
[0050] Step 1011, determining the target object;
[0051] Specifically, the target audience is determined according to preset criteria, which can be customized based on user needs. For example, in one application scenario, the preset criteria are Traditional Chinese Medicine professionals in the first-tier talent program who are under 40 years old. Here, the first-tier talent program is determined based on actual conditions and can include at least one talent project or program.
[0052] Step 1012, obtaining multi-source data of the target object; the multi-source data includes basic data, education data, work data, academic achievement data and influence data of the target object.
[0053] Specifically, the multi-source data of the target object can be obtained through surveys or online information acquisition, and the type of multi-source data can be selected according to the actual situation.
[0054] Here, taking the application scenario of traditional Chinese medicine talents as an example, the basic data may include: (1) basic information: age, gender, professional title, talent name and classification (clinical type, scientific research type) when selected for the first-level talent project; (2) educational background: academic qualifications, degree, name of the graduating school and school level, whether there is internship work experience in multiple units, the level of the instructor, whether there are multiple famous teachers guiding at the same time, whether there is multi-professional learning experience, etc.; (3) professional skills: professional direction of long-term research, including the third-level discipline of clinical research, basic research direction, etc.; (4) time of studying medicine: years of studying traditional Chinese medicine; (5) family background: whether there is a family background of traditional Chinese medicine, the degree of support for the study of traditional Chinese medicine in the family environment, etc.
[0055] Educational data may include: (1) Study abroad / visiting scholar experience: age of the scholar at the time of study abroad / visiting scholar, study abroad institution, institution level and industry status, main research direction during study abroad, duration of study abroad, etc.; (2) Study abroad / visiting scholar supervisor information: supervisor’s unit, whether there are multiple supervisors, each supervisor’s academic status, supervisor’s research direction, whether there are interdisciplinary subjects, etc.; (3) Major achievements during study abroad / visiting scholar period: whether high-scoring articles were published, the impression factor of the highest-scoring articles published, whether the scholar presided over or participated in first-level major projects, etc.
[0056] Work data may include: (1) practical experience: the specialty and professional direction of the work, whether there is work experience in multiple units, whether there is work experience in multiple majors, whether there is work experience in multiple positions, etc.; (2) post-employment continuing education experience: post-employment learning experience, including whether to learn from famous teachers, the level of the tutor, whether there are multiple famous teachers for guidance, the professional direction of the tutor, etc.; (3) the situation of the unit: the type of unit organization, the level of the unit, the situation of the first-level platform of the hospital, the situation of the first-level talents of the institution, etc.; (4) the construction situation of the discipline and specialty: the scale of the department, whether it is a first-level key discipline / specialty, whether the academic leader is a first-level talent, etc.; (5) the construction situation of the talent echelon of the team: whether the team has multiple first-level talents, whether it has an interdisciplinary or cross-institutional team, etc.
[0057] Academic achievement data may include: (1) the total number of SCI impact factors of the first author and corresponding author as of the year of selection, the highest SCI impact factor of a single article, etc.; (2) the level and ranking of scientific and technological progress awards received by the first and second completers; (3) whether the first completer has transformed the results, the amount of the results transformed, etc.; (4) whether the first major scientific research project is presided over, the number of participants, etc.
[0058] Influence data may include: (1) the appointment of chairpersons and vice chairpersons of first-level academic societies; (2) administrative positions, etc.; (3) international exchanges: whether the person has been invited to speak or give academic reports at first-level international academic societies, etc.; (4) whether the person has multiple other honors above the preset level: honor level, number of honors, etc.; (5) whether the unit has independent publicity and reporting by individuals; (6) whether the person has established a personal public account, the number of other self-media publicity, the number of followers, etc.
[0059] It should be noted that the first level, level, etc. mentioned above can be selected as a specific level or type according to specific circumstances.
[0060] In an optional embodiment of the present invention, step 102 includes:
[0061] Step 1021, determining a preprocessing method based on the multi-source data;
[0062] Specifically, multi-source data includes text data, numerical data, and so on. Therefore, it is necessary to determine the corresponding preprocessing method based on the different data types. In this embodiment, the preprocessing method for basic data is to handle missing values and encode categorical features; the preprocessing method for education data is to discretize rankings and normalize average grade points; the preprocessing method for work data is to logarithmically transform salaries and discretize years of work experience; the preprocessing method for academic achievements is to logarithmically transform the number of papers and quantify conference levels; and the preprocessing method for influence data is to normalize the number of followers and binarize awards.
[0063] Step 1022: preprocess the multi-source data according to the preprocessing method to obtain preprocessed data.
[0064] Specifically, different data types in the multi-source data are preprocessed according to the determined preprocessing method. For example, the missing values in the basic data are filled with the median (such as age) or mode (such as gender). At the same time, the gender is categorized using the one-hot encoding method (such as encoding male as [1,0] and female as [0,1]); for the school rankings in the education data, Discretize, where R is the discretized school ranking, and then perform one-hot encoding. For example, if the school ranking in the education data is 5, it will be encoded as [1,0,0]; Normalize the credit points, where x nrom is the normalized credit score, x1 is the original credit score, and x max is the maximum value of the original credit points, x min is the minimum value of the original credit score; for the salary in the work data, use x log =log(x2+1), where x log is the salary after logarithmic transformation, and x2 is the original salary; the number of papers / citations in the academic achievement data can also be preprocessed by the above logarithmic transformation method; the number of fans in the influence data can be preprocessed by the above normalization method; the above is only an example, and the specific preprocessing method can be selected according to the actual situation.
[0065] By preprocessing multi-source data, the data format is unified, providing a basis for subsequent data fusion and avoiding the impact of missing data, inconsistent data and abnormal data on subsequent processing efficiency and accuracy.
[0066] In an optional embodiment of the present invention, step 103 includes:
[0067] Step 1031, determining a fusion method according to the dimension of the pre-processed data;
[0068] Specifically, the dimension of the preprocessed data refers to the total dimension of the preprocessed basic data, education data, work data, academic achievement data, and influence data. For example, if the basic data includes the gender "female," and the basic data after preprocessing using one-hot encoding is [0, 1], then the dimension of the basic data is 2. When the dimension of the preprocessed data is less than the preset value, the first fusion method is selected, which directly concatenates the preprocessed basic data, education data, work data, academic achievement data, and influence data in the preprocessed data. When the dimension of the preprocessed data is less than the preset value, the second fusion method is selected, which fuses the data using weighted summation.
[0069] Step 1032: perform fusion processing on the pre-processed data according to the fusion method to obtain fused data.
[0070] Specifically, if the fusion method is the first fusion method, the preprocessed basic data, education data, work data, academic achievement data and influence data in the preprocessed data are directly spliced, that is, Y = [y1, y2, y3, y4, y5], where Y is the fused data, and y1, y2, y3, y4, y5 are the preprocessed basic data, education data, work data, academic achievement data and influence data respectively. If the fusion method is the second fusion method, the pre-processed basic data, education data, work data, academic achievement data and influence data are fused according to the preset weights, that is, Y = [w1y1, w2y2, w3y3, w3y4, w3y5], where Y is the fused data, y1, y2, y3, y4, y5 are the pre-processed basic data, education data, work data, academic achievement data and influence data respectively, and y1, y2, y3, y4, y5 are the weights of the pre-processed basic data, the weight of the pre-processed education data, the weight of the pre-processed work data, the weight of the pre-processed academic achievement data and the weight of the pre-processed influence data respectively. Each weight reflects the importance of various data, which is conducive to improving the accuracy of subsequent predictions.
[0071] In an optional embodiment of the present invention, step 104 includes:
[0072] Step 10411: Input the fused data into the first processing layer of the talent prediction model to obtain a first output result; wherein the first processing layer includes z1=W1Y+b1, h1=ReLU(z1); wherein z1 is the inactivated output of the first processing layer, W1 is the weight matrix of the first processing layer, Y is the fused data, b1 is the bias term of the first processing layer, h1 is the first output result, and ReLU is the activation function;
[0073] Step 10412: input the first output result into the second processing layer of the talent prediction model to obtain a second output result; wherein the second processing layer includes in, is the inactivated output of the second processing layer, W2 is the weight matrix of the second processing layer, h1 is the first output result, b2 is the bias term of the second processing layer, is the second output result, ε is the activation function (which can be a Sigmoid activation function);
[0074] Step 10412: Determine a prediction result based on the second output result and a preset threshold.
[0075] Specifically, if the value of the second output result is greater than a preset threshold, the prediction result is that the person is a talent, otherwise the person is not a talent. By setting a preset threshold, a screening basis is provided for whether the person is a talent, thereby improving the accuracy of talent screening.
[0076] In an optional embodiment of the present invention, step 104 includes:
[0077] Step 10421: Collect historical multi-source data; the historical multi-source data includes basic data, education data, work data, academic achievement data, and influence data;
[0078] Specifically, relevant data previously identified as talents and non-talents can be obtained from the talent database and bound through transformation (such as binding talent-related data to the "talent" label and binding non-talent-related data to the "non-talent" label). The bound data is used as data for model training.
[0079] Step 10422: determining training samples based on the historical multi-source data;
[0080] Specifically, the historical multi-source data are normalized, one-hot encoded, and then spliced to obtain the spliced data, which are then divided into training set, validation set, and test set in a ratio of 7:1.5:1.5 as training samples.
[0081] Step 10423: training the preset network model according to the training samples to obtain a trained model;
[0082] Specifically, the training set can be used to train the preset network model. The training goal is to obtain the optimal weight matrix W1 of the first processing layer, the weight matrix W2 of the second processing layer, the bias item b1 of the first processing layer, and the bias item b2 of the second processing layer, so as to minimize the prediction error. During the training process, the weight matrix W1 of the first processing layer, the weight matrix W2 of the second processing layer, the bias item b1 of the first processing layer, and the bias item b2 of the second processing layer will be automatically updated through back propagation. Here, the preset network model structure used includes the first processing layer (z1=W1Y+b1, h1=ReLU(z1)), the second processing layer includes
[0083] After training, the validation set is used to verify the trained model. If the accuracy does not meet expectations, the trained model is optimized.
[0084] Step 10424, according to m t =β1m t-1 +(1-β1)g t 、 The model parameters of the trained model are optimized to obtain a talent prediction model; wherein, m t is the gradient mean of the t-th iteration, β1 is the first-order moment decay rate, m t-1 is the mean gradient of the t-1th iteration, g t is the gradient of the tth iteration, v t is the second-order moment estimate of the gradient at the t-th iteration, β2 is the second-order moment decay rate, is the square of the gradient of the tth iteration, v t-1 is the second-order moment estimate of the gradient at the t-1th iteration, is the revised first-order moment estimate, t is the current iteration number, as t increases, Approaching 0, is β1 raised to the power of t, is β2 raised to the power of t, is the modified second-order moment estimate, θ t+1 is the updated model parameter, θ t is the model parameter at the tth iteration, γ is the learning rate, and τ is the preset value, which can be set to 10 -8 The model parameters include the weight matrix W1 of the first processing layer, the weight matrix W2 of the second processing layer, the bias term b1 of the first processing layer, and the bias term b2 of the second processing layer.
[0085] Specifically, the above formula can be used to optimize the model's weights and biases, and based on the updated (or optimized) model parameters, a talent prediction model can be determined. By collecting historical multi-source data to train a preset network model and optimizing the trained model, the model's output prediction results become more accurate, which helps improve prediction accuracy.
[0086] In an optional embodiment of the present invention, step 105 includes:
[0087] Step 1051: Determine a talent training plan based on the prediction result;
[0088] Specifically, the training program corresponding to the predicted results or multi-source data can be searched from the historical talent database to provide a basis and reference for subsequent talent screening results. Here, the historical talent database may include multi-source data related to historical talents, predicted results, and corresponding training programs.
[0089] Step 1052: Determine the talent screening result according to the talent training plan.
[0090] Specifically, the talent screening results may include multi-source data, prediction results and talent training plans, providing sufficient and objective data basis for talent screening.
[0091] A specific embodiment of the method for processing talent information according to the embodiment of the present invention includes:
[0092] Step 111, obtaining multi-source data of the target object;
[0093] Obtain the target object's basic data, education data, work data, academic achievement data and influence data through surveys or online information acquisition, providing a data basis for subsequent talent prediction.
[0094] Step 112, pre-processing;
[0095] Different preprocessing methods are determined according to different data types, and then multi-source data are preprocessed separately to achieve the unification of data formats and provide a basis for subsequent data fusion.
[0096] Step 113: data fusion processing;
[0097] Data fusion processing is achieved by splicing multi-source data, which facilitates subsequent direct input into the model for processing.
[0098] Step 114, talent prediction;
[0099] The fused data is input into the talent prediction model to perform talent prediction and obtain the prediction results.
[0100] Step 115: Determine the talent screening result.
[0101] Search the historical talent database for training programs corresponding to the predicted results or multi-source data, and form talent screening results together with the multi-source data and predicted results for screening personnel to review and screen, thereby improving screening efficiency.
[0102] The talent information processing method according to the embodiment of the present invention can provide objective data basis and talent screening basis for talent screening personnel, effectively reducing the workload of screening personnel and improving work efficiency.
[0103] like Figure 2 As shown, an embodiment of the present invention provides a talent information processing device 200, comprising:
[0104] Acquisition module 201, for multi-source data of target object;
[0105] The processing module 202 is used to preprocess the multi-source data to obtain preprocessed data; perform fusion processing on the preprocessed data to obtain fused data; obtain a prediction result based on the fused data and a trained talent prediction model; the talent prediction model is obtained by training a preset network model based on the collected historical multi-source data; and determine a talent screening result based on the prediction result.
[0106] Optionally, obtain multi-source data for the target object, including:
[0107] Identify target audiences;
[0108] Acquire multi-source data of the target object; the multi-source data includes basic data, education data, work data, academic achievement data and influence data of the target object.
[0109] Optionally, preprocessing the multi-source data to obtain preprocessed data includes:
[0110] Determining a preprocessing method based on the multi-source data;
[0111] The multi-source data is preprocessed according to the preprocessing method to obtain preprocessed data.
[0112] Optionally, performing fusion processing on the preprocessed data to obtain fused data includes:
[0113] Determining a fusion method according to the dimension of the preprocessed data;
[0114] The pre-processed data is fused according to the fusion method to obtain fused data.
[0115] Optionally, obtaining a prediction result based on the fused data and the trained talent prediction model includes:
[0116] Inputting the fused data into the first processing layer of the talent prediction model to obtain a first output result; wherein the first processing layer includes z1=W1Y+b1, h1=ReLU(z1); wherein z1 is the inactivated output of the first processing layer, W1 is the weight matrix of the first processing layer, Y is the fused data, b1 is the bias term of the first processing layer, h1 is the first output result, and ReLU is the activation function;
[0117] The first output result is input into the second processing layer of the talent prediction model to obtain a second output result; wherein the second processing layer includes in, is the inactivated output of the second processing layer, W2 is the weight matrix of the second processing layer, h1 is the first output result, b2 is the bias term of the second processing layer, is the second output result, ε is the activation function;
[0118] A prediction result is determined according to the second output result and a preset threshold.
[0119] Optionally, the training process of the talent prediction model includes:
[0120] Collect historical multi-source data; the historical multi-source data includes basic data, education data, work data, academic achievement data, and influence data;
[0121] Determining training samples based on the historical multi-source data;
[0122] Train the preset network model according to the training samples to obtain a trained model;
[0123] According to m t =β1m t-1 +(1-β1)g t 、 The trained model is optimized to obtain a talent prediction model; wherein, m t is the gradient mean of the t-th iteration, β1 is the first-order moment decay rate, m t-1 is the mean gradient of the t-1th iteration, g t is the gradient of the tth iteration, v t is the second-order moment estimate of the gradient at the t-th iteration, β2 is the second-order moment decay rate, is the square of the gradient of the tth iteration, v t-1 is the second-order moment estimate of the gradient at the t-1th iteration, is the revised first-order moment estimate, t is the current iteration number, as t increases, Approaching 0, is β1 raised to the power of t, is β2 raised to the power of t, is the modified second-order moment estimate, θ t+1 is the updated model parameter, θ t is the model parameter at the tth iteration, γ is the learning rate, and τ is the preset value.
[0124] Optionally, determining a talent screening result based on the prediction result includes:
[0125] Determine a talent training plan based on the prediction results;
[0126] Determine the talent screening results based on the talent training plan.
[0127] The talent information processing device of the embodiment of the present invention obtains multi-source data of the target object, fuses the multi-source data to obtain fused data, pre-processes the fused data to obtain pre-processed data, obtains prediction results based on the pre-processed data and a trained talent prediction model, and finally determines the talent training plan based on the prediction results, thereby reducing the difficulty and subjectivity of manual screening, and can quickly, effectively and automatically process and screen data, provide support and assistance for determining talent screening results, and help improve work efficiency and accuracy.
[0128] It should be noted that the device is a device corresponding to the above method, and all implementations in the above method embodiment are applicable to the embodiment of the device and can achieve the same technical effects, which will not be described in detail in this embodiment.
[0129] An embodiment of the present invention further provides a computing device comprising: a processor and a memory storing a computer program. When the computer program is executed by the processor, the computer program performs the method described in any of the above embodiments. All implementations in the above method embodiments are applicable to the embodiments of this device and can achieve the same technical effects. These are not further described in this embodiment.
[0130] An embodiment of the present invention further provides a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the method described in any of the above embodiments. All implementations in the above method embodiments are applicable to the embodiments of the device and can achieve the same technical effects. These are not further described in this embodiment.
[0131] It should be noted that, in the apparatus and method of the present invention, it is apparent that each component or step can be decomposed and / or recombined. Such decomposition and / or recombination should be considered equivalent solutions of the present invention. Furthermore, the steps of performing the above series of processes can naturally be performed in chronological order according to the order described, but do not necessarily need to be performed in chronological order. Certain steps can be performed in parallel, interleaved, or independently of each other.
[0132] It should be noted that, in the above embodiments, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or device comprising the element. In addition, it should be pointed out that the scope of the methods and devices in the implementation methods of the above embodiments is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in the opposite order according to the functions involved. For example, the described method may be performed in an order different from that described, and various steps may also be added, omitted, or combined. In addition, the features described with reference to certain examples may be combined in other examples.
[0133] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.
Claims
1. A method for processing talent information, characterized in that: include: Obtain multi-source data of the target object; Preprocessing the multi-source data to obtain preprocessed data; Performing fusion processing on the pre-processed data to obtain fused data; Obtaining a prediction result based on the fused data and the trained talent prediction model; The talent prediction model is obtained by training a preset network model based on collected historical multi-source data; A talent screening result is determined based on the prediction result.
2. The method for processing talent information according to claim 1, characterized in that: Get multi-source data of the target object, including: Identify target audiences; Acquire multi-source data of the target object; the multi-source data includes basic data, education data, work data, academic achievement data and influence data of the target object.
3. The method for processing talent information according to claim 1, wherein: Preprocessing the multi-source data to obtain preprocessed data includes: Determining a preprocessing method based on the multi-source data; The multi-source data is preprocessed according to the preprocessing method to obtain preprocessed data.
4. The method for processing talent information according to claim 1, wherein: Performing fusion processing on the pre-processed data to obtain fused data includes: Determining a fusion method according to the dimension of the preprocessed data; The pre-processed data is fused according to the fusion method to obtain fused data.
5. The method for processing talent information according to claim 1, wherein: Based on the fused data and the trained talent prediction model, a prediction result is obtained, including: Inputting the fused data into the first processing layer of the talent prediction model to obtain a first output result; wherein the first processing layer includes z1=W1Y+b1, h1=ReLU(z1); wherein z1 is the inactivated output of the first processing layer, W1 is the weight matrix of the first processing layer, Y is the fused data, b1 is the bias term of the first processing layer, h1 is the first output result, and ReLU is the activation function; The first output result is input into the second processing layer of the talent prediction model to obtain a second output result; wherein the second processing layer includes in, is the inactivated output of the second processing layer, W2 is the weight matrix of the second processing layer, h1 is the first output result, b2 is the bias term of the second processing layer, is the second output result, ε is the activation function; A prediction result is determined according to the second output result and a preset threshold.
6. The method for processing talent information according to claim 5, characterized in that: The training process of the talent prediction model includes: Collect historical multi-source data; the historical multi-source data includes basic data, education data, work data, academic achievement data, and influence data; Determining training samples based on the historical multi-source data; Train the preset network model according to the training samples to obtain a trained model; According to m t =β1m t-1 +(1-β1)g t 、 The trained model is optimized to obtain a talent prediction model; wherein, m t is the gradient mean of the t-th iteration, β1 is the first-order moment decay rate, m t-1 is the mean gradient of the t-1th iteration, g t is the gradient of the tth iteration, v t is the second-order moment estimate of the gradient at the t-th iteration, β2 is the second-order moment decay rate, is the square of the gradient of the tth iteration, v t-1 is the second-order moment estimate of the gradient at the t-1th iteration, is the revised first-order moment estimate, t is the current iteration number, as t increases, Approaching 0, is β1 raised to the power of t, is β2 raised to the power of t, is the modified second-order moment estimate, θ t+1 is the updated model parameter, θ t is the model parameter at the tth iteration, γ is the learning rate, and τ is the preset value.
7. The method for processing talent information according to claim 1, wherein: Determine talent screening results based on the prediction results, including: Determine a talent training plan based on the prediction results; Determine the talent screening results based on the talent training plan.
8. A talent information processing device, characterized in that: include: Acquisition module, used for multi-source data of target objects; A processing module, configured to preprocess the multi-source data to obtain preprocessed data; The pre-processed data is fused to obtain fused data; a prediction result is obtained based on the fused data and a trained talent prediction model; the talent prediction model is obtained by training a preset network model based on the collected historical multi-source data; and a talent screening result is determined based on the prediction result.
9. A computing device, characterized in that include: A processor and a memory storing a computer program, wherein when the computer program is executed by the processor, the method according to any one of claims 1 to 7 is performed.
10. A computer-readable storage medium, characterized in that The device stores instructions, which, when executed on a computer, enable the computer to execute the method according to any one of claims 1 to 7.