Scientific and technological talent evaluation management method and system based on artificial intelligence

Abstract

The invention provides a scientific and technological talent evaluation management method and system based on artificial intelligence, and the method comprises the steps: S1, collecting first historical talent evaluation data of talent units in a plurality of fields, drawing a first grade distribution diagram of the first historical talent evaluation data, and carrying out the normalization of the first grade distribution diagram; s2, training the established artificial intelligence evaluation model of the unit to which the talent belongs according to the normalized first grade distribution diagram and the actual post rating data; s3, collecting second historical talent evaluation data of talent evaluation corresponding to the same post rating in the field by each expert to form a plurality of corresponding second grade distribution diagrams; and S4, making a talent cultivation direction and a talent strategy adjustment basis according to a condition proportion condition corresponding to the final actual post rating data in the predicted post rating data. According to the invention, a more objective talent evaluation scheme is obtained through comparison of intelligent rating of the unit and rating of different experts, so that a correct talent cultivation and human resource adjustment strategy is correspondingly made.

Description

Technical Field

[0001] The present invention relates to a talent evaluation management method and system, and in particular to a scientific and technological talent evaluation management method and system based on artificial intelligence. Background Art

[0002] The dimensions involved in the evaluation of talents include academic qualifications, publications, work experience, performance, peer evaluation, as well as IQ and EQ. However, existing technologies only evaluate emotions, which is accurate for areas that require psychological assistance, but not for some industries that are not inclined to emotional evaluation, such as scientific research in the field of natural sciences, company research and development, etc. Therefore, how to intelligently evaluate the evaluation factors of multiple dimensions is an urgent problem to be solved.

[0003] In addition, it is noted that even in the same field, different experts may be selected for evaluation of talents in the same field due to their different experiences, different projects and the requirements of different enterprises. Excellent experts in the field may not necessarily meet the standard requirements of the enterprise for evaluating the talents of the enterprise. Therefore, the experts who evaluate talents themselves also need to be screened and evaluated, which is a consideration that the existing technology does not have. Summary of the invention

[0004] In order to solve the above problems, the present invention will consider the following points: first, collect historical talent evaluation data in different fields and analyze the priority levels of talent evaluation factors in this field; second, establish an artificial intelligence evaluation system according to different priority levels; third, in order to balance the talent evaluation cognition of different experts in this field, study and analyze the distribution of emphasis of different experts in talent evaluation factors, so as to compensate for the balance of different experts in talent evaluation.

[0005] In view of the above considerations, the present invention provides, on the one hand, a scientific and technological talent evaluation and management method based on artificial intelligence, comprising the following steps: S1 collects the first historical talent evaluation data of the units to which talents belong in multiple fields, divides the first historical talent evaluation data into evaluation factors of multiple dimensions, sets the level of each evaluation factor, obtains actual job rating data, draws the first level distribution map of the first historical talent evaluation data, and normalizes it; S2 For the same field, the artificial intelligence evaluation model of the unit to which the talent belongs is trained based on the normalized first-level distribution map and actual job rating data; S3 collects, for each field, the second historical talent evaluation data of the talent evaluation corresponding to the same position rating in the field by each expert, forms a plurality of corresponding second level distribution graphs according to step S1, and inputs them one by one into the artificial intelligence evaluation model trained in step S2 after normalization to obtain a plurality of predicted position rating data; S4 takes the average of each predicted job rating data and the actual job rating data to obtain corresponding multiple groups of final job rating data, marks the talents whose predicted job rating data is higher than the corresponding final actual job rating data, and substitutes the normalized first-level distribution map of the talent into the artificial intelligence evaluation model of other units to obtain the corresponding additional multiple predicted job rating data, and takes the highest rating among the additional multiple predicted job rating data as the direction of talent training; similarly marks the talents with a small proportion of the situation, and substitutes the normalized first-level distribution map of the talent into the artificial intelligence evaluation model of other units to obtain the highest rating among the corresponding additional multiple predicted job rating data as the basis for adjusting the talent strategy of the affiliated unit.

[0006] Therefore, the rating of the talent in each position may not be the most objective in the unit to which they belong, and there may be different evaluations in other units. Therefore, it is necessary not only to evaluate in the same unit, but also to see how other experts in the same field and position level evaluate, so as to achieve a more accurate and objective evaluation of talents; and through the majority of high-rated information in the evaluation of each expert, the actual evaluation status of the low or high evaluation in the unit can be obtained, so as to adjust the training direction or talent strategy adjustment basis. For example, if the evaluation is low, further emphasis will be placed on training and giving full play to the initiative of the talent, while if the evaluation is high, the talent needs to be re-examined, and the characteristics of the product or service of the unit with the highest rating will be studied, so as to adjust the talent to a work direction that is more suitable for his ability or find other more suitable talents.

[0007] Optionally, the affiliated units include corporate units and public institutions.

[0008] Optionally, the evaluation factors include academic qualifications, publications, work experience, performance, peer evaluation, as well as IQ and EQ.

[0009] Optionally, different colors are used to represent different levels, and any one of bar charts, pie charts, and bubble charts is used as the type of first-level distribution charts and second-level distribution charts. A convolutional neural network (CNN) or a recurrent neural network (RNN) is used to build an artificial intelligence evaluation model for the talent's unit, and the high level of an evaluation factor is normalized as 1.

[0010] Optionally, the training method comprises: S2-1 divides the normalized first-level distribution map into a training set and a validation set, with the ratio of (5-3): (1-2); S2-2 Construct a convolutional neural network (CNN) or a recurrent neural network (RNN), use the training set for training, the validation set to verify the job rating prediction accuracy, calculate the loss function and back-propagate to optimize the model parameters until the job rating accuracy is the highest and stop training.

[0011] Another object of the present application is to provide an artificial intelligence-based scientific and technological talent evaluation and management system, comprising a computer, wherein the computer comprises a memory storing a computer-readable program capable of implementing the aforementioned method.

[0012] The artificial intelligence-based scientific and technological talent evaluation management method and system of the present invention have the following beneficial effects: By comparing the intelligent ratings of affiliated units with the ratings of different experts, a more objective talent evaluation plan can be obtained, so that correct talent training and human resource adjustment strategies can be made accordingly. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 It is a flow chart of a scientific and technological talent evaluation management method based on artificial intelligence according to an embodiment of the present invention; Figure 2 It is a flowchart of steps S2-S4 in the solution of an embodiment of the present invention. DETAILED DESCRIPTION

[0014] Figure 1 This is a flow chart of the scientific and technological talent evaluation management method based on artificial intelligence, including the following steps: S1 collects the first historical talent evaluation data of the units to which talents belong in multiple fields, divides the first historical talent evaluation data into evaluation factors of multiple dimensions, sets the level of each evaluation factor, obtains actual job rating data, draws the first level distribution map of the first historical talent evaluation data, and normalizes it; The fields here include multiple fields divided according to natural sciences and social sciences, according to patent technologies, and according to the functional positions of enterprises and institutions. For example, in the education system, there are lecturers, associate professors, professors, and principals. For example, in a company, there are chairman, general manager, manager, department head, sales, R&D department leader, R&D team member, clerk, secretary, accounting, market research department personnel, etc. The specific division can be made according to the standards selected by the unit to which the talent belongs. It is preferred to divide the fields to which the talent belongs according to the above three methods. The first historical talent evaluation data here is the evaluation data of individual talents, and the units to which they belong include enterprises and institutions.

[0015] The evaluation factors include academic qualifications, publications, work experience, performance, peer evaluation, and IQ and EQ. Publications include published journal articles, published books, program writing, integrated circuit layout author, etc. Performance includes past years of excellence evaluation, job assessment, peer evaluation includes annual grading in the unit to which they belong, and the review opinions on the works in the same field, etc.

[0016] S2 For the same field, the artificial intelligence evaluation model of the unit to which the talent belongs is trained based on the normalized first-level distribution map and actual job rating data; S3 collects, for each field, the second historical talent evaluation data of the talent evaluation corresponding to the same position rating in the field by each expert, forms a plurality of corresponding second level distribution graphs according to step S1, and inputs them one by one into the artificial intelligence evaluation model trained in step S2 after normalization to obtain a plurality of predicted position rating data; S4 takes the average of each predicted job rating data and the actual job rating data to obtain corresponding multiple groups of final job rating data, marks the talents whose predicted job rating data is higher than the corresponding final actual job rating data, and substitutes the normalized first-level distribution map of the talent into the artificial intelligence evaluation model of other units to obtain the corresponding additional multiple predicted job rating data, and takes the highest rating among the additional multiple predicted job rating data as the direction of talent training; similarly marks the talents with a small proportion of the situation, and substitutes the normalized first-level distribution map of the talent into the artificial intelligence evaluation model of other units to obtain the highest rating among the corresponding additional multiple predicted job rating data as the basis for adjusting the talent strategy of the affiliated unit.

[0017] The process of steps S2-S4 is summarized in Figure 2 Among them, N experts evaluate two types of talents, A and B, from the same field and different actual positions in two units.

[0018] The N experts for each type of evaluation may be different from each other, or not completely the same; and the experts between the two types are different from each other. Figure 2 In the figure, the level is used as the vertical axis and the evaluation factor is used as the horizontal axis. From left to right, they are academic qualifications, works, work experience, performance, peer evaluation, as well as IQ and EQ. A columnar normalized first and second level distribution diagram is formed. Except for the boxes in the figure, which indicate that academic qualifications and works (mainly based on circulation or impact factor) are stable. In the normalized columnar second level distribution diagram, the degree of recognition of work experience, performance, peer evaluation, IQ, and EQ by N different experts will be different, which ultimately leads to deviations in the ratings of the affiliated units between experts and between different units. In this application, the experts can be professional managers, personnel from the human resources department of the affiliated unit, or social science researchers, etc.

[0019] Therefore, in the evaluation of many talents, due to the different emphases and angles of experts, personal research preferences, and different experiences, the evaluation of the same type of talents is different. Therefore, after the evaluation of different experts, a variety of intelligent job rating prediction results are generated, and the objective final job rating results are obtained after averaging with the actual job rating of the unit to which the talent belongs. At the same time, the difference between the final rating and the predicted job rating is still observed. For the cases where the predicted rating is higher than the final actual job rating, it means that the rating of the talent in the unit is low, otherwise it is high. Therefore, targeted training and talent strategy adjustments are given to solve the problem of more objective talent evaluation.

[0020] The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A scientific and technological talent evaluation and management method based on artificial intelligence, characterized in that: The steps include: S1 collects the first historical talent evaluation data of the units to which talents belong in multiple fields, divides the first historical talent evaluation data into evaluation factors of multiple dimensions, sets the level of each evaluation factor, obtains actual job rating data, draws the first level distribution map of the first historical talent evaluation data, and normalizes it; S2 For the same field, the artificial intelligence evaluation model of the unit to which the talent belongs is trained based on the normalized first-level distribution map and actual job rating data; S3 collects, for each field, the second historical talent evaluation data of the talent evaluation corresponding to the same position rating in the field by each expert, forms a plurality of corresponding second level distribution graphs according to step S1, and inputs them one by one into the artificial intelligence evaluation model trained in step S2 after normalization to obtain a plurality of predicted position rating data; S4 takes the average of each predicted job rating data and the actual job rating data to obtain corresponding multiple groups of final job rating data, marks the talents whose predicted job rating data is higher than the corresponding final actual job rating data, and substitutes the normalized first-level distribution map of the talent into the artificial intelligence evaluation model of other units to obtain the corresponding additional multiple predicted job rating data, and takes the highest rating among the additional multiple predicted job rating data as the direction of talent training; similarly marks the talents with a small proportion of the situation, and substitutes the normalized first-level distribution map of the talent into the artificial intelligence evaluation model of other units to obtain the highest rating among the corresponding additional multiple predicted job rating data as the basis for adjusting the talent strategy of the affiliated unit.

2. The method according to claim 1, characterized in that The affiliated units include corporate units and public institutions.

3. The method according to claim 1, characterized in that The evaluation factors include academic qualifications, publications, work experience, performance, peer evaluation, as well as IQ and EQ.

4. The method according to claim 3, characterized in that Different colors represent different levels, and any of the bar charts, pie charts, and bubble charts are used as the types of first-level distribution charts and second-level distribution charts. The convolutional neural network CNN or the recurrent neural network RNN ​​is used to build an artificial intelligence evaluation model for the talent's unit, and the high level of an evaluation factor is normalized as 1.

5. The method according to any one of claims 1 to 4, characterized in that The training method of the artificial intelligence evaluation model includes: S2-1 divides the normalized first-level distribution map into a training set and a validation set, with the ratio of (5-3): (1-2); S2-2 Construct a convolutional neural network (CNN) or a recurrent neural network (RNN), use the training set for training, the validation set to verify the job rating prediction accuracy, calculate the loss function and back-propagate to optimize the model parameters until the job rating accuracy is the highest and stop training.

6. A scientific and technological talent evaluation and management system based on artificial intelligence, characterized in that: The method comprises a computer comprising a memory in which a computer-readable program capable of implementing the method according to any one of claims 1 to 5 is stored.

Images