Learning preference analysis method based on big data information analysis

Abstract

The invention relates to the field of learning preference analysis, in particular to a learning preference analysis method based on big data information analysis, and the method comprises the following steps: S1, obtaining basic information of a target employee, carrying out the data processing of the basic information, and generating a first basic information quantification parameter set; s2, calculating a first learning preference of the target staff for each type of learning content according to the first basic information quantization parameter set; s3, calculating a preference influence coefficient used for representing the influence of other employees on the first learning preference of the target employee, and correcting the first learning preference of the target employee to generate a second learning preference; compared with a common calculation method, the method has the advantages that the influence of surrounding colleagues on the learning preference of the target staff is considered, the preference influence coefficient is calculated according to the similarity between other staff and the target staff, the calculation precision of the second learning preference can be further improved, and the calculation efficiency is improved. And thus, more accurate learning content pushing can be carried out on the target staff.

Description

Technical Field

[0001] The present invention relates to the field of learning preference analysis, and specifically to a learning preference analysis method based on big data information analysis. Background Art

[0002] Modern mines have higher and higher requirements for employees, and the knowledge that employees need to learn also increases accordingly. In order to ensure that each employee can quickly adapt to the equipment of the post and the safety and environmental protection requirements of national policies, the mine will provide a learning platform for employees to browse and learn the corresponding content.

[0003] In order to improve the user experience and learning efficiency, the learning platform generally creates user portraits based on the basic information of employees and actively pushes some learning content according to the user portraits. However, this method does not consider the factor that the learning tendencies of colleagues around employees will affect the learning preferences of employees, resulting in the learning content pushed to employees not meeting the needs of employees, thereby reducing the user experience and learning efficiency. Summary of the Invention

[0004] In view of the deficiencies of the prior art, the present invention provides a learning preference analysis method based on big data information analysis, which solves the technical problem that the influence of colleagues on the learning preferences of employees is not considered in the prior art, resulting in the learning content pushed not meeting the needs of employees.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A learning preference analysis method based on big data information analysis, comprising the following steps:

[0007] S1. Obtain the basic information of the target employee and perform data processing on it to generate the first basic information quantization parameter set A;

[0008] S2. Calculate the first learning preference of the target employee for each type of learning content according to the first basic information quantization parameter set A

[0009] S3. Calculate the preference influence coefficient I representing the influence of other employees on the first learning preference of the target employee and correct the first learning preference of the target employee ij to generate the second learning preference

[0010] S4. Perform learning content recommendation according to the second learning preference

[0011] Further, the basic information includes: education level, job position, working hours, and learning content. ​​

[0012] Further, in step S1, it specifically includes the following steps:

[0013] S11. Divide all educational attainment levels into several categories, assign an educational attainment parameter to each category, and match the corresponding educational attainment parameter according to the employee's educational attainment;

[0014] S12. Assign a position parameter to each type of job position, and construct a historical position parameter based on the job positions the employee has held in the past;

[0015] S13. Calculate the active learning influence factor S according to the learning content j ;

[0016] S14. Construct the first basic information quantization parameter set A based on the educational attainment parameter, historical position parameter, working hours, and active learning influence factor of each employee, A = {a1, a2, a3, a4}, where a1, a2, a3, and a4 represent the educational attainment parameter, historical position parameter, working hours, and active learning influence factor S respectively j .

[0017] Further, in step S13, it specifically includes the following steps:

[0018] S131. Define the search basic weight w s and the browsing basic weight w b ;

[0019] S132. Calculate the basic weight factor w s based on the search basic weight w b and the browsing basic weight w i , and its calculation formula is:

[0020]

[0021] In the formula, s i represents the flag indicating whether the i-th learning content is a search behavior. If so, s i is 1; if not, s i is 0; d i represents the residence time of the employee on the i-th learning content; d max represents the maximum residence time threshold; min represents the min function;

[0022] S133. Calculate the time decay factor w t based on the browsing time of the learning content, and its calculation formula is:

[0023]

[0024] Wherein, e represents the base of the natural logarithm; λ represents the time decay coefficient; T represents the current time point; t i represents the time point of the learning content;

[0025] S134. Classify all browsable learning contents and calculate the active learning influence factor S corresponding to each learning content for each type of learning content j , and its calculation formula is:

[0026]

[0027] Wherein, n represents the total number of employees' learning contents; f ij represents the flag indicating whether the i-th learning content belongs to the j-th category. If so, f ij is 1; if not, f ij is 0.

[0028] Furthermore, in step S2, it specifically includes the following steps:

[0029] S21. Obtain the second set of quantified basic information parameters B of all historical employees and the third learning preferences of each historical employee at the corresponding time point, and construct a training set therewith, B = {b1, b2, b3, b4};

[0030] S22. Use the training set to train the preference analysis model to obtain a target model, and analyze the fourth learning preferences of the target user for each type ;

[0031] S23. Calculate the comprehensive similarity S(a i , b i ) between the first set of quantified basic information parameters A of the target employee and the second set of quantified basic information parameters B of each historical employee;

[0032] S24. Select several historical employees with the highest comprehensive similarity S(a i , b i ) to the target employee, and calculate the offset parameter ΔP w ;

[0033] S25. Calculate the first learning preference k according to the offset parameter ΔP , and its calculation formula is:

[0034]

[0035] Wherein, represents the preference degree of the target employee for the k-th type of learning content.

[0036] Furthermore, in step S23, it specifically includes the following steps:

[0037] S231. Calculate the similarity of educational attainment parameters based on the educational attainment parameters of the target employee and historical employees Its calculation formula is:

[0038]

[0039] In the formula, and respectively represent the educational attainment parameters of the i-th target employee and the j-th historical employee; e represents the base of the natural logarithm; σ represents the similarity decay rate parameter;

[0040] S232. Calculate the similarity of historical position parameters based on the historical position parameters of the target employee and historical employees Its calculation formula is:

[0041]

[0042] In the formula, and respectively represent the historical position parameters of the i-th target employee and the j-th historical employee; represents the similarity decay rate coefficient;

[0043] S233. Calculate the similarity of working hours based on the working hours of the target employee and historical employees Its calculation formula is:

[0044]

[0045] In the formula, and respectively represent the working hours of the i-th target employee and the j-th historical employee; max(T S ) and min(T S ) respectively represent the maximum and minimum working hours among the target employees;

[0046] S234. Calculate the similarity of the active learning factor based on the active learning impact factor S j of the target employee and historical employees Its calculation formula is:

[0047]

[0048] In the formula, and respectively represent the active learning impact factor S j of the i-th target employee and the j-th historical employee; max(A S ) and min(A Srespectively represent the maximum and minimum active learning influence factors S among the target employees j ;

[0049] S235, calculate the comprehensive similarity S(a i , b i ), and its calculation formula is:

[0050]

[0051] In the formula, W E , W J , W T and W A respectively represent the first, second, third, and fourth weight parameters regarding S(a i , b i ).

[0052] Furthermore, in step S24, the calculation formula for the offset parameter ΔP k is:

[0053]

[0054] In the formula, L represents the total number of historical employees with the highest similarity to the target employee; P i，k represents the third learning preference of the i-th historical employee for the k-th type of learning content; represents the mean of the third learning preferences of L historical employees for the k-th type of learning content.

[0055] Furthermore, in step S3, it specifically includes the following steps:

[0056] S31. Obtain the basic information of other employees in the same mining area as the target employee, and calculate the age similarity S age (i, j) between the target employee and each other employee, and its calculation formula is:

[0057]

[0058] In the formula, S age (i, j) represents the age similarity between the target employee and the i-th other employee; age i and age j respectively represent the ages of the target employee and the i-th other employee; max(Δage) represents the maximum difference in the age range of the mine;

[0059] S32. Calculate the position similarity S job (i, j) between the target employee and each other employee, and its calculation formula is:

[0060]

[0061] Wherein, S job (i, j) represents the job similarity between the target employee and the i-th other employee; job i and job j represent the job parameters of the target employee and the i-th other employee respectively; max(Δjob) represents the maximum difference in job parameters;

[0062] S33. Calculate the educational background similarity S edu (i, j) between the target employee and each other employee, and its calculation formula is:

[0063]

[0064] Wherein, S edu (i, j) represents the educational background similarity between the target employee and the i-th other employee; edu i and edu j represent the educational attainment parameters of the target employee and the i-th other employee respectively; max(Δedu) represents the maximum difference in educational attainment parameters;

[0065] S34. Calculate the employee similarity S ij between the target employee and each other employee, and its calculation formula is:

[0066] S ij = w age × S age (i, j) + w job × S job (i, j) + w edu × S edu (i, j)

[0067] Wherein, S ij represents the employee similarity between the target employee and the i-th other employee; w age , w job and w edu represent the weight parameters for S age (i, j), S job (i, j) and S edu (i, j) respectively;

[0068] S35. Calculate the personalized weight γ ij according to the employee similarity S ij , and its calculation formula is:

[0069] γ ij = exp(-β × (1 - S ij ))

[0070] Wherein, β represents the decay factor for γ ij ;

[0071] S36. According to the personalized weight γ ij Calculate the preference influence coefficient I of each other employee on the target employee ij ;

[0072] S37. According to the preference influence coefficient I ij Calculate the second learning preference

[0073] Furthermore, in step S36, the preference influence coefficient I ij The calculation formula is:

[0074]

[0075] In the formula, L j (k) and L i (k) respectively represent the number of learning times of the i-th other employee and the target employee for the k-th type of learning content; I ij Represents the preference influence coefficient of the i-th other employee on the target employee.

[0076] Furthermore, in step S37, the second learning preference The calculation formula is:

[0077]

[0078] In the formula, Represents the first learning preference of the target employee.

[0079] Compared with the prior art, the present invention provides a learning preference analysis method based on big data information analysis, having the following beneficial effects:

[0080] 1. Compared with common calculation methods, the present invention takes into account the influence of surrounding colleagues on the learning preferences of the target employee, and calculates the preference influence coefficient according to the similarity between other employees and the target employee, which can further improve the calculation accuracy of the second learning preference, so as to push more accurate learning content to the target employee.

[0081] 2. Compared with common employee learning preference analysis methods, when calculating the first learning preference of the target employee, the present invention takes into account that different employees have different basic information, which will lead to different employees choosing different learning content when learning, effectively improving the accuracy of the first learning preference of the target employee, and facilitating the later accurate push of learning content to the target employee.

[0082] 3. Compared with common calculation methods, the present invention can effectively improve the accuracy of the active learning influence factor S j , and further improve the accuracy of calculating the learning preferences of employees in the later stage. BRIEF DESCRIPTION OF THE DRAWINGS

[0083] The accompanying drawings described herein are used to provide a further understanding of the present application and form a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation of the present application. In the drawings:

[0084] Figure 1 It is a flowchart of a learning preference analysis method based on big data information analysis of the present invention. Specific embodiments

[0085] In order to make the above objects, features, and advantages of the present invention more obvious and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. Thereby, a full understanding of how the present application uses technical means to solve technical problems and achieve the realization process of technical effects can be obtained and implemented accordingly.

[0086] Since modern mining operations increasingly use mechanical equipment and automated systems during the mining process, and there is an increasing emphasis on safety and efficiency, workers need to continuously learn knowledge such as equipment operation methods, risk identification, accident prevention, and efficient mining methods during their work to meet the requirements of safe and efficient coal mine mining. In addition, workers also need to learn knowledge of different positions among themselves, which not only helps workers better complete their existing work, but also provides them with promotion opportunities and the possibility of job transfer, which is beneficial to the career development of workers. In order to improve work efficiency, safety, and the skills of workers, as Figure 1 shown, the present invention proposes a learning preference analysis method based on big data information analysis, including the following steps:

[0087] S1. Obtain the basic information of the target worker and perform data processing on it to generate the first basic information quantization parameter set A; specifically, the target worker refers to the worker currently working in the coal mine. During the mining of the coal mine, the worker needs to learn relevant knowledge of the corresponding position, relevant knowledge of work safety, and relevant knowledge of environmental protection, etc. Since different workers have different learning degrees of the above information, their demands for the above knowledge are different. For example, workers who have been in a certain position for many years have much lower demands for the above knowledge than workers who have just started in that position; workers seeking to change positions have higher demands for relevant knowledge of other positions than other workers; in addition, the educational level of workers and the demand for workers' knowledge reserves due to the development of industry technology will both affect the learning preferences of workers. Therefore, the basic information includes: educational level, work position, work duration, and learning content; specifically, the educational level of the worker is based on the academic information in the information entered by the mine, and the work position and work duration of the worker are directly obtained according to the work information table recorded by the mine;

[0088] It should be noted that the learning content includes the information browsed by employees, browsing time, browsing duration, and specific operations (active search or recommended browsing).

[0089] In addition, specific parameters are required for calculation when establishing the first portrait of employees in the later stage, and the basic information is not specific parameters, so it needs to be quantified for calculation. Therefore, in step S1, it specifically includes the following steps:

[0090] S11. Divide all education levels into several categories, assign an education level parameter to each category, and match the corresponding education level parameter according to the employee's education level; specifically, generally, the higher the education level, the higher the learning requirements of the employee, so the higher the education level, the higher the education level parameter. In the present invention, if the employee's education level is primary school or below, the education level parameter is 1; if the employee's education level is junior high school, the education level parameter is 2; if the employee's education level is high school or secondary school, the education level parameter is 3; if the employee's education level is junior college, the education level parameter is 4; if the employee's education level is undergraduate or above, the education level parameter is 5.

[0091] S12. Assign a position parameter to each type of job position, and construct a historical position parameter based on the job positions the employee has held in the past; in the present invention, employees are divided into four categories, namely ordinary employees, technical workers (such as electricians, mechanical maintenance workers), management personnel (such as team leaders, department heads), and professional and technical personnel (such as engineers, safety experts); the position parameters of the above positions are: ordinary employees - 1, technical workers - 2, management personnel - 3, and professional and technical personnel - 4; the historical position parameter is the sum of the position parameters of all the job positions the employee has worked in.

[0092] S13. Calculate the active learning influence factor S according to the learning content j ; specifically, since the content browsed actively by employees in the process of daily learning can better represent the learning tendency of employees than the recommended browsing content, when calculating the learning influence factor according to the learning content, it is necessary to consider the influence of different operation behaviors of employees on the learning influence factor. Therefore, in step S13, it specifically includes the following steps:

[0093] S131. Define the search base weight w s and the browsing base weight w b ; in the present invention, since the content browsed actively can better represent the learning tendency of employees than the recommended browsing content, so w s and w b are 1.5 and 1.0 respectively;

[0094] S132. According to the search base weight w sand the browsing basic weight w b Calculate the basic weight factor w i , and its calculation formula is:

[0095]

[0096] In the formula, s i represents the flag indicating whether the i-th learning content is a search behavior. If so, s i is 1; if not, s i is 0; d i represents the residence time of the employee in the i-th learning content; d max represents the maximum residence time threshold; min represents the min function; in the present invention, d max is 10 minutes;

[0097] S133. Calculate the time decay factor w according to the browsing time of the learning content t , and its calculation formula is:

[0098]

[0099] In the formula, e represents the base of the natural logarithm; λ represents the time decay coefficient; T represents the current time point; t i represents the time point of the learning content;

[0100] S134. Classify all browsable learning contents, and calculate the active learning influence factor S corresponding to each learning content for each type of learning content j , and its calculation formula is:

[0101]

[0102] In the formula, n represents the total number of the employee's learning contents; f ij represents the flag indicating whether the i-th learning content belongs to the j-th category. If so, f ij is 1; if not, f ij is 0; specifically, all browsable learning contents represent all learning contents available for employees to browse and learn on the platform, which are the information of their own positions, non-own position information, safety training information, and environmental protection knowledge information respectively, and they are uniquely encoded, and each code corresponds to a type of learning content.

[0103] S14. Construct the first basic information quantization parameter set A according to the education level parameter, historical position parameter, working hours, and active learning influence factor of each employee. A = {a1, a2, a3, a4}, where a1, a2, a3, and a4 represent the education level parameter, historical position parameter, working hours, and active learning influence factor S respectively j; It should be noted that the working hours are directly represented by the number of working months of the employees.

[0104] In step S1 of the present invention, an education level parameter, a historical position parameter, working hours, and an active learning influence factor S are constructed based on the educational information, working hours, job position, and learning content of the employees j , which is convenient for later analyzing the learning preferences of employees based on it. Among them, when calculating the active learning influence factor S j , considering that the active search and recommended browsing of learning content by employees represent different learning intentions, different basic weights are set for active search and recommended browsing, and the basic weight factor w is calculated based on them i , and also considering that the learning preferences of employees change over time, so the time decay factor w is calculated t , and finally the active learning influence factor S calculated based on the above information j , compared with common calculation methods, can effectively improve the accuracy of the active learning influence factor S j , thereby improving the accuracy of calculating the learning preferences of employees in the later stage.

[0105] S2. Calculate the first learning preference of the target employee for each type of learning content according to the first basic information quantization parameter set A Specifically, when calculating the first learning preference of each employee, a neural network model is generally constructed, and then the first basic information quantization parameter set is input into the neural network model to directly obtain the first learning preference. This method can reflect the learning preferences of most employees. Since each employee is different due to factors such as education level and work experience, their future career plans are also different, resulting in different learning preferences among different employees. Therefore, directly calculating the first learning preference according to the basic weight coefficient cannot adapt to the differences of each employee, resulting in the inability to accurately recommend corresponding learning content for employees based on the first learning preference in the later stage. Therefore, in step S2, it specifically includes the following steps:

[0106] S21. Obtain the second basic information quantization parameter set B of all historical employees and the third learning preference of each historical employee at the corresponding time point and construct a training set with them, B = {b1, b2, b3, b4}; specifically, historical employees refer to employees who have worked in coal mines, and the second basic information quantization parameter set B refers to the education level parameter, historical position parameter, working hours, and active learning influence factor S of the above-mentioned several employees at several past time points j , and the second basic information quantization parameter set is used as a training sample; then, the ratio of each type of learning content learned by historical employees in the later stage to all the learning content is used as the third learning preference of each employee, and the active learning influence factor of each category represents the preference degree for the learning content of this category, and the third learning preference is used as a sample label;

[0107] S22. Use the training set to train the preference analysis model to obtain the target model, and analyze the fourth learning preference of the target user for each type for analysis. and respectively represent the preference degrees of historical employees for job information, non-job information, safety training information, and environmental protection knowledge information; specifically, the preference analysis model is constructed based on the long short-term memory network.

[0108] S23. Calculate the comprehensive similarity S(a i , b i ) between the first set of basic information quantization parameters A of the target employee and the second set of basic information quantization parameters B of each historical employee; specifically, in order to make the first learning preference closer to the actual situation of the target employee, some historical employees similar to the target employee can be selected, and the fourth learning preference can be corrected according to their similarity. Therefore, in step S23, the following steps are specifically included:

[0109] S231. Calculate the similarity of the education level parameters according to the education level parameters of the target employee and the historical employee The calculation formula is:

[0110]

[0111] In the formula, and respectively represent the education level parameters of the i-th target employee and the j-th historical employee; e represents the base of the natural logarithm; σ represents the similarity attenuation rate parameter; in the present invention, σ is 0.15.

[0112] S232. Calculate the similarity of the historical position parameters according to the historical position parameters of the target employee and the historical employee The calculation formula is:

[0113]

[0114] In the formula, and respectively represent the historical position parameters of the i-th target employee and the j-th historical employee; represents the similarity attenuation rate coefficient; in the present invention, is 0.18;

[0115] S233. Calculate the duration similarity according to the working hours of the target employee and the historical employee The calculation formula is:

[0116]

[0117] In the formula, and respectively represent the working hours of the i-th target employee and the j-th historical employee; max(T S ) and min(T S ) respectively represent the maximum and minimum working hours among the target employees;

[0118] S234. Calculate the similarity of the active learning factor according to the active learning influence factors S j of the target employee and the historical employee The calculation formula thereof is:

[0119]

[0120] In the formula, and respectively represent the active learning influence factors S j of the i-th target employee and the j-th historical employee; max(A S ) and min(A S ) respectively represent the maximum and minimum active learning influence factors S j among the target employees;

[0121] S235. Calculate the comprehensive similarity S(a i , b i ), and the calculation formula thereof is:

[0122]

[0123] In the formula, W E , W J , W T and W A respectively represent the first, second, third, and fourth weight parameters for S(a i , b i ); in the present invention, W E , W J , W T and W A are 0.2, 0.3, 0.1, and 0.4 respectively.

[0124] S24. Select several historical employees with the highest comprehensive similarity S(a i , b i ) to the target employee, and calculate the offset parameter ΔP w according to them; specifically, in order to adapt to the differences between different employees, some historical employees most similar to the target employee can be selected, and these historical employees may learn similar learning contents. Therefore, in step S24, the calculation formula of the offset parameter ΔP k is:

[0125]

[0126] In the formula, L represents the total number of historical employees with the highest similarity to the target employee; P i，k represents the third learning preference of the i-th historical employee for the k-th type of learning content; represents the mean value of the third learning preferences of L historical employees for the k-th type of learning content.

[0127] S25. Calculate the first learning preference according to the offset parameter ΔP k The calculation formula is as follows: The calculation formula is:

[0128]

[0129] In the formula, represents the preference degree of the target employee for the k-th type of learning content.

[0130] In step S2 of the present invention, after obtaining the fourth learning preference of the target employee, considering that each employee has different educational backgrounds, work experiences, etc., their future career plans are also different, which will lead to different learning preferences of different employees. Therefore, some historical employees most similar to the target employee are selected. These historical employees may learn similar learning contents, and the comprehensive similarity S(a i , b i ) between the target employee and the historical employees is calculated, and then the fourth learning preference is adjusted according to the comprehensive similarity S(a i , b i ) to obtain the first learning preference Compared with the common methods for analyzing employees' learning preferences, when calculating the first learning preference of the target employee, considering that different employees have different basic information, which will lead to different learning contents selected by employees during learning, the accuracy of the first learning preference of the target employee is effectively improved, facilitating the accurate push of learning contents to the target employee in the later stage.

[0131] S3. Calculate the preference influence coefficient I representing the influence of other employees on the first learning preference of the target employee ij , and correct the first learning preference of the target employee to generate the second learning preference Specifically, since the learning preference of the target employee is also affected by the surrounding employees. For example, when other colleagues learn a certain type of learning content, the possibility of some employees who are easily influenced by their surrounding colleagues to learn this type of learning content will increase significantly. Ignoring this factor will affect the accuracy of the push of learning contents. Therefore, in step S3, the following steps are specifically included:

[0132] S31. Obtain the basic information of other employees in the same mining area as the target employee, and calculate the age similarity S between the target employee and each other employee age (i, j), and its calculation formula is:

[0133]

[0134] In the formula, S age (i, j) represents the age similarity between the target employee and the i-th other employee; age i and age j respectively represent the ages of the target employee and the i-th other employee; max(Δage) represents the maximum difference in the age range of the mine; in the present invention, max(Δage) is 50;

[0135] S32. Calculate the position similarity S between the target employee and each other employee job (i, j), and its calculation formula is:

[0136]

[0137] In the formula, S job (i, j) represents the position similarity between the target employee and the i-th other employee; job i and job j respectively represent the position parameters of the target employee and the i-th other employee; max(Δjob) represents the maximum difference in the position parameters;

[0138] S33. Calculate the education similarity S between the target employee and each other employee edu (i, j), and its calculation formula is:

[0139]

[0140] In the formula, S edu (i, j) represents the education similarity between the target employee and the i-th other employee; edu i and edu j respectively represent the education level parameters of the target employee and the i-th other employee; max(Δedu) represents the maximum difference in the education level parameters;

[0141] S34. Calculate the employee similarity S between the target employee and each other employee ij , and its calculation formula is:

[0142] S ij = w age × S age (i, j)+ w job × S job(i, j) + w edu × S edu (i, j)

[0143] Wherein, S ij represents the employee similarity between the target employee and the i-th other employee; w age , w job and w edu respectively represent the weight parameters for S age (i, j), S job (i, j) and S edu (i, j); in the present invention, w age , w job and w edu are 0.3, 0.4, and 0.3 respectively;

[0144] S35. Calculate the personalized weight γ ij according to the employee similarity S ij , and its calculation formula is:

[0145] γ ij = exp(-β × (1 - S ij ))

[0146] Wherein, β represents the decay factor for γ ij ; in the present invention, β is 0.5;

[0147] S36. Calculate the preference influence coefficient I ij of each other employee on the target employee according to the personalized weight γ ij , and its calculation formula is:

[0148]

[0149] Wherein, L j (k) and L i (k) respectively represent the number of learning times of the i-th other employee and the target employee for the k-th type of learning content; I ij represents the preference influence coefficient of the i-th other employee on the target employee;

[0150] S37. Calculate the second learning preference ij according to the preference influence coefficient I , and its calculation formula is:

[0151]

[0152] Wherein, represents the first learning preference of the target employee.

[0153] In step S3 of the present invention, considering that the learning preferences of employees are very likely to be influenced by their surrounding colleagues, among which the employees with conditions similar to those of the target employee are undoubtedly the ones most likely to affect the learning preferences of the target employee. Therefore, the personalized weight γ of each other employee is calculated based on this ij , and then the preference influence coefficient I is calculated according to it ij , and finally the second learning preference is calculated Compared with the common calculation methods, it takes into account the influence of surrounding colleagues on the learning preferences of the target employee, and calculates the preference influence coefficient according to the similarity between other employees and the target employee, which can further improve the calculation accuracy of the second learning preference, so as to push more accurate learning content to the target employee.

[0154] S4. Recommend learning content according to the second learning preference . After calculating the second learning preference of the target employee , accurate learning content can be pushed to the target employee according to it, which can effectively save the time for the target employee to search for corresponding content on the learning platform, so as to improve learning efficiency and user experience.

[0155] Those of ordinary skill in the art can understand that all or part of the steps in the above embodiments can be completed by instructing relevant hardware through a program. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0156] The above embodiments have introduced the present invention in detail. Specific examples are used in this article to elaborate on the principles and implementation methods of the present invention. The descriptions of the above embodiments are only used to help understand the method and its core idea of the present invention; at the same time, for those of ordinary skill in the art, according to the idea of the present invention, there will be changes in the specific implementation methods and application scopes. In summary, the content of this specification should not be construed as a limitation to the present invention.

Claims

1. A learning preference analysis method based on big data information analysis, characterized in that, Including the following steps: S1. Obtain the basic information of the target employee, perform data processing on it, and generate the first set of basic information quantization parameters A; S2. Calculate the first learning preference of the target employee for each type of learning content according to the first basic information quantization parameter set A S3. Calculate the first learning preference of other employees for the target employee The preference influence coefficient I that causes an impact ij , and correct the first learning preference of the target employee to generate the second learning preference S4. Recommend learning content according to the second learning preference ​ 2. The learning preference analysis method according to claim 1, wherein The basic information includes: education level, job position, working hours, and learning content.

3. The learning preference analysis method according to claim 1, wherein In step S1, it specifically includes the following steps: S11. Divide all education levels into several categories, assign an education level parameter to each category, and match the corresponding education level parameter according to the employee's education level; S12. Assign a job parameter to each job position, and construct a historical job parameter based on the job positions the employee has held in the past; S13. Calculate the active learning influence factor S according to the learning content j ; S14. Construct the first set A of quantified basic information parameters according to the education level parameter, historical position parameter, working hours, and active learning influence factor of each employee. A = {a1, a2, a3, a4}, where a1, a2, a3, and a4 represent the education level parameter, historical position parameter, working hours, and active learning influence factor respectively.S j .

4. The learning preference analysis method according to claim 3, wherein In step S13, it specifically includes the following steps: S131. Define the search base weight w s and the browsing base weight w b ; S132. Calculate the basic weight factor w s based on the search basic weight w b and the browsing basic weight w i . The calculation formula is as follows: Where s i represents a flag indicating whether the i-th learning content is a search behavior. If so, s i is 1; if not, s i is 0; d i represents the residence time of the employee in the i-th learning content; d max represents the maximum residence time threshold; min represents the min function; S133. Calculate the time decay factor w based on the browsing time of the learning content t , and its calculation formula is: where e represents the base of the natural logarithm; λ represents the time decay coefficient; T represents the current time point; t i represents the time point of the learning content; S134. Classify all browsable learning content and calculate the active learning influence factor S corresponding to each type of learning content for each piece of learning content j , and its calculation formula is: Where n represents the total number of the learning content of employees; f ij represents the flag indicating whether the i-th learning content belongs to the j-th category. If so, f ij is 1; if not, f ij is 0.

5. The learning preference analysis method according to claim 1, wherein In step S2, it specifically includes the following steps: S21. Obtain the second set of basic information quantization parameters B of all historical employees and the third learning preference of each historical employee at the corresponding time point, and construct a training set based on this, B = {b1, b2, b3, b4}; S22. Use the training set to train the preference analysis model to obtain a target model, and analyze the fourth learning preference of the target user for each type for analysis; S23. Calculate the comprehensive similarity S(a i , b i ) between the first set of quantified basic information parameters A of the target employee and the second set of quantified basic information parameters B of each historical employee; S24. Select several historical employees with the highest comprehensive similarity S(a i , b i ) to the target employee, and calculate the offset parameter ΔP based on them w ; S25. Calculate the first learning preference according to the offset parameter ΔP k The calculation formula thereof is as follows: ​ In the formula, represents the preference degree of the target employee for the k-th type of learning content.

6. The learning preference analysis method according to claim 5, wherein In step S23, it specifically includes the following steps: S231. Calculate the similarity of the education level parameters based on the education level parameters of the target employee and the historical employees The calculation formula thereof is as follows: In the formula, and respectively represent the education level parameters of the i-th target employee and the j-th historical employee; e represents the base of the natural logarithm; σ represents the similarity decay rate parameter; S232. Calculate the historical post parameter similarity based on the historical post parameters of the target employee and the historical employees Its calculation formula is as follows: Wherein, and respectively represent the historical position parameters of the i-th target employee and the j-th historical employee; represents the similarity attenuation rate coefficient; S233. Calculate the duration similarity based on the working hours of the target employee and historical employees The calculation formula is as follows: In the formula, and respectively represent the working hours of the i-th target employee and the j-th historical employee; max(T S ) and min(T S ) respectively represent the maximum and minimum working hours among the target employees; S234. Calculate the similarity of active learning factors based on the active learning impact factor S of the target employee and the historical employees j Calculate the similarity of active learning factors The calculation formula thereof is as follows: Wherein, and respectively represent the active learning influence factors S of the i-th target employee and the j-th historical employee j ; max(A S ) and min(A S ) respectively represent the maximum and minimum active learning influence factors S among the target employees j ; S235. Calculate the comprehensive similarity S(a i , b i ), and its calculation formula is as follows: Where, W E , W J , W T and W A respectively represent the first, second, third, and fourth weight parameters with respect to S(a i , b i ).

7. The learning preference analysis method according to claim 5, wherein In step S24, the offset parameter ΔP k is calculated by the following formula: where L represents the total number of historical employees with the highest similarity to the target employee; P i，k represents the third learning preference of the i-th historical employee for the k-th type of learning content; represents the mean value of the third learning preferences of L historical employees for the k-th type of learning content.

8. The learning preference analysis method according to claim 1, wherein In step S3, it specifically includes the following steps: S31. Obtain the basic information of other employees in the same mining area as the target employee, and calculate the age similarity S between the target employee and each other employee age (i, j), and its calculation formula is: Where S age (i, j) represents the age similarity between the target employee and the i-th other employee; age i and age j represent the age of the target employee and the i-th other employee respectively; max(Δage) represents the maximum difference in the age range of the mine site; S32. Calculate the job similarity S between the target employee and each other employee job (i, j), and its calculation formula is: Where S job (i, j) represents the job similarity between the target employee and the i-th other employee; job i and job j represent the job parameters of the target employee and the i-th other employee respectively; max(Δjob) represents the maximum difference in job parameters; S33. Calculate the educational background similarity S between the target employee and each other employee edu (i, j), and its calculation formula is: Where S edu (i, j) represents the educational similarity between the target employee and the i-th other employee; edu i and edu j represent the education level parameters of the target employee and the i-th other employee respectively; max(Δedu) represents the maximum difference in education level parameters. S34. Calculate the employee similarity S between the target employee and each other employee ij , and its calculation formula is: S ij = w age × S age (i, j) + w job × S job (i, j) + w edu × S edu (i, j) Wherein, S ij represents the employee similarity between the target employee and the i-th other employee; w age , w job and w edu respectively represent the weight parameters for S age (i, j), S job (i, j) and S edu (i, j); S35. Calculate the personalized weight γ according to the employee similarity S ij The calculation formula is as follows: ij ​ γ ij = exp(-β × (1 - S ij )) where β represents the attenuation factor with respect to γ ij ; S36. According to the personalized weight γ ij Calculate the preference influence coefficient I of each other employee on the target employee ij ; S37. Calculate the second learning preference according to the preference influence coefficient I ij ​ 9. The learning preference analysis method according to claim 8, wherein In step S36, the preference influence coefficient I ij is calculated by the formula: where L j (k) and L i (k) respectively represent the number of times the i-th other employee and the target employee learn the k-th type of learning content; I ij represents the preference influence coefficient of the i-th other employee on the target employee.

10. The learning preference analysis method according to claim 8, characterized in that In step S37, the second learning preference is calculated by the formula: In the formula, represents the first learning preference of the target employee.

Images