User portrait generation method based on big data

Abstract

The invention relates to the technical field of portrait generation, and discloses a big data-based user portrait generation method, which comprises the steps of data collection, feature extraction, portrait generation and portrait updating. By widely collecting multi-dimensional data such as professional backgrounds, research results, work experiences and review processes, a user image is comprehensively drawn, and it is ensured that portraits are rich and accurate. In the feature extraction link, scientific quantification is carried out on various data, such as educational background, professional field popularity and various indexes of research results, and user features are accurately reflected; normalization processing eliminates dimension difference, and weighting calculation is combined with weight coefficients in all aspects, so that the comprehensive score is fair and reasonable. A regular updating mechanism can adjust portraits in time along with changes of user conditions, and timeliness is guaranteed; clear portrait labels are divided according to comprehensive scores, rapid screening and application in scenes such as talent management and project allocation are facilitated, the working efficiency and decision scientificity are improved, and powerful support is provided for related services.

Description

Technical Field

[0001] The present invention relates to the technical field of image generation, and particularly to a method for generating user portraits based on big data. Background Art

[0002] In the fields of scientific research and academia, researchers and scholars with different professional backgrounds, research achievements, and work experiences have different needs and performances in aspects such as academic exchanges, cooperation projects, and the allocation of scientific research resources. However, traditional evaluation methods for researchers often rely on simple resume information and limited personal introductions, and cannot comprehensively and accurately reflect their comprehensive strength. This evaluation method lacks systematicness and quantitative criteria, making it difficult to classify personnel in detail and match them precisely, resulting in problems such as unreasonable allocation of scientific research resources and low efficiency of academic exchange and cooperation. In professional service industries, such as consulting and review fields, service providers need to provide personalized service solutions for customers based on multi-dimensional information such as the customers' professional backgrounds, past project experiences, and review experiences. However, previous customer information collection and analysis methods are relatively single, unable to fully explore the potential value of customers and difficult to meet the increasingly diverse and personalized needs of customers. At the same time, due to the lack of an effective mechanism for dynamically updating customer portraits, as time goes by, the changes in customer information cannot be reflected in the portrait in a timely manner, causing the service to gradually deviate from the actual situation of the customer.

[0003] Currently, the application of methods for generating user portraits based on big data in specific fields such as scientific research, academia, and professional services still has many deficiencies, lacking a systematic solution for the characteristics and needs of users in these fields. Some existing user portrait generation methods often ignore the internal relationships between key data such as professional backgrounds, research achievements, work experiences, and review processes, and fail to fully utilize these data to construct a comprehensive and accurate portrait reflecting user characteristics. Moreover, in the process of feature extraction, the quantitative criteria are not scientific and reasonable enough, resulting in the generated portrait being unable to truly reflect the actual capabilities and values of users. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for generating user portraits based on big data, which solves the technical problems proposed in the background art.

[0005] The purpose of the present invention can be achieved by the following technical solutions:

[0006] A method for generating user portraits based on big data includes the following steps:

[0007] The first step, data collection:

[0008] Collect multi-dimensional data corresponding to the user's professional background data, research achievement data, work experience data, and review process data;

[0009] The second step, feature extraction:

[0010] Perform feature extraction processing based on the user's professional background data, research achievement data, work experience data, and review process data;

[0011] Step 3. Portrait generation:

[0012] First, perform normalization processing on the comprehensive feature values of the professional background, research achievements, work experience, and review process. Then, perform comprehensive scoring processing on the normalized comprehensive feature values through the weighted calculation method. After that, determine the corresponding portrait labels for the user according to the results of the comprehensive scoring processing;

[0013] Step 4. Portrait update:

[0014] According to the data collection method corresponding to the first step, regularly collect the new multi-dimensional data corresponding to the user's new professional background data, new research achievement data, new work experience data, and new review process data; then, according to the feature extraction method corresponding to the second step, recalculate the comprehensive feature values obtained by the user; after that, according to the portrait generation method corresponding to the third step, re-determine the portrait labels of the user.

[0015] As a further solution of the present invention: wherein:

[0016] Professional background data: including the user's academic degree and professional field;

[0017] Research achievement data: including the papers, patents, and project achievements published by the user;

[0018] Work experience data: including the user's job position and working years;

[0019] Review process data: including the review projects participated by the user, review opinions, and review results.

[0020] As a further solution of the present invention: the feature extraction processing method is as follows:

[0021] StepK1. Professional background feature extraction:

[0022] StepK1.1. Quantification of academic degree level:

[0023] Convert the user's academic degree information into numerical features and label it as A0;

[0024] When the user's academic degree information is undergraduate, quantify its academic degree level as 1, that is, A0 = 1;

[0025] When the user's academic degree information is master's, quantify its academic degree level as 2, that is, A0 = 2;

[0026] When the user's educational attainment information is a doctorate, the educational attainment level is quantified as 3, that is, A0 = 3;

[0027] StepK1.2. Popularity Score of Professional Field:

[0028] Score according to the popularity of the user's professional field, and the method is as follows:

[0029] First, count the number of papers, the number of projects, and the total research funds in the user's professional field, and mark them as B1, B2, and B3 respectively;

[0030] At the same time, in all professional fields, count the number of papers, the number of projects, or the total research funds in each professional field, and extract the largest number of papers, the number of projects, and the total research funds, and mark them as B1 max , B2 max and B3 max ;

[0031] Then, through: Calculate the popularity B0 of the user's professional field;

[0032] StepK1.3. Comprehensive Characteristics of Professional Background:

[0033] Multiply the educational attainment level by the popularity of the professional field to obtain the comprehensive characteristics of the professional background, that is, A0 × B0 = AB, where AB represents the comprehensive characteristic value of the professional background;

[0034] StepK2. Extraction of Research Result Characteristics:

[0035] First, count the number of papers published by the user, the number of citations of all the user's papers, the number of patents applied by the user, and the total research funds corresponding to all the projects the user participated in, and mark them as C1, C2, C3, and C4 respectively;

[0036] At the same time, among all users, count the number of papers published by each user, the number of citations of all the papers of each user, the number of patents applied by each user, and the total research funds corresponding to all the projects each user participated in, and extract the largest number of papers, the number of citations, the number of patents, and the total research funds, and mark them as C1 max , C2 max , C3 max and C4 max ;

[0037] Then, through: Calculate the comprehensive characteristic value C of the research results;

[0038] In the formula, α1, α2, α3, α4 are the corresponding preset weight coefficients;

[0039] Step K3. Feature extraction of work experience:

[0040] Step K3.1. Quantification of position level:

[0041] Convert the user's position information into a numerical feature and label it as D0;

[0042] If the user's position information is a junior position, quantify its position level as 1, i.e., D0 = 1;

[0043] If the user's position information is a mid-level position, quantify its position level as 2, i.e., D0 = 2;

[0044] If the user's position information is a senior position, quantify its position level as 3, i.e., D0 = 3;

[0045] Step K3.2. Work experience feature:

[0046] First, count the user's work experience and label it as E1;

[0047] At the same time, among all users, count the work experience of each user and extract the maximum work experience value, and label it as E1 max ;

[0048] Subsequently, through: Calculate the work experience feature value E0 of the user;

[0049] Step K3.3. Comprehensive work experience feature:

[0050] Multiply the quantified value of the position level and the work experience feature value to obtain the comprehensive work experience feature, i.e., D0 × E0 = DE, where DE represents the comprehensive work experience feature value;

[0051] Step K4. Feature extraction of the review process:

[0052] First, count the number of review projects participated by the user, the average number of words in all the user's corresponding review comments, and the number of projects passed by the user, and label them as F1, F2, and F3 respectively;

[0053] At the same time, among all users, count the number of review projects participated by each user and the average number of words in all the review comments corresponding to each user, and extract the maximum number of review projects participated and the average number of words in all the review comments, and label them as F1 max and F2 max ;

[0054] Subsequently, through: Calculate the comprehensive eigenvalue F of the review process;

[0055] Wherein, β1, β2, and β3 are the corresponding preset weight coefficients.

[0056] As a further solution of the present invention: The normalization process is as follows:

[0057] Among the comprehensive eigenvalues of the professional background, research achievements, work experience, and review process, extract the maximum and minimum comprehensive eigenvalues and mark them as H max and H min ;

[0058] Then, through: Calculate the normalized value H1 of each comprehensive eigenvalue;

[0059] Wherein, H is the substitution vector of the comprehensive eigenvalues of the professional background, research achievements, work experience, and review process.

[0060] As a further solution of the present invention: The comprehensive scoring process is as follows:

[0061] Mark the normalized values corresponding to the comprehensive eigenvalues of the professional background, research achievements, work experience, and review process as H11, H12, H13, and H14 respectively;

[0062] Then, through: M = H11×γ1 + H12×γ2 + H13×γ3, calculate the comprehensive score value M of the relevant user;

[0063] Wherein, γ1, γ2, γ3, and γ4 are the weight coefficients preset according to the professional background, research achievements, work experience, and review process respectively.

[0064] As a further solution of the present invention: The portrait label determination method is as follows:

[0065] Match the comprehensive score value M of the relevant user with multiple preset score division intervals:

[0066] Among them, the score division intervals include: [m y1 , m y2 , [m y2 , m y3 , and [m y3 , m y4 , and m y1 <m y2 <m y3 <m y4 ;

[0067] When M is in [m y1, m y2 , add the portrait label "Junior Expert" for the relevant user;

[0068] When M is in [m y2 , m y3 , add the portrait label "Intermediate Expert" for the relevant user;

[0069] When M is in [m y3 , m y4 , add the portrait label "Intermediate Expert" for the relevant user.

[0070] Advantages of the present invention:

[0071] In the present invention, by collecting multi-dimensional data such as the professional background data, research achievement data, work experience data, and review process data of users, it is possible to comprehensively and meticulously understand all aspects of users, making the generated user portraits richer, more accurate, and more capable of reflecting the true characteristics and ability levels of users.

[0072] In the present invention, during the feature extraction process, reasonable quantization and calculation methods are adopted for different types of data. In the extraction of professional background features, not only the academic qualifications are quantized by grade, but also the popularity of the professional field is comprehensively considered, making the professional background features more scientific and representative; in the extraction of research achievement features, multiple factors such as the number of papers, the number of citations, the number of patents, and the total amount of project research funds are comprehensively considered, and calculations are performed through preset weight coefficients, which can accurately reflect the achievements and abilities of users in research; in the extraction of work experience features, the position level and work years are quantized and comprehensively calculated, and in the extraction of review process features, factors such as the number of projects participated in the review, the number of words in the review opinions, and the number of projects passed in the review are comprehensively considered, all of which can effectively extract the key features related to the work and review of users, providing strong support for accurately generating user portraits.

[0073] In the present invention, normalization processing is performed on the comprehensive feature values of professional background, research achievements, work experience, and review process, eliminating the dimensional differences between different feature values, enabling each feature value to be compared on the same scale. Then, through the weighted calculation method, comprehensive scoring processing is performed on the normalized comprehensive feature values, and weight coefficients are preset according to the importance of different aspects, which can evaluate users more fairly and reasonably, making the comprehensive score value more accurately reflect the comprehensive ability and level of users.

[0074] According to the process of regularly collecting new data, recalculating eigenvalue, and re-determining portrait labels, the present invention can timely reflect the changes of users in aspects such as professional background, research achievements, work experience, and review process, so that the generated user portrait always conforms to the latest status of users, ensuring the timeliness and accuracy of the user portrait, and providing a more reliable basis for relevant decisions and applications.

[0075] The present invention matches the comprehensive score value of users with multiple preset score division intervals, and adds clear portrait labels to users. This clear division method is convenient for classifying and managing users and applications. In aspects such as talent selection, project allocation, and resource allocation, appropriate personnel can be quickly and accurately screened according to the portrait labels of users, improving work efficiency and the scientific nature of decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] The present invention will be further described below in conjunction with the accompanying drawings.

[0077] Figure 1 is a schematic flowchart of a method for generating a user portrait based on big data according to the present invention.

[0078] Figure 2 is a schematic flowchart of portrait generation in a method for generating a user portrait based on big data according to the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0079] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0080] Embodiment 1

[0081] Please refer to Figure 1 and Figure 2 As shown, the present invention is a method for generating a user portrait based on big data, including the following steps:

[0082] The first step, data collection:

[0083] Collect multi-dimensional data corresponding to the professional background data, research achievement data, work experience data, and review process data of users;

[0084] Among them:

[0085] Professional background data: including the user's education background and professional field;

[0086] Research achievement data: including papers, patents, and project achievements published by users;

[0087] Work experience data: including the user's job position and working years;

[0088] Review process data: including the review projects participated by the user, review opinions, and review results;

[0089] In this embodiment, the professional background data can be obtained from the educational institution database and the relevant academic association database; among them, the user's education level is divided into undergraduate, master, and doctor;

[0090] The research achievement data collects the user's research achievements through academic databases such as CNKI and Web of Science;

[0091] The work experience data is collected from the user's personal resume, enterprise employee database, and industry talent exchange platform; among them, the user's positions are divided into junior positions, intermediate positions, and senior positions;

[0092] The review process data is to record the relevant data during the review process when the user participates in the review activity;

[0093] For example:

[0094] Taking a scientific research personnel as an example, obtaining his professional background data from the educational institution database, it is known that the scientific research personnel has a doctorate degree and the professional field is artificial intelligence;

[0095] Collecting the research achievement data through CNKI, it is found that he has published 10 papers, the total number of citations is 50 times, applied for 3 patents, and the total research funds for the participated projects reached 500,000 yuan;

[0096] Collecting the work experience data from the personal resume, his position is an intermediate position and the working years are 5 years;

[0097] Recording the review process data when he participates in the review activity, he participates in 5 review projects, the average number of words in the review opinions is 500 words, and 3 projects have passed the review;

[0098] Second step, feature extraction:

[0099] Perform feature extraction processing according to the user's professional background data, research achievement data, work experience data, and review process data;

[0100] The feature extraction processing method is as follows:

[0101] Step K1, professional background feature extraction:

[0102] Step K1.1, quantification of education level:

[0103] Convert the user's education information into numerical features and mark them as A0;

[0104] When the user's educational attainment information is undergraduate, quantify its educational level as 1, i.e., A0 = 1;

[0105] When the user's educational attainment information is master's degree, quantify its educational level as 2, i.e., A0 = 2;

[0106] When the user's educational attainment information is doctorate, quantify its educational level as 3, i.e., A0 = 3;

[0107] StepK1.2, Popularity Score of Professional Field:

[0108] Score according to the popularity of the professional field to which the user belongs, and the method is as follows:

[0109] First, count the number of papers, the number of projects, and the total research funds within the professional field to which the user belongs, and label them as B1, B2, and B3 respectively;

[0110] At the same time, within all professional fields, count the number of papers, the number of projects, or the total research funds in each professional field, and extract the largest number of papers, the number of projects, and the total research funds, and label them as B1 max 、B2 max and B3 max ;

[0111] Then, through: Calculate the popularity B0 of the professional field to which the user belongs;

[0112] StepK1.3, Comprehensive Characteristics of Professional Background:

[0113] Multiply the educational level and the popularity of the professional field to obtain the comprehensive characteristics of the professional background, i.e., A0 × B0 = AB, where AB represents the comprehensive characteristic value of the professional background;

[0114] StepK2, Extraction of Research Result Characteristics:

[0115] First, count the number of papers published by the user, the number of citations of all the user's papers, the number of patents applied by the user, and the total research funds corresponding to all the projects participated by the user, and label them as C1, C2, C3, and C4 respectively;

[0116] At the same time, among all users, count the number of papers published by each user, the number of citations of all the papers of each user, the number of patents applied by each user, and the total research funds corresponding to all the projects participated by each user, and extract the largest number of papers, the number of citations, the number of patents, and the total research funds, and label them as C1 max 、C2 max 、C3 maxand C4 max ;

[0117] Subsequently, through: Calculate the comprehensive eigenvalue C of the research results;

[0118] In the formula, α1, α2, α3, α4 are the corresponding preset weight coefficients;

[0119] StepK3. Extraction of work experience characteristics:

[0120] StepK3.1. Quantification of job level:

[0121] Convert the user's job information into numerical features and label them as D0;

[0122] When the user's job information is a junior position, quantify its job level as 1, i.e., D0 = 1;

[0123] When the user's job information is a mid-level position, quantify its job level as 2, i.e., D0 = 2;

[0124] When the user's job information is a senior position, quantify its job level as 3, i.e., D0 = 3;

[0125] StepK3.2. Work experience characteristics:

[0126] First, count the user's work experience and label it as E1;

[0127] At the same time, among all users, count the work experience of each user, extract the maximum work experience value, and label it as E1 max ;

[0128] Subsequently, through: Calculate the work experience eigenvalue E0 of the user; StepK3.3. Comprehensive work experience characteristics:

[0129] Multiply the quantified job level value by the work experience eigenvalue to obtain the comprehensive work experience characteristic, i.e., D0 × E0 = DE, where DE represents the comprehensive work experience eigenvalue;

[0130] StepK4. Extraction of review process characteristics:

[0131] First, count the number of projects the user participates in the review, the average number of words in all the user's review opinions, and the number of projects the user approves, and label them as F1, F2, F3 respectively;

[0132] Meanwhile, among all users, count the number of review projects participated by each user, the average number of words in all review comments corresponding to each user, extract the largest number of review projects participated and the average number of words in all review comments, and mark them as F1 max and F2 max ;

[0133] Subsequently, through: Calculate the comprehensive eigenvalue F of the review process;

[0134] In the formula, β1, β2, and β3 are the corresponding preset weight coefficients;

[0135] Illustrative example:

[0136] In terms of extracting professional background features, the academic level of this researcher is quantified as 3;

[0137] Suppose the number of papers B1 of this researcher in the field of artificial intelligence is 1000, the number of projects B2 is 50, and the total research funds B3 is 10 million yuan. The maximum values of these three items in all professional fields are B1max = 5000, B2max = 200, and B3max = 50 million yuan. Calculate the popularity B0 of his professional field through the formula, and then obtain the comprehensive eigenvalue AB of the professional background;

[0138] When extracting research result features, calculate the comprehensive eigenvalue C of research results according to the paper, citation, patent and project fund data of this researcher, combined with the preset weight coefficients;

[0139] In the extraction of work experience features, the position level is quantified as 2. Suppose the maximum working years E1max of all users is 20 years, calculate the working years eigenvalue E0 of this researcher, and then obtain the comprehensive eigenvalue DE of work experience;

[0140] When extracting review process features, calculate the comprehensive eigenvalue F of the review process according to the review data of this researcher, combined with the preset weight coefficients.

[0141] Step 3. Portrait generation:

[0142] First, perform normalization processing on the comprehensive eigenvalues of professional background, research results, work experience, and review process. Subsequently, perform comprehensive scoring on the normalized comprehensive eigenvalues through the weighted calculation method. Then, determine the corresponding portrait labels for users according to the results of the comprehensive scoring;

[0143] The normalization processing method is as follows:

[0144] Among the comprehensive eigenvalues of professional background, comprehensive eigenvalues of research achievements, comprehensive eigenvalues of work experience, and comprehensive eigenvalues of the review process, extract the comprehensive eigenvalue with the largest value and the comprehensive eigenvalue with the smallest value, and mark them as H max and H min ;

[0145] Then, through: Calculate the normalized value H1 of each comprehensive eigenvalue;

[0146] In the formula, H is the substitution vector of the comprehensive eigenvalue of professional background, the comprehensive eigenvalue of research achievements, the comprehensive eigenvalue of work experience, and the comprehensive eigenvalue of the review process;

[0147] The comprehensive scoring processing method is as follows:

[0148] Mark the normalized values corresponding to the comprehensive eigenvalue of professional background, the comprehensive eigenvalue of research achievements, the comprehensive eigenvalue of work experience, and the comprehensive eigenvalue of the review process as H11, H12, H13, and H14 respectively;

[0149] Then, through: M = H11×γ1 + H12×γ2 + H13×γ3, calculate the comprehensive scoring value M of the relevant user;

[0150] In the formula, γ1, γ2, γ3, and γ4 are the preset weight coefficients corresponding to professional background, research achievements, work experience, and review process respectively;

[0151] The method for determining portrait labels is as follows:

[0152] Match the comprehensive scoring value M of the relevant user with multiple preset scoring division intervals:

[0153] Among them, the scoring division intervals include: [m y1 , m y2 , [m y2 , m y3 , and [m y3 , m y4 , and m y1 <m y2 <m y3 <m y4 ;

[0154] When M is in [m y1 , m y2 , add the portrait label "junior expert" to the relevant user;

[0155] When M is in [m y2 , m y3 , add the portrait label "intermediate expert" to the relevant user;

[0156] When M is in [my3 , m y4 , add the portrait label "Intermediate Expert" to the relevant user;

[0157] Illustrative example:

[0158] Suppose among a group of scientific researchers, the maximum comprehensive eigenvalue of the professional background is H max is 9, and the minimum H min is 1.5. Calculate the normalized value H1 of each comprehensive eigenvalue of this scientific researcher through the normalization formula;

[0159] Then, based on the preset weight coefficients γ1, γ2, γ3, γ4, calculate the comprehensive score value M of this scientific researcher;

[0160] If the score division interval is [m y1 , m y2 is [0, 0.3], [m y2 , m y3 is (0.3, 0.6], [m y3 , m y4 is (0.6, 1], and assume the M value is 0.4, then this scientific researcher is added the portrait label "Intermediate Expert";

[0161] In this embodiment, the multi-dimensional data collection is comprehensive and specific, covering multi-faceted data such as professional background, research achievements, work experience, and review process, and can depict user characteristics from multiple perspectives; the feature extraction method is scientific and reasonable, quantifies and scores different types of data, and converts complex information into computable eigenvalue, facilitating subsequent analysis; through normalization processing and weighted calculation method for comprehensive scoring, it can comprehensively consider various factors, determine accurate portrait labels for users, and provide an effective method and basis for the generation of user portraits.

[0162] Embodiment 2

[0163] Please refer to Figure 1 and Figure 2 . As the second embodiment of the present invention, when this application is specifically implemented, compared with Embodiment 1, the technical solution of this embodiment is only different from that of Embodiment 1 in that this embodiment further includes the step: portrait update;

[0164] The portrait update is as follows:

[0165] First, according to the data collection method corresponding to the first step, regularly collect the new multi-dimensional data corresponding to the user's new professional background data, new research achievement data, new work experience data, and new review process data;

[0166] Then, in combination with the new multi-dimensional data, and according to the feature extraction method corresponding to the second step, recalculate the comprehensive eigenvalue obtained by the user;

[0167] Next, combine the re-derived comprehensive eigenvalue for each item, and re-determine the user's portrait tags according to the corresponding portrait generation method in the third step;

[0168] For example: Still taking the scientific research personnel in Embodiment 1 as an example, after a period of time, this scientific research personnel was promoted to a senior position, published 5 more papers, the number of citations increased to 80 times, participated in 4 new review projects, and the average number of words in the review opinions increased to 600 words;

[0169] Collect these new data according to the data collection method, and recalculate the comprehensive eigenvalue for each item. For example, the comprehensive eigenvalue of the professional background, the comprehensive eigenvalue of research results, etc. may change;

[0170] After recalculation, if the comprehensive score value M of this scientific research personnel becomes 7, according to the portrait generation method, his portrait tag will be updated from "intermediate expert" to "senior expert" to reflect his latest situation;

[0171] This embodiment adds a portrait update step, which can update the user portrait in a timely manner according to the newly generated data of the user, ensure the timeliness and accuracy of the portrait, and enable the portrait to dynamically reflect the latest situation of the user.

[0172] Embodiment 3

[0173] Please refer to Figure 1 and Figure 2 As shown in, as Embodiment 3 of the present invention, when this application is specifically implemented, compared with Embodiment 1 and Embodiment 2, the technical solution of this embodiment is to combine and implement the solutions of the above-mentioned Embodiment 1 and Embodiment 2. The difference between the technical solution of this embodiment and Embodiment 1 and Embodiment 2 is only that in this embodiment, it further includes the step: portrait application;

[0174] Portrait application is to provide personalized services for relevant users according to the user's portrait tags, and the methods are as follows:

[0175] StepU1, personalized recommendation:

[0176] Provide personalized recommendation services for users according to the user portrait;

[0177] For example, according to the user's professional field and research direction, match relevant academic resources; recommend high-level academic conferences or projects for "senior expert" users, and recommend basic training courses for "junior expert" users;

[0178] StepU2, talent assessment:

[0179] Conduct talent assessment according to the user portrait;

[0180] For example, during the recruitment process, evaluate the professional background, research achievements, and work experience of candidates based on their portrait tags;

[0181] If a candidate's portrait tag is "senior expert" and the eigenvalue of their research achievements is 0.8, it indicates that they have strong capabilities and influence in the scientific research field;

[0182] Suppose a company gives priority to candidates with the portrait tag of "senior expert" and a relatively high eigenvalue of work experience when recruiting a technical director;

[0183] StepU3. Resource Allocation:

[0184] Conduct resource allocation according to the user portrait;

[0185] For example, in the evaluation of scientific research projects, assign high-difficulty evaluation tasks based on the portrait tags of the evaluation experts; and when a certain journal assigns paper review tasks, it gives priority to users with the portrait tag of "senior expert" and a relatively high eigenvalue of professional review opinions;

[0186] This embodiment adds portrait application steps, applies the generated user portraits to practical scenarios such as personalized recommendation, talent evaluation, and resource allocation, gives full play to the value of user portraits, provides strong support for relevant decisions, and improves the pertinence and efficiency of services.

[0187] Embodiment 4

[0188] Please refer to Figure 1 and Figure 2 As shown, as Embodiment 4 of the present invention, when this application is specifically implemented, compared with Embodiment 1, Embodiment 2, and Embodiment 3, the technical solution of this embodiment lies in combining the solutions of the above Embodiment 1, Embodiment 2, and Embodiment 3 for implementation.

[0189] This embodiment combines the solutions of Embodiment 1, 2, and 3, which not only ensures the scientificity and accuracy of user portrait generation, but also can update the portrait in a timely manner to adapt to changes. It can also apply the portrait to practical scenarios, achieving full-process coverage from data collection, portrait generation, update to application, and has comprehensiveness and systematicness.

[0190] It should be stated that all multi-dimensional data of users collected in this application are collected with the consent and authorization of the users, and the uses of the multi-dimensional user data are all legal and compliant, and the use and processing of user data comply with the relevant laws, regulations, and standards of the relevant regions.

[0191] The above formulas are all dimensionless and take their numerical values for calculation. The formulas are obtained by collecting a large amount of data for software simulation to obtain a formula closest to the actual situation. The preset parameters and threshold selections in the formulas are set by those skilled in the art according to the actual situation.

[0192] As described above, it is only the specific implementation manner of the present application. However, the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of changes or substitutions, which should all be covered within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims described above.

Claims

1. A method for generating user portraits based on big data, characterized in that, It includes the following steps: Data collection: Collect multi-dimensional data corresponding to the user's professional background data, research achievement data, work experience data, and review process data; Feature extraction: Perform feature extraction processing based on the user's multi-dimensional data, and obtain the comprehensive feature values of the professional background, research achievements, work experience, and review process; Portrait generation: First, perform normalization processing on each comprehensive feature value, then perform comprehensive scoring processing on the normalized comprehensive feature values through the weighted calculation method, and then determine the corresponding portrait labels for the user according to the results of the comprehensive scoring processing; Portrait update: According to the data collection method, regularly collect new multi-dimensional data corresponding to the user's new professional background data, new research achievement data, new work experience data, and new review process data; then, according to the feature extraction method, recalculate the comprehensive feature values obtained by the user; afterwards, according to the portrait generation method, re-determine the user's portrait labels.

2. The method for generating a user profile based on big data according to claim 1, wherein The extraction method of the comprehensive feature value of the professional background is as follows: StepK1.1, Quantification of educational level: Convert the user's educational information into numerical features: When the user's educational information is undergraduate, quantify its educational level as 1; when the user's educational information is master's, quantify its educational level as 2; when the user's educational information is doctorate, quantify its educational level as 3; StepK1.2, Scoring of the popularity of the professional field: Score according to the popularity of the professional field to which the user belongs, and the method is as follows: First, count the number of papers, the number of projects, and the total amount of scientific research funds in the professional field to which the user belongs, and mark them as B1, B2, and B3 respectively; Meanwhile, in all professional fields, count the number of papers, the number of projects, or the total amount of scientific research funds in each professional field, extract the largest number of papers, the number of projects, and the total amount of scientific research funds, and label them as B1 max , B2 max and B3 max ; Subsequently, by: calculating the popularity B0 of the user's professional field; StepK1.3, Comprehensive feature of the professional background: Multiply the educational level and the popularity of the professional field to obtain the comprehensive feature of the professional background, and mark it as AB.

3. The method for generating a user profile based on big data according to claim 2, wherein The extraction method of the comprehensive feature value of research achievements is as follows: First, count the number of papers published by the user, the number of citations of all the user's papers, the number of patents applied by the user, and the total amount of scientific research funds corresponding to all the projects participated by the user, and mark them as C1, C2, C3, and C4 respectively; Meanwhile, among all users, count the number of papers published by each user, the number of citations of all papers of each user, the number of patents applied by each user, and the total research funds corresponding to the number of all projects participated by each user, and extract the largest number of papers, number of citations, number of patents, and total research funds, and label them as C1 max , C2 max , C3 max and C4 max ; Subsequently, by: calculating the comprehensive eigenvalue C of the research results; In the formula, α1, α2, α3, and α4 are the corresponding preset weight coefficients.

4. The method for generating a user profile based on big data according to claim 3, wherein, The extraction method of the comprehensive feature value of work experience is as follows: StepK3.1, Quantification of position level: Convert the user's position information into numerical features: When the user's position information is a junior position, quantify its position level as 1; when the user's position information is a mid-level position, quantify its position level as 2; when the user's position information is a senior position, quantify its position level as 3; StepK3.2, Work experience feature: First, count the user's work years; at the same time, among all users, count the work years of each user, and extract the maximum work years from them, and then calculate the proportion of the user's work years in the maximum work years to obtain the work experience feature value of the user; StepK3.3, Comprehensive feature of work experience: Multiply the quantified value of the position level and the work experience feature value to obtain the comprehensive feature of work experience, and mark it as DE.

5. The method for generating a user profile based on big data according to claim 4, wherein, The extraction method of the comprehensive eigenvalue in the review process is as follows: First, count the number of review projects participated by the user, the average number of words in all the review opinions corresponding to the user, and the number of projects passed by the user, and mark them as F1, F2, and F3 respectively; Meanwhile, among all users, count the number of review projects each user participates in, and the average number of words in all the review comments corresponding to each user. Then extract the largest number of review projects participated in and the average number of words in all review comments, and mark them as F1 max and F2 max ; Subsequently, by: Calculating the comprehensive eigenvalue F of the review process; In the formula, β1, β2, and β3 are the corresponding preset weight coefficients.

6. The method for generating a user profile based on big data according to claim 5, wherein The professional background data includes the user's education background and professional field; the research result data includes the papers, patents, and project results published by the user; the work experience data includes the user's job position and working years; the review process data includes the review projects, review opinions, and review results participated by the user.

7. The method for generating a user portrait based on big data according to claim 5, wherein The normalization processing method is as follows: Among the comprehensive eigenvalue of professional background, the comprehensive eigenvalue of research achievements, the comprehensive eigenvalue of work experience, and the comprehensive eigenvalue of the review process, extract the comprehensive eigenvalue with the largest value and the comprehensive eigenvalue with the smallest value, and mark them as H max and H min ; Followed by: Calculating the normalized value H1 of each comprehensive feature value; In the formula, H is the substitution vector of the comprehensive eigenvalue of the professional background, the comprehensive eigenvalue of the research results, the comprehensive eigenvalue of the work experience, and the comprehensive eigenvalue of the review process.

8. The method for generating a user profile based on big data according to claim 7, wherein, The comprehensive score processing method is as follows: Mark the normalized values corresponding to the comprehensive eigenvalue of the professional background, the comprehensive eigenvalue of the research results, the comprehensive eigenvalue of the work experience, and the comprehensive eigenvalue of the review process as H11, H12, H13, and H14 respectively; Then, through: M = H11×γ1 + H12×γ2 + H13×γ3, calculate the comprehensive score value M of the relevant user; In the formula, γ1, γ2, γ3, and γ4 are the corresponding preset weight coefficients based on the professional background, research results, work experience, and review process respectively.

9. The method for generating a user profile based on big data according to claim 8, wherein, The method for determining the portrait label is as follows: Match the comprehensive score value M of the relevant user with multiple preset score division intervals: The score division intervals include: [m y1 , m y2 , [m y2 , m y3 , and [m y3 , m y4 ​ When M is in [m y1 , m y2 , then add the portrait label "Junior Expert" to the relevant user; When M is in [m y2 , m y3 , then add the portrait label "Intermediate Expert" to the relevant user; When M is in the range of [m y3 , m y4 , add the portrait label "Intermediate Expert" to the relevant users.

10. The method for generating a user profile based on big data according to claim 9, wherein Among them, m y1 <m y2 <m y3 <m y4 。

Images