Resume Screening Method, Computer Device, and Storage Medium

Abstract

The present application provides a resume screening method, a computer device, and a storage medium. Among them, the method includes: respectively inputting the job description information and the resume text corresponding to the job into a trained first neural network model for information extraction to obtain corresponding job key information and resume key information; based on a pre-configured knowledge base, post-processing the job key information and the resume key information to obtain structured job description information and structured resume information; obtaining additional features corresponding to the job description information and the resume text; inputting the structured job description information, the structured resume information, and the additional features into a trained second neural network model for matching degree analysis to obtain corresponding job matching degree scores, so as to screen resumes based on the job matching degree scores. The present application can improve the reliability and efficiency of resume screening.

Description

Technical Field

[0001] This application relates to the field of information technology, and in particular to a resume screening method, a computer device, and a storage medium. Background Art

[0002] Currently, when screening resumes, relevant personnel usually manually screen the resumes of each applicant based on aspects such as the applicant's education background, major, work experience, etc., and then interview the applicants corresponding to the screened resumes. Since the operation of resume screening relies on manual experience, some applicants who match the position may be screened out, or some applicants who do not match the position may be selected, which cannot ensure the reliability of resume screening; moreover, the operation of manual screening has a large workload and low efficiency.

[0003] Therefore, how to improve the reliability and efficiency of resume screening has become an urgent problem to be solved. Summary of the Invention

[0004] This application provides a resume screening method, a computer device, and a storage medium, which can improve the reliability and efficiency of resume screening.

[0005] In a first aspect, this application provides a resume screening method, which includes:

[0006] Input the job description information and the resume text corresponding to the job into a trained first neural network model for information extraction to obtain corresponding job key information and resume key information;

[0007] Based on a pre-configured knowledge base, post-process the job key information and the resume key information to obtain structured job description information and structured resume information;

[0008] Obtain additional features corresponding to the job description information and the resume text;

[0009] Input the structured job description information, the structured resume information, and the additional features into a trained second neural network model for matching degree analysis to obtain corresponding job matching degree scores, and perform resume screening based on the job matching degree scores.

[0010] In a second aspect, this application also provides a computer device, which includes:

[0011] A memory and a processor;

[0012] Among them, the memory is connected to the processor and is used to store programs;

[0013] The processor is used to implement the steps of the resume screening method according to any one of the embodiments provided in the present application by running the program stored in the memory.

[0014] In a third aspect, the present application further provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the processor is enabled to implement the steps of the resume screening method according to any one of the embodiments provided in the present application.

[0015] For the resume screening method, computer device, and storage medium disclosed in the present application, by respectively inputting the job description information and the resume text corresponding to the job into the trained first neural network model for information extraction, the corresponding job key information and resume key information are obtained. Then, based on the pre-configured knowledge base, post-processing is performed on the job key information and resume key information to obtain structured job description information and structured resume information, and additional features corresponding to the job description information and resume text are obtained. The structured job description information, structured resume information, and additional features are input into the trained second neural network model for matching degree analysis to obtain the corresponding job matching degree score, and resume screening is performed based on the obtained job matching degree score. The process of resume screening reduces the dependence on manual experience. Therefore, the reliability of resume screening is improved; and since the operation of manual screening is omitted, the efficiency of resume screening is also improved.

[0016] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and cannot limit the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0018] Figure 1 It is a schematic diagram of the steps of a resume screening method provided by an embodiment of the present application;

[0019] Figure 2 It is a schematic diagram of a resume screening system provided by an embodiment of the present application;

[0020] Figure 3 It is a schematic diagram of the steps of inputting the structured job description information, the structured resume information, and the additional features into the trained second neural network model for matching degree analysis to obtain the corresponding job matching degree score provided by an embodiment of the present application;

[0021] Figure 4 It is a schematic flowchart of a process for obtaining a job matching score based on a second neural network model provided by an embodiment of the present application;

[0022] Figure 5 It is a schematic block diagram of a computer device provided by an embodiment of the present application.

[0023] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and cannot limit the present application. Detailed implementation manners

[0024] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present application without creative efforts shall fall within the protection scope of the present application.

[0025] The flowchart shown in the accompanying drawings is only an example, and does not necessarily include all the contents and operations / steps, nor does it necessarily execute in the described order. For example, some operations / steps can also be decomposed, combined or partially merged. Therefore, the actual execution order may be changed according to the actual situation.

[0026] It should be understood that the terms used in the specification of the present application are only for the purpose of describing specific embodiments and are not intended to limit the present application. As used in the specification of the present application and the appended claims, unless otherwise clearly specified in the context, the singular forms of "a", "an" and "the" are intended to include the plural forms.

[0027] It should be understood that in order to facilitate the clear description of the technical solutions in the embodiments of the present application, in the embodiments of the present application, terms such as "first" and "second" are used to distinguish the same items or similar items with basically the same functions and effects. For example, the first callback function and the second callback function are only used to distinguish different callback functions, and do not limit their sequence. Those skilled in the art can understand that the terms "first" and "second" do not limit the quantity and execution order, and the terms "first" and "second" do not necessarily mean different.

[0028] It should also be understood that the term "and / or" used in the specification of the present application and the appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0029] Currently, when screening resumes, relevant personnel usually manually screen the resumes of each applicant based on aspects such as the applicant's education background, major, work experience, etc., and then interview the applicants corresponding to the screened resumes. Since the operation of resume screening relies on manual experience, some applicants who match the position may be eliminated, or some applicants who do not match the position may be selected, which cannot ensure the reliability of resume screening; moreover, the operation of manual screening has a large workload and low efficiency.

[0030] Therefore, embodiments of the present application provide a resume screening method, a computer device, and a storage medium to improve the reliability and efficiency of resume screening.

[0031] The following will describe some embodiments of the present application in detail with reference to the accompanying drawings. Without conflict, the following embodiments and the features in the embodiments can be combined with each other.

[0032] Please refer to Figure 1 , Figure 1 which is a schematic flowchart of a resume screening method provided by an embodiment of the present application. The resume screening method can be applied to a terminal device, where the terminal device can be a mobile phone, a tablet computer, a laptop computer, a desktop computer, a personal digital assistant, a wearable device, or a smart screen, a smart TV, a projector, a point-reading pen, a computer device, etc., and of course, it is not limited thereto. In the following, the resume screening method provided by the present application will be described by taking the resume screening method applied to a computer device as an example.

[0033] As Figure 1 shown, the resume screening method includes steps S101 to S104.

[0034] S101. Input the job description information and the resume text corresponding to the job into the trained first neural network model for information extraction to obtain the corresponding job key information and resume key information.

[0035] Before resume screening, a neural network model for information extraction is pre-constructed. For the sake of easy description, this neural network model will be referred to as the first neural network model below. Exemplarily, the first neural network model includes a BERT (Bidirectional Encoder Representation from Transformers) model, a CNN (Convolutional Neural Networks) model, etc. Moreover, multiple job description information and resumes corresponding to the jobs are collected and labeled to generate sample data, and the first neural network model is trained through the sample data to obtain the trained first neural network model.

[0036] When it is necessary to screen one or more resumes corresponding to a position, first, the trained first neural network model is used to extract information from the position description information corresponding to the position, and the trained first neural network model is used to extract information from the resume text. Among them, the position description information includes information such as position type, educational requirements for the position, professional requirements for the position, skill requirements for the position, and work experience requirements for the position. The resume text includes information such as the applicant's name, age, expected salary, self-introduction, educational experience (school, educational level, start and end times, major), work experience, project experience (project name, start and end times, position, specific description), etc.

[0037] After the first neural network model extracts information from the position description information, the corresponding position key information is obtained. Among them, the position key information includes, but is not limited to, educational requirements for the position, professional requirements for the position, skill requirements for the position, work experience requirements for the position, etc. And by using the first neural network model to extract information from the resume text, the corresponding resume key information is obtained. Among them, the resume key information includes, but is not limited to, expected salary, educational level, graduation school, major, work experience, project experience, etc.

[0038] In some embodiments, the operation of inputting the position description information and the resume text corresponding to the position into the trained first neural network model to extract information and obtaining the corresponding position key information and resume key information includes: inputting the position description information and the corresponding first tag group into the first neural network model, predicting the start position and end position of each tag in the first tag group corresponding to the position description information, and obtaining the position key information; inputting the resume text and the corresponding second tag group into the first neural network model, predicting the start position and end position of each tag in the second tag group corresponding to the resume text, and obtaining the resume key information.

[0039] In the operation of extracting information by the first neural network model, the data input into the model includes two parts. One part is the position description information / resume text, and the other part is the corresponding tag group information. The tag group information is used to indicate the field key values of the corresponding tags to be specifically extracted from the position description information / resume text. For the convenience of distinction and description, hereinafter, the tag group information corresponding to the position description information is called the first tag group, and the tag group information corresponding to the resume is called the second tag group. Among them, the first tag group includes various tags corresponding to the position description information, such as tags like "educational requirements for the position", "professional requirements for the position", "skill requirements for the position", "work experience requirements for the position", etc.; the second tag group includes various tags corresponding to the resume text, such as "educational level" tag, "school" tag, "major" tag, etc.

[0040] Exemplarily, the job description information is input into the BERT model for encoding. Meanwhile, the first tag group is input into the CNN model for encoding. The two encoding results are interacted to calculate the similarity, and the specific tag content corresponding to each tag in the first tag group for the job description information is predicted, including predicting the start position, end position, sequence, etc. of the field key values corresponding to each tag, so as to obtain the job key information corresponding to the job description information for the first tag group.

[0041] Similarly, the resume text is input into the BERT model for encoding. Meanwhile, the second tag group is input into the CNN model for encoding. The two encoding results are interacted to calculate the similarity, and the specific tag content corresponding to each tag in the second tag group for the resume text is predicted, so as to obtain the resume key information corresponding to the resume text for the second tag group.

[0042] It should be noted that not all the extracted information will be used. Instead, some key information that can be used for matching is selected as the job key information and the resume key information. For example, when extracting information from the resume text, information such as "marital status" may be obtained, but this information is rarely required in the job description information, so it will not be used as the resume key information.

[0043] Exemplarily, a method for selecting and constructing each tag in the first tag group / second tag group is preset and can be adjusted according to specific needs. Based on the first tag group and the second tag group, information extraction is performed on the job description information and the resume text respectively, so as to obtain the corresponding job key information and resume key information.

[0044] S102. Based on the pre-configured knowledge base, post-process the job key information and the resume key information to obtain structured job description information and structured resume information.

[0045] Exemplarily, the structured job description information and the structured resume information are composed of corresponding field names and field values. For example, taking the structured resume information as an example, the resume key information obtained by information extraction for a certain resume is: {"education": "master", "major": "computer technology", "school": "Peking University"}, then the corresponding structured resume information generated is: "[Basic Information][education]Master[major]Computer Technology[school]Peking University". Here, the two square brackets represent the coarse-grained and fine-grained field names respectively, which is convenient for the model to identify.

[0046] Exemplarily, a knowledge base is pre-configured, which is generated by professionals collecting and summarizing public information. The knowledge base contains standardized field knowledge of various information, including but not limited to the standard names and aliases of various majors and schools, the grade levels of different schools, the description information of each position level (such as experts, person in charge, etc.), job hunting status, political status, professional skills, industry names, exam names, etc. The knowledge base is mainly used for post-processing the obtained structured job description information and structured resume information. On the one hand, it further calculates the information required for subsequent matching. On the other hand, it restores some non-standardized field contents (such as aliases) in the key information of the job and the key information of the resume to the standard form for easy storage and subsequent matching.

[0047] Exemplarily, since various information may change, such as school name changes, etc., the knowledge base is updated regularly or irregularly, or new information content is added to the knowledge base, such as newly added majors, etc.

[0048] In some embodiments, the post-processing of the key information of the job and the key information of the resume based on the pre-configured knowledge base includes: based on the knowledge base, performing a matching query on the non-standardized content in the key information of the job and the key information of the resume, and performing a standardization process on the non-standardized content according to the queried matching data.

[0049] Since some field contents in the obtained key information of the job and key information of the resume may not be standardized, by calling the configured knowledge base, a matching query is performed on the non-standardized content in the obtained key information of the job and key information of the resume. For example, query the standardized field corresponding to the non-standardized school information in the key information of the resume in the knowledge base, and replace the non-standardized school information in the key information of the resume with the corresponding standardized field in the knowledge base to achieve the standardization process of the key information of the resume.

[0050] Exemplarily, the post-processing of the key information of the job and the key information of the resume may further include: processing null values and duplicate values in the key information of the job and the key information of the resume; packing the same set of elements together. For example, the start time, end time, company name, position name, etc. of a work experience are the same set of elements, and these information are packed together; calculating some values that cannot be directly extracted. For example, the total working duration of a work experience is a value that cannot be directly extracted, and the total working duration is calculated according to the start time and end time of the work experience.

[0051] By post-processing the key information of the job and the key information of the resume, the corresponding structured job description information and structured resume information are finally obtained.

[0052] S103. Obtain the additional features corresponding to the job description information and the resume text.

[0053] There are some features that have a greater impact on the person-job matching situation and cannot be directly obtained through the model. Therefore, in addition to obtaining structured job description information and structured resume information, additional features corresponding to the job description information and the resume text are also obtained. The additional features serve as supplementary information for the structured job description information and the structured resume information.

[0054] Exemplarily, the additional features include feature items and the corresponding feature values. It should be noted that the types of feature items included in the additional features can be set flexibly, and operations such as adding, deleting, and modifying the used feature items can also be performed according to actual needs.

[0055] For example, one type of feature item can be "the comparison situation between the applicant's education level and the education level required for the job", and the corresponding situations may be "lower than", "equal to", "higher than"; another type of feature item can be "the school level of the applicant's graduated school", which may be "world top", "domestic first-class", "key university", "ordinary university", etc.; another type of feature item can be "the applicant's internship experience", which may be "no internship experience", "has internship experience in an ordinary company", "has internship experience in a leading company", etc. By standardizing different situations of each feature item to corresponding feature values, such as taking values of 0, 1, 2... etc., the feature values corresponding to the feature items are obtained, thereby obtaining the additional features, and the additional features serve as a supplement to the structured job description information and the structured resume information.

[0056] In some embodiments, the obtaining of the additional features corresponding to the job description information and the resume text includes:

[0057] Receiving the additional features input manually, where the additional features are determined manually based on the job description information and the resume text; or

[0058] Performing analysis and processing on the job description information and the resume text to automatically generate the additional features.

[0059] In one implementation, the additional features can be constructed and input by personnel such as HR (Human Resources) and interviewers based on information such as job description information and resume text, and the additional features are directly obtained by receiving manual input.

[0060] For example, assume that there are a total of m types of corresponding feature items, and among them, n (n is less than m) types of feature items have a greater impact on the person-job matching situation. Then, the n types of feature items can be manually selected and the corresponding feature values of each type of feature item can be determined for input, and additional features can be obtained by receiving the manual input.

[0061] In another implementation, relevant feature items corresponding to the position are automatically selected, and by automatically processing and comparing information such as the job description information and the resume text, the feature values corresponding to each feature item are obtained, thereby automatically generating additional features.

[0062] In yet another implementation, after the feature items corresponding to the position are manually selected, by automatically processing and comparing information such as the job description information and the resume text, the feature values corresponding to each manually selected feature item are obtained to obtain additional features.

[0063] It should be noted that step S103 only needs to be executed before step S104, and does not necessarily need to be executed after step S102. There is no specific limitation on the execution order of step S103.

[0064] S104. Input the structured job description information, the structured resume information, and the additional features into the trained second neural network model for matching degree analysis, and obtain the corresponding job matching degree score, so as to screen the resume based on the job matching degree score.

[0065] A neural network model for scoring the job matching of resumes is pre-constructed. For the convenience of description, this neural network model will be referred to as the second neural network model below. Exemplarily, the second neural network model includes a BERT model. The second neural network model will perform model training for the scoring task. The training samples used for model training can be sourced from the human resources department. Each sample includes a resume, the job description information corresponding to the resume, and the final status of the resume. Among them, the final status of the resume includes but is not limited to being screened through, passing the initial test, passing the second test, and being finally admitted, etc. The second neural network model is trained with the samples to obtain the trained second neural network model. During the model training process, different scores can be assigned to the samples in different states according to the different specific uses of the second neural network model finally.

[0066] For example, if the second neural network model is mainly used to screen resumes, the scores of samples that pass the screening can be marked as a first score, such as 1, and the scores of samples that fail the screening can be marked as a second score, such as 0. The second neural network model is trained using these samples. For example, during the training process, the second neural network model uses binary cross entropy as the loss function until the second neural network model converges and the model training is completed. Afterwards, for a given input, the trained second neural network model outputs 0 or 1 points.

[0067] For another example, if you want the second neural network model to carefully consider the entire process, you can assign different scores to the samples according to the final status. For example, different scores are assigned to samples that pass screening, pass preliminary examination, pass re-examination, pass re-examination, and are finally admitted. For example, the score of the sample that passes the screening is marked as 0.1 points, the score of the sample that passes the preliminary examination is marked as 0.3 points, the score of the sample that passes the re-examination is marked as 0.5 points, and the score of the sample that is finally admitted is marked as 1 point. The second neural network model is trained with these samples until the second neural network model converges and the model training is completed. Afterwards, for a given input, the trained second neural network model outputs a score between 0 and 1.

[0068] like Figure 2 As shown, Figure 2 This is a schematic diagram of a system for resume screening for job matching. The system mainly includes a knowledge base, a first neural network model, and a second neural network model. The first neural network model and the knowledge base are used to parse job description information to obtain corresponding structured job description information. The first neural network model and the knowledge base are used to parse resume text to obtain corresponding structured resume information. The structured job description information and structured resume information are used as inputs to the second neural network model, and job matching is performed through the second neural network model to obtain a job matching score.

[0069] In some embodiments, as Figure 3 As shown, step S104 may include sub-step S1041 and sub-step S1042.

[0070] S1041. Concatenate the structured job description information and the structured resume information to obtain machine text;

[0071] S1042: Input the machine text and the additional features into the second neural network model, and output the job matching score.

[0072] The structured job description information and the structured resume information are used as the input of the second neural network model. The structured job description information and the structured resume information are concatenated, and the resulting machine text is used as the input of the second neural network model. The additional features are supplementary information of the machine text and are also used as the input of the second neural network model.

[0073] Exemplarily, the second neural network model includes a BERT model. In some embodiments, the step of inputting the machine text and the additional features into the second neural network model and outputting the job matching degree score includes: inputting the machine text into the BERT model to obtain a corresponding first feature vector; and processing the additional features through the embedding layer, activation function and linear layer of the second neural network model to obtain a corresponding second feature vector; processing the first feature vector and the second feature vector through the activation function and linear layer of the second neural network model to obtain a fusion vector; and processing the fusion vector through the dropout and linear layer of the second neural network model to obtain the job matching degree score.

[0074] For example, as Figure 4 shown, the machine text and the additional features are used as two parts of the input of the second neural network model. The machine text obtained by concatenating the structured job description information and the structured resume information is input into the BERT model to obtain a corresponding semantic vector, which is hereinafter referred to as the first feature vector for the convenience of distinction. Also, the additional features are input into the second neural network model. The additional features include feature 1, feature 2... feature N, etc. Through the embedding layer (Embedding) of the second neural network model, they are mapped to corresponding feature vectors, and the feature vectors 1, feature vectors 2... feature vectors N corresponding to each additional feature such as feature 1, feature 2... feature N are obtained. The feature vectors 1, feature vectors 2... feature vectors N are aggregated and processed through a linear layer transformation and an activation function (Linear + activation + Linear) to be combined into a feature vector, which is hereinafter referred to as the second feature vector for the convenience of distinction. Then, the first feature vector and the second feature vector are concatenated again and processed through a linear layer transformation and an activation function (Linear + activation + Linear) to obtain a fusion vector. Finally, the fusion vector is processed through Dropout and a linear layer (Dropout + Linear) to obtain the final job matching degree score.

[0075] Exemplarily, the activation function can be selected as the ReLU (Linear rectification function) function. It should be noted that other functions other than the ReLU function can also be selected as the activation function.

[0076] In some embodiments, the job matching score includes a preset first score and a second score. Screening the resume based on the job matching score includes: when the obtained job matching score is the first score, determining that the resume passes the screening; when the obtained job matching score is the second score, determining that the resume fails the screening.

[0077] Exemplarily, the first score is set to 1 point and the second score is set to 0 point. It should be noted that the specific values of the first score and the second score can be flexibly set according to the actual situation and are not specifically limited in this application.

[0078] If the second neural network model is trained with samples marked with the first score and the second score, the job matching score of the resume relative to the applied job output by the trained second neural network model is the first score or the second score. When the second neural network model outputs the first score, such as 1 point, it is determined that the resume passes the screening. When the second neural network model outputs the second score, such as 0 point, it is determined that the resume fails the screening. By scoring the job matching scores of each resume and screening the resumes based on the job matching scores, the manual screening process is omitted, greatly improving the efficiency.

[0079] In some embodiments, the job matching score is a non-fixed value. Screening the resume based on the job matching score includes: screening the resumes according to the job matching scores corresponding to each resume, and determining a preset number of resumes that pass the screening, where the job matching scores corresponding to the resumes that pass the screening are higher than the job matching scores corresponding to other resumes that fail the screening; or, when the job matching score corresponding to the resume is greater than or equal to the first preset threshold, determining that the resume passes the screening; when the job matching score corresponding to the resume is less than the first preset threshold, determining that the resume fails the screening.

[0080] If the second neural network model is trained with samples marked with corresponding different scores, the job matching score of the resume relative to the applied job output by the trained second neural network model is a non-fixed value, such as a non-fixed value within the range of 0 - 1.

[0081] Exemplarily, after obtaining the different job matching scores of each resume relative to the applied job, compare each job matching score, and determine the preset number of resumes with relatively higher job matching scores as the resumes that pass the screening, and the other resumes with relatively lower job matching scores as the resumes that fail the screening.

[0082] It should be noted that the preset number can be flexibly set according to the number of recruits required for the job and is not specifically limited here.

[0083] Exemplarily, a first preset threshold for resume screening is preset in advance. For example, the first preset threshold is set to a certain score within the range of 0-1. It should be noted that the first preset threshold can be flexibly set according to the actual situation, and no specific setting is made here.

[0084] After obtaining the job matching degree scores of each resume for the applied position, compare the job matching degree scores with the first preset threshold. If the job matching degree score corresponding to the resume is greater than or equal to the first preset threshold, it is determined that the resume screening is passed. On the contrary, if the job matching degree score corresponding to the resume is less than the first preset threshold, it is determined that the resume screening is not passed.

[0085] In some embodiments, after obtaining the corresponding job matching degree score, it further includes: if the job matching degree score exceeds the preset score range, perform data update processing on the job matching degree score, and the updated job matching degree score is within the score range; and / or, if the proportion of the job matching degree scores corresponding to each resume that is less than the second preset threshold reaches the preset ratio, perform equal-proportion amplification processing on the job matching degree scores corresponding to each resume.

[0086] Taking any resume as an example, after obtaining the job matching degree score of this resume for the applied position, determine whether the job matching degree score is within the preset score range. If the job matching degree score exceeds the preset score range, perform data update processing on the job matching degree score so that the updated job matching degree score is within the preset score range, and use the updated job matching degree score as the job matching degree score finally corresponding to the resume.

[0087] For example, the preset score range is preset to be 0-1. If the job matching degree score of the resume for the applied position is 1.1 points, that is, the job matching degree score exceeds the preset score range, then clip the job matching degree score of 1.1 points and update the job matching degree score of 1.1 points to 1 point. That is, the job matching degree score finally corresponding to the resume is 1 point.

[0088] In another implementation manner, a second preset threshold and a preset ratio are preset in advance. For example, the second preset threshold is set to a certain score within the range of 0-1, such as 0.5 points, and the preset ratio is set to 50%. It should be noted that the second preset threshold and the preset ratio can be flexibly set according to the actual situation, and no specific setting is made here.

[0089] After obtaining the job matching degree scores of each resume with respect to the applied position, if the proportion of the job matching degree scores corresponding to each resume that is less than the second preset threshold reaches the preset ratio, then perform an equal-proportion amplification process on the job matching degree scores corresponding to each resume, and use the processed job matching degree scores as the final job matching degree scores. In this way, it is achieved that the proportion of the final job matching degree scores corresponding to each resume that is less than the second preset threshold is lower than the preset ratio, making the final job matching degree scores corresponding to each resume more evenly distributed within the preset score range. Subsequently, when the user views the job matching degree scores corresponding to each resume, the resume will not be eliminated due to too low job matching degree scores, thereby further improving the reliability.

[0090] For example, assume that the job matching degree scores corresponding to 10 resumes are obtained. Among them, the job matching degree scores corresponding to 5 resumes are 0.4 points, and the job matching degree scores corresponding to 5 resumes are 0.2 points. Then, among the 10 job matching degree scores, the proportion of those less than 0.5 points is 100%, exceeding 50%. Then, the job matching degree scores of 0.4 points corresponding to 5 resumes and the job matching degree scores of 0.2 points corresponding to 5 resumes can be amplified by 2 times. Update 0.4 points to 0.8 points and 0.2 points to 0.4 points. That is, the final job matching degree scores corresponding to 5 resumes are 0.8 points, and the final job matching degree scores corresponding to 5 resumes are 0.4 points. In this way, the job matching degree scores corresponding to each resume are more evenly distributed within the 0-1 interval, thereby further improving the reliability of resume screening.

[0091] The resume screening method disclosed in the above embodiments extracts information by inputting the job description information and the resume text corresponding to the position into the trained first neural network model respectively, obtains the corresponding key job information and key resume information, and then based on the pre-configured knowledge base, post-processes the key job information and key resume information to obtain structured job description information and structured resume information, and obtains additional features corresponding to the job description information and the resume text. Input the structured job description information, structured resume information, and additional features into the trained second neural network model for matching degree analysis, obtain the corresponding job matching degree scores, and perform resume screening through the obtained job matching degree scores. The process of resume screening reduces the dependence on manual experience. Therefore, the reliability of resume screening is improved; and since the operation of manual screening is omitted, the efficiency of resume screening is also improved.

[0092] Please refer to Figure 5 , Figure 5 which is a schematic block diagram of a computer device provided by an embodiment of the present application. As Figure 5As shown in the figure, the computer device 300 includes one or more processors 301 and a memory 302. The processor 301 and the memory 302 are connected by a bus, such as an I2C (Inter-integrated Circuit) bus.

[0093] Among them, one or more processors 301 work alone or jointly to execute the steps of the resume screening method provided in the above embodiments.

[0094] Specifically, the processor 301 can be a microcontroller unit (MCU), a central processing unit (CPU), a digital signal processor (DSP), etc.

[0095] Specifically, the memory 302 can be a Flash chip, a read-only memory (ROM), a magnetic disk, an optical disc, a USB flash drive, a mobile hard disk, etc.

[0096] Among them, the processor 301 is used to run the computer program stored in the memory 302 and implement the steps of the resume screening method provided in the above embodiments when executing the computer program.

[0097] Exemplarily, the processor 301 is used to run the computer program stored in the memory 302 and, when executing the computer program, implement the following steps:

[0098] Input the job description information and the resume text corresponding to the job into the trained first neural network model for information extraction to obtain the corresponding job key information and resume key information;

[0099] Based on a pre-configured knowledge base, post-process the job key information and the resume key information to obtain structured job description information and structured resume information;

[0100] Obtain the additional features corresponding to the job description information and the resume text;

[0101] Input the structured job description information, the structured resume information, and the additional features into the trained second neural network model for matching degree analysis to obtain the corresponding job matching degree score, so as to screen resumes based on the job matching degree score.

[0102] In some embodiments, when the processor 301 implements the operation of inputting the structured job description information, the structured resume information, and the additional features into the trained second neural network model for matching degree analysis to obtain the corresponding job matching degree score, it is used to implement:

[0103] Concatenate the structured job description information and the structured resume information to obtain a machine text;

[0104] Input the machine text and the additional features into the second neural network model, and output the job matching degree score.

[0105] In some embodiments, when the processor 301 implements the operation of obtaining the additional features corresponding to the job description information and the resume text, it is used to implement:

[0106] Receive the additional features input manually, where the additional features are determined manually based on the job description information and the resume text; or

[0107] Analyze and process the job description information and the resume text to automatically generate the additional features.

[0108] In some embodiments, the second neural network model includes a BERT model. When the processor 301 implements the operation of inputting the machine text and the additional features into the second neural network model and outputting the job matching degree score, it is used to implement:

[0109] Input the machine text into the BERT model to obtain the corresponding first feature vector; and process the additional features through the embedding layer, activation function, and linear layer of the second neural network model to obtain the corresponding second feature vector;

[0110] Process the first feature vector and the second feature vector through the activation function and linear layer of the second neural network model to obtain a fusion vector;

[0111] Process the fusion vector through the dropout and linear layer of the second neural network model to obtain the job matching degree score.

[0112] In some embodiments, the job matching degree score includes a preset first score and a second score. When the processor 301 implements the operation of screening the resume based on the job matching degree score, it is used to implement:

[0113] When the obtained job matching degree score is the first score, determine that the resume passes the screening;

[0114] When the obtained job matching score is the second score, it is determined that the resume screening fails.

[0115] In some embodiments, when the job matching score implemented by the processor 301 is a non-fixed value and the resume is screened based on the job matching score, it is used to implement:

[0116] Resume screening is performed according to the job matching scores corresponding to each resume, and a preset number of resumes are determined to pass the screening, where the job matching scores corresponding to the resumes that pass the screening are higher than the job matching scores corresponding to other resumes that fail the screening; or

[0117] When the job matching score corresponding to the resume is greater than or equal to the first preset threshold, it is determined that the resume passes the screening; when the job matching score corresponding to the resume is less than the first preset threshold, it is determined that the resume fails the screening.

[0118] In some embodiments, after the processor 301 implements obtaining the corresponding job matching score, it is used to implement:

[0119] If the job matching score exceeds the preset score range, data update processing is performed on the job matching score, and the updated job matching score is within the score range; and / or

[0120] If the proportion of the job matching scores corresponding to each resume that is less than the second preset threshold reaches the preset ratio, equal-proportion amplification processing is performed on the job matching scores corresponding to each resume.

[0121] In some embodiments, when the processor 301 implements inputting the job description information and the resume text corresponding to the job into the trained first neural network model for information extraction to obtain the corresponding job key information and resume key information, it is used to implement:

[0122] Input the job description information and the corresponding first tag group into the first neural network model, predict the start position and end position of each tag in the first tag group corresponding to the job description information, and obtain the job key information;

[0123] Input the resume text and the corresponding second tag group into the first neural network model, predict the start position and end position of each tag in the second tag group corresponding to the resume text, and obtain the resume key information.

[0124] In some embodiments, when the processor 301 implements post-processing the job key information and the resume key information based on the pre-configured knowledge base, it is used to implement:

[0125] Based on the knowledge base, perform a matching query on the non-standardized content in the key information of the position and the key information of the resume, and perform standardization processing on the non-standardized content according to the retrieved matching data.

[0126] An embodiment of the present application further provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the processor is caused to implement the steps of the resume screening method provided in the above embodiment.

[0127] Among them, the computer-readable storage medium may be an internal storage unit of the computer device described in any of the foregoing embodiments, such as the hard disk or memory of the terminal device. The computer-readable storage medium may also be an external storage device of the terminal device, such as a plug-in hard disk equipped on the terminal device, a smart media card (SMC), a secure digital (SD) card, a flash card, etc.

[0128] As described above, the above are only specific embodiments of the present application, but 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 various equivalent modifications or substitutions, and these modifications or substitutions 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.

Claims

1. A resume screening method, characterized in that: The method comprises: Input the job description information and the resume text corresponding to the job into the trained first neural network model to extract information and obtain the corresponding job key information and resume key information; Based on a preconfigured knowledge base, post-processing the key information of the position and the key information of the resume is performed to obtain structured position description information and structured resume information; Obtaining additional features corresponding to the job description information and the resume text; The structured job description information, the structured resume information, and the additional features are input into a trained second neural network model for matching analysis to obtain corresponding job matching scores, so as to screen resumes based on the job matching scores.

2. The method according to claim 1, characterized in that Inputting the structured job description information, the structured resume information, and the additional features into a trained second neural network model for matching analysis to obtain a corresponding job matching score includes: splicing the structured job description information and the structured resume information to obtain machine text; The machine text and the additional features are input into the second neural network model, and the job matching score is output.

3. The method according to claim 1, characterized in that The obtaining of the additional features corresponding to the job description information and the resume text includes: receiving the additional features manually input, wherein the additional features are manually determined based on the job description information and the resume text; or The job description information and the resume text are analyzed and processed to automatically generate the additional features.

4. The method according to claim 2, characterized in that The second neural network model includes a BERT model, and inputting the machine text and the additional features into the second neural network model and outputting the job matching score includes: Inputting the machine text into the BERT model to obtain a corresponding first feature vector; and processing the additional features through an embedding layer, an activation function, and a linear layer of the second neural network model to obtain a corresponding second feature vector; Processing the first eigenvector and the second eigenvector through the activation function and the linear layer of the second neural network model to obtain a fusion vector; The fusion vector is processed by the dropout and linear layers of the second neural network model to obtain the job matching score.

5. The method according to claim 1, wherein The job matching score includes a preset first score and a second score, and screening the resumes based on the job matching score includes: When the obtained job matching score is the first score, determining that the resume has passed the screening; When the obtained job matching score is the second score, it is determined that the resume screening has failed.

6. The method according to claim 1, characterized in that The job matching score is a non-fixed value, and screening the resumes based on the job matching score includes: Screen resumes based on the job matching scores corresponding to each resume, and determine a preset number of resumes that have passed the screening, wherein the job matching scores corresponding to the resumes that have passed the screening are higher than the job matching scores corresponding to the resumes that have failed the screening; or When the job matching score corresponding to the resume is greater than or equal to the first preset threshold, it is determined that the resume screening has passed; when the job matching score corresponding to the resume is less than the first preset threshold, it is determined that the resume screening has failed.

7. The method according to claim 6, characterized in that After obtaining the corresponding job matching score, the method further includes: If the job matching score exceeds the preset score range, the job matching score is updated so that the updated job matching score is within the score range; and / or If the proportion of the job matching scores corresponding to the resumes that are less than the second preset threshold reaches a preset ratio, the job matching scores corresponding to the resumes are amplified in proportion.

8. The method according to claim 1, characterized in that The job description information and the resume text corresponding to the job are respectively input into the trained first neural network model for information extraction to obtain the corresponding job key information and resume key information, including: Inputting the job description information and the corresponding first tag group into the first neural network model, predicting the starting position and ending position of each tag in the first tag group corresponding to the job description information, and obtaining the key information of the job; The resume text and the corresponding second tag group are input into the first neural network model, and the starting position and ending position of each tag in the second tag group corresponding to the resume text are predicted to obtain the resume key information.

9. The method according to any one of claims 1 to 8, characterized in that The post-processing of the key position information and the key resume information based on the pre-configured knowledge base includes: Based on the knowledge base, a matching query is performed on the non-standardized content in the job key information and the resume key information, and the non-standardized content is standardized according to the queried matching data.

10. A computer device, characterized in that: The computer device comprises: memory and processor; Wherein, the memory is connected to the processor and is used to store programs; The processor is configured to implement the steps of the resume screening method according to any one of claims 1 to 9 by running the program stored in the memory.

11. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, which, when executed by a processor, enables the processor to implement the steps of the resume screening method according to any one of claims 1 to 9.

Images