Resume evaluation method and device, electronic equipment and storage medium

Abstract

The invention relates to a resume evaluation method and device, electronic equipment and a storage medium, and the method comprises the steps: obtaining the description information of target information based on the target information provided by a first user; extracting a resume fragment related to the description information of the target information from the resume information of the second user; and inputting description information of the target information and the resume fragment to a large language model, and generating multi-dimensional resume evaluation information through processing of the large language model, wherein the large language model is finely adjusted and trained to understand description information and resume information of the target information and generate resume evaluation information expressing matching between the resume information and the target information. According to the embodiment of the invention, the resume can be accurately and objectively evaluated, the processing efficiency is high, and time and energy are saved for users.

Description

Technical Field

[0001] The present invention relates to the technical field of Internet data processing, and in particular to a resume evaluation method, device, electronic device and storage medium. Background Art

[0002] In people's life and work, people need to evaluate certain information. For example, in the field of recruitment, the human resources (HR) department faces a lot of resume evaluation and screening work during the recruitment process. The traditional method mainly relies on manual reading and analysis of resumes to evaluate whether the job seeker of the resume is suitable for the position to be recruited. This method is not only inefficient, but also easily affected by subjective factors, resulting in the omission of excellent candidates. In order to improve the efficiency and quality of screening, some recruitment platforms provide some resume screening tools based on matching algorithms (such as text matching of keywords, feature matching, etc.). The text matching method mainly screens resumes by preset keywords or rules. For example, a Chinese patent with announcement number CN 109165295 B discloses a resume screening method, extracting data from a recruitment data set, extracting data from resume text data, and then matching the data extracted from the resume text data with the data extracted from the recruitment data set, and writing the matched data into a database. The feature matching method mainly extracts the recruiter data and the job seeker data respectively, and converts them into feature vectors or matrices respectively, and matches them by comparing the two feature vectors or feature matrices. However, these two matching algorithms can only serve as preliminary selection, and the fixed screening conditions may lead to the omission of some suitable candidates. Recruiters are still required to review resumes one by one, so they are of limited help to recruiters.

[0003] With the development of intelligent technologies (such as deep learning, neural networks, etc.), machine learning models have also been applied to resume screening methods to help recruiters screen resumes, and more types of data can be used in the evaluation process. For example, a Chinese patent with a publication number of CN 108182512 B discloses a resume screening method, which uses an evaluation model to calculate the probability that the resume to be evaluated is associated with a target feedback action item. In addition to using the resume data itself, it also uses the feedback data of the recruiter when browsing the resume. There are also methods that use other types of data to assist in screening resumes, which will not be described one by one here.

[0004] Although intelligent technology provides more methods for resume screening, there are still some problems. For example, due to the dimensions and conditions of data extraction, there are limitations in the depth of analyzing resume content, making it difficult to accurately evaluate the actual ability of candidates and their match with the position; there is a lack of interaction with users, and it is impossible to receive user feedback and make real-time adjustments.

[0005] In some other scenarios where evaluation information is needed, it is usually done manually. For example, course providers usually only evaluate students' learning information through test scores, which is obviously too one-sided. In addition, for career planners, when providing career planning for career consultants, they first need to evaluate the suitability of the career consultants for one or more occupations. The current evaluation method mainly relies on the subjective judgment of career planners, which is time-consuming and labor-intensive, and is seriously affected by subjective factors. Summary of the invention

[0006] In view of the technical problems existing in the prior art, the present invention proposes a resume evaluation method, device, electronic device and storage medium, which accurately evaluate the matching between the resume and the target information based on the specific information of the resume in a specific scenario.

[0007] According to one aspect of the present invention, the present invention provides a resume evaluation method, comprising the following steps:

[0008] Acquire description information of the target information based on the target information provided by the first user;

[0009] Extracting a resume fragment related to the description information of the target information from the resume information of the second user; and

[0010] Inputting the description information of the target information and the resume fragment into a large language model, and generating multi-dimensional resume evaluation information through processing by the large language model;

[0011] The large language model is fine-tuned and trained to understand the descriptive information and resume information of the target information and generate resume evaluation information that expresses the matching between the resume information and the target information.

[0012] Optionally, the step of extracting a resume segment related to the description information of the target information from the resume information of the second user includes:

[0013] Semantically grouping the resume information of the second user, and taking each obtained semantic group as a resume segment;

[0014] Generate a vector for each resume fragment, and multiple resume fragments and their vectors constitute a resume fragment index;

[0015] Semantically grouping the description information of the target information, and taking each obtained semantic group as a target description sub-information;

[0016] Generate a vector for each target descriptor information;

[0017] Calculate the cosine similarity between each resume segment vector and each target descriptor information vector; and

[0018] Corresponding to each target descriptor information, the resume segment with the maximum cosine similarity with its vector or the cosine similarity greater than a threshold is extracted from the resume information of the second user as the resume segment related to the corresponding target descriptor information.

[0019] Optionally, after extracting the resume fragment related to the description information of the target information from the resume information of the second user, the method further includes:

[0020] Based on the description information of the target information, a historical recall database is retrieved to determine the resume description information with the highest satisfaction with the description information of the target information as the resume reference information; correspondingly, the description information of the target information, the resume fragment and the resume reference information are input into the large language model.

[0021] Optionally, the dimensions of the resume evaluation information include: one or more of the second user's strengths, the degree of matching with the target information, the second user's weaknesses and resume description information.

[0022] Optionally, each dimension information in the resume evaluation information is provided to the first user in the form of descriptive text, rating or structured label.

[0023] Optionally, the resume evaluation method further comprises:

[0024] An interactive interface is provided, through which input information of a first user is received and generated resume evaluation information is output.

[0025] Optionally, the resume evaluation method outputs the generated resume evaluation information in the form of text, audio or video.

[0026] Optionally, the resume evaluation method further comprises: analyzing input information, and when the input information includes feedback information on resume evaluation information, further comprises:

[0027] extracting satisfaction information from the feedback information; and

[0028] The resume evaluation information is annotated based on the satisfaction information and stored in a historical recall database.

[0029] Optionally, when the input information includes user preference information, the method further includes:

[0030] Setting a weight for the preference information;

[0031] Correspondingly, the preference information and its corresponding weight, the description information of the target information and the resume fragment are input into the large language model, and multi-dimensional resume evaluation information is generated after being processed by the large language model.

[0032] Optionally, when the input information includes user evaluation index information, the method further includes:

[0033] Setting weights for the evaluation indicator information;

[0034] Correspondingly, the evaluation index information and its corresponding weight, the description information of the target information and the resume fragment are input into a large language model, and multi-dimensional resume evaluation information is generated after being processed by the large language model; or

[0035] Extracting a second resume segment related to the evaluation indicator information from the resume information, wherein the resume segment related to the description information of the target information extracted from the resume information is the first resume segment;

[0036] Correspondingly, the description information of the target information, the evaluation index information, the first resume fragment and the second resume fragment are input into the large language model, and multi-dimensional resume evaluation information is generated after being processed by the large language model.

[0037] Optionally, the first user is a recruiting user, the corresponding target information is position information, the corresponding second user is a job seeker, and the corresponding resume is a job application resume; or, the first user is a career planner, the corresponding target information is career information, the corresponding second user is a career consultant, and the corresponding resume is the resume of the career consultant; or the first user is a course trainer, the corresponding target information is course information, the corresponding second user is a learner, and the corresponding resume is the learner's academic performance information.

[0038] According to another aspect of the present invention, the present invention also provides a resume evaluation device, including a data acquisition module, a data orientation module and an evaluation module, wherein the data acquisition module is configured to acquire descriptive information of the target information based on the target information provided by the first user; the data orientation module is configured to extract a resume fragment related to the descriptive information of the target information from the resume information of the second user; the evaluation module is configured to input the descriptive information of the target information and the resume fragment into a large language model, and generate multi-dimensional resume evaluation information after processing by the large language model; wherein the large language model is fine-tuned and trained to understand the descriptive information and resume information of the target information and generate resume evaluation information expressing the matching between the resume information and the target information.

[0039] According to another aspect of the present invention, the present invention also provides an electronic device, which includes a processor and a memory, wherein a computer program instruction set is stored in the memory, and the aforementioned resume evaluation method is implemented when the processor executes the computer program instruction set in the memory.

[0040] According to another aspect of the present invention, the present invention also provides a computer-readable storage medium, on which a computer program instruction set is stored, and when the computer program instruction set is executed by a processor, the resume evaluation method described above is implemented.

[0041] According to another aspect of the present invention, the present invention further provides a computer program product, comprising a computer program instruction set, wherein the computer program instruction set implements the aforementioned resume evaluation method when executed by a processor.

[0042] The present invention realizes efficient retrieval and accurate evaluation of resume content, improves processing efficiency and evaluation accuracy, and saves users a lot of time and energy. Since the interference of human subjective factors is reduced when evaluating resumes, the objectivity of the evaluation is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The preferred embodiments of the present invention will be further described in detail below with reference to the accompanying drawings, wherein:

[0044] Figure 1 is a flow chart of a resume evaluation method according to an embodiment of the present invention;

[0045] Figure 2 is a functional block diagram of a resume evaluation device according to an embodiment of the present invention;

[0046] Figure 3 is a flow chart of a method for evaluating a resume based on a position ID and a resume ID provided by a user according to Embodiment 1 of the present invention;

[0047] Figure 4 is a flow chart of a method for evaluating resumes based on a position ID provided by a user according to Embodiment 2 of the present invention;

[0048] Figure 5 is a flow chart of a method for evaluating a resume based on a position ID and a resume ID provided by a user according to Embodiment 3 of the present invention;

[0049] Figure 6 is a block diagram of the architecture of a resume evaluation system according to another embodiment of the present invention; and

[0050] Figure 7 It is a schematic diagram of the hardware structure principle of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0051] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0052] In the following detailed description, reference may be made to the various specification drawings that are part of the present application and are used to illustrate specific embodiments of the present application. In the accompanying drawings, similar reference numerals describe substantially similar components in different figures. The various specific embodiments of the present application are described below in sufficient detail so that a person of ordinary skill in the art with relevant knowledge and skills in the art can implement the technical solutions of the present application. It should be understood that other embodiments may also be used or structural, logical or electrical changes may be made to the embodiments of the present application.

[0053] The resume in the present invention has different contents according to different scenarios. For example, the resume can be a job seeker's resume, which includes personal information, educational background and professional experience. The resume can also be a resume of a career consultant, which contains similar contents as a job seeker's resume, such as personal information, educational background and professional experience, and can also include the interests and hobbies of the career consultant. The resume can also be the academic performance information of the course learner (referred to as the learner), such as the test site information of each stage of the course and the corresponding test scores, the class attendance information of the learner, the class performance information, etc.

[0054] In the present invention, the information corresponding to the resume is called target information. For example, when the resume is a job seeker's resume, the target information is the position information provided by the recruiter, which can be the position name, the position ID on the recruitment platform, and its specific description information includes, for example, the industry, job responsibilities, skill requirements, salary and benefits, etc.; when the resume is a resume of a career consultant, the target information can be the position information, and its specific description information includes, for example, the industry, skill requirements, job responsibilities, job advancement space, salary range, etc.; when the resume is the student's academic performance information, the target information can be course information, including knowledge point content, test point content and importance level, evaluation level, etc. Unless otherwise specified, the resume in the following description can be a resume in any scenario.

[0055] See also Figure 1 , Figure 1 1 is a flow chart of a resume evaluation method according to an embodiment of the present invention. In this embodiment, the resume evaluation method comprises the following steps:

[0056] Step S11, acquiring description information of the target information based on the target information provided by the first user;

[0057] Step S12, extracting a resume fragment related to the description information of the target information from the resume information of the second user;

[0058] Step S13, inputting the description information of the target information and the resume fragment into a large language model (Large Language Model, LLM for short), and generating multi-dimensional resume evaluation information after processing by the large language model; wherein the large language model is fine-tuned and trained to understand the description information and resume information of the target information and generate resume evaluation information expressing the matching of the resume information with the target information.

[0059] See also Figure 2 , Figure 2 It is a principle block diagram of a resume evaluation device according to an embodiment of the present invention. In this embodiment, the resume evaluation device includes a data acquisition module 11, a data orientation module 12 and an evaluation module 13, wherein the data acquisition module 11 is configured to obtain the description information of the target information from the database 101 based on the target information provided by the first user; the data orientation module 12 is configured to extract the resume fragment related to the description information of the target information from the resume information of the second user; the evaluation module 13 is configured to input the description information of the target information and the resume fragment into the LLM, and generate multi-dimensional resume evaluation information after processing by the LLM.

[0060] In one embodiment, the data orientation module 12 first generates a resume fragment index based on the resume, for example, semantically grouping the resume information of the second user, and taking each obtained semantic group as a resume fragment; calculating the vector of each resume fragment, and multiple resume fragments and their vectors constitute the resume fragment index. Then, multiple target descriptor information is extracted from the description information of the target information, for example, semantic analysis is performed on the description information of the target information to obtain multiple semantic groups, each semantic group is taken as a target descriptor information and a corresponding vector is generated, and then the corresponding resume fragment is retrieved for each target descriptor information in the resume fragment index in turn through the RAG (Retrieval-Augmented Generation) model. The retrieval process is the process of calculating cosine similarity and sorting. Taking a target descriptor information as an example, the cosine similarity between it and each resume fragment vector is calculated respectively, and the cosine similarity is sorted from large to small, and the resume fragment corresponding to the resume fragment vector with the highest sorting is taken as the resume fragment related to the target descriptor information and extracted, that is, a resume fragment is extracted based on a target descriptor information. Alternatively, the resume fragments with cosine similarity greater than a threshold are extracted from the resume information of the second user as resume fragments related to the corresponding target descriptor information, that is, based on one target descriptor information, multiple resume fragments can be extracted or no resume fragment can be extracted.

[0061] The evaluation module 13 generates resume evaluation information using LLM, wherein the LLM is a large language model that has been fine-tuned and trained, such as the Generative Pre-Trained Transformer (GPT) series models developed by Open AI, LaMDA and PaLM developed by Google, LLaMA developed by Meta, Wenxin Yiyan developed by Baidu, Tongyi Qianwen of Alibaba, Hunyuan of Tencent, and a large language model named "AliceMind" developed by Alibaba Damo Academy, etc. Developers in this field can apply any LLM according to actual conditions.

[0062] Taking the recruitment field as an example, the following is an explanation of the LLM training process:

[0063] First, data is collected and preprocessed. In this embodiment, a large amount of job description information, user resume information, and evaluation information of a resume based on the specific job description information are collected. Among them, the job description information and user resume information cover different positions, industries, work experience, etc., to ensure the generalization ability of LLM.

[0064] The collected data is preprocessed, including text cleaning (such as removing irrelevant characters, unifying the format, etc.), word segmentation, and removing stop words.

[0065] Create a sample database. Specifically, for a job description, a user resume, and evaluation information of the user resume based on the job description, extract one or more key contents from the user resume as resume fragments according to the job description, take the job description, the resume fragment, and the corresponding resume evaluation information as a sample, and mark the satisfaction of the resume evaluation information. In addition, the sample can also be constructed in the form of a dialogue to adapt to the dialogue scenario.

[0066] Then select an LLM as a pre-trained model, such as one of the various models mentioned above.

[0067] Then, supervised fine-tuning of the model is performed. Specifically, the samples in the sample database are divided into training set, test set and evaluation set, and the samples in the training set are input to LLM, so that LLM can learn the matching between job description information and resume fragments and generate resume evaluation information in multiple dimensions. Then, the samples are tested with the test set and verified with the samples in the validation set, and the model parameters are stored after meeting the requirements.

[0068] In one embodiment, the LLM can be deployed on a private server to protect the privacy of data, or the training data can be sent to the open source platform for LLM training by calling the LLM interface on the open source platform. The specific application mode of LLM can be selected according to the actual scenario.

[0069] Taking the recruitment field as an example, the resume evaluation method and resume evaluation device of the present invention are described in detail. In the recruitment field, the recruitment platform includes two types of users, one of which is the recruitment end user, such as Figure 1 The first user in the above example is a human resources staff (HR for short) of an enterprise or institution that posts a job opening on the recruitment platform. Figure 1 The target information described in is, for example, the job title or job ID posted by the recruiting user. The other type is the job seeker on the job search side, such as Figure 1 The second user described in the above provides a resume on the recruitment platform. The database 101 of the recruitment platform includes databases for storing various information, such as a user database, including personal information of job seekers, HR information of recruiters, and company information; another example is a resume database, which includes personal resumes of job seekers, which have specific IDs or names and correspond one-to-one with the IDs or names of job seekers in the user database; another example is a position database, which includes description information of each position published on the recruitment platform, which has a specific ID or name and corresponds one-to-one with the ID or name of the position and the ID or name of the recruiter.

[0070] In one scenario, when HR receives a large number of resumes from job seekers for job positions posted on a recruitment platform, HR no longer needs to manually read and evaluate the resumes one by one, but can obtain evaluation information for one or some resumes through the method provided in this embodiment.

[0071] In order to facilitate user operation and obtain evaluation results in real time, the resume evaluation device of the present invention also provides an interactive interface, through which the HR enters the job title or job ID. Figure 1 The resume evaluation information can be obtained by processing the method shown and displayed in real time in the interactive interface.

[0072] Embodiment 1

[0073] See also Figure 3 , Figure 3 4 is a flow chart of a method for evaluating a resume based on a position ID and a resume ID provided by a user according to Embodiment 1 of the present invention.

[0074] Step S111: A user (such as HR) starts a resume evaluation system.

[0075] Step S211: providing an interactive interface based on the startup operation of HR.

[0076] In one embodiment, the resume evaluation system can be implemented as a specific application (App), a recruitment system function module or plug-in, etc. When the HR clicks the login entrance to start, the interactive interface is entered. When it is used as an independent App, the corresponding security verification process is also included when logging in, which is similar to the login process of other Apps and will not be repeated here. The interactive interface at least includes an information input window and an information display window.

[0077] Step S112: HR enters the position title or position ID in the information input window.

[0078] Step S113, HR enters the job seeker's name or resume ID in the information input window.

[0079] Step S212, based on the job title or job ID provided by the HR, the job description information is obtained. The HR identity is determined based on the recruitment user ID determined when the HR logs in, and the job database is queried based on the job title or job ID input by the HR. The corresponding job description information is found based on the correspondence between the job title or job ID and the job description information and the recruitment user identity. In this process, if the corresponding job description information cannot be found, an error message is generated and displayed in the information display window to prompt the HR that the job title or job ID currently input by the HR is invalid.

[0080] Step S213, extracting resume fragments related to the job description information from the target resume information. In this embodiment, the RAG (Retrieval-Augmented Generation) model is used to extract resume fragments related to the job description information from the target resume information. For example, when the job description includes certain skill information, the RAG model is used to extract only information related to the skill from the target resume information, and other information in the resume that is not related to the skills in the job description is discarded. Through the RAG model, the resume content that matches the position can be extracted from the resume information, achieving efficient retrieval of the resume content, thereby making the subsequent evaluation of the resume more targeted and more accurate.

[0081] Step S214, the job description information and the resume fragment are input to the LLM, and the LLM is processed to generate multi-dimensional resume evaluation information. In one embodiment, the dimensions of the resume evaluation information include the second user's strengths, the degree of matching with the target information, the second user's weaknesses and the resume description information. The information content of the strengths and weaknesses, such as skills, education, work experience, personality, etc., and the degree of matching with the position can be measured by grades, scores, etc.

[0082] Step S215, output the resume evaluation information. For example, it is displayed in the information output window of the interactive interface in the form of text. Among them, in order to facilitate user understanding, each dimension information in the resume evaluation information is displayed in the information output window in the form of description text, score or structured label. The description text is a section of content described in natural language, wherein each dimension information is separated by corresponding symbols, such as using semicolons, vertical lines, slashes, etc. to separate adjacent two dimensional contents, or placing the text of one dimension between quotation marks, brackets or book title marks. The score, for example, includes a score and a full score value represented by a specific numerical value, such as 90 (100), 90 / 100, 8 / 10, etc., wherein the preceding numerical value represents a specific evaluation score, and the numerical value in brackets or after the slash represents the full score value of the score system. The evaluation information can also be represented in the form of structured labels, for example, expressing one-dimensional content in the form of AA:BB, the AA before the colon represents the dimension name, the BB after the colon represents the specific content of the dimension, the dimensions are separated by separators such as semicolons, or the overall label is placed between quotation marks, brackets or book title marks. For example: "Advantages: XXXX; Rating: X; Disadvantages: XXXX". Or "Advantages: XXXX", "Rating: X", "Disadvantages: XXXX", "Resume Description: XXXX...XX". The resume description is, for example, a paragraph extracted from a job application resume, or a collection of multiple resume fragments or a paragraph extracted based on multiple resume fragments.

[0083] In addition, in another output embodiment, the output can also be in the form of audio. For example, the text of the resume evaluation information is converted into a voice and output. The voice of the application can be a male voice or a female voice, and the user can set it according to preference. When outputting, check the output audio preference parameters set by the user. If there are any, the audio parameters set by the user are used for output. If not, the system default audio parameters are used for output. In another output embodiment, it can also be output in the form of video / animation. For example, a virtual character is provided as a human character in a video / animation, and virtual characters with different images are provided for users to choose. The user can set the image and voice of the virtual character according to his or her preferences. When determining to output the resume evaluation information in the form of video, check the video and audio parameters set by the user. If there are any, the image of the virtual character set by the user is used as the person in the video, and the set audio parameters are used to convert the text into voice output. When outputting the voice, a specific algorithm is used to calculate the lip shape and expression of the virtual character that matches the voice content, so that the human character in the video is more realistic. If the user does not set a preferred virtual character image or audio parameter, the default virtual character and audio parameters are used to output the resume evaluation information. This embodiment displays the resume evaluation information to the user in a user-friendly manner when outputting the resume evaluation information, so as to facilitate the user to quickly screen and make decisions.

[0084] In another embodiment, the resume evaluation system further includes: Figure 3 In step S216, the feedback request information is output and displayed in the information display window of the interactive interface. The feedback request information can be output in a variety of ways, for example, in a dialog box in the form of text, or in a pop-up window, or in the form of a button in the interactive interface, and when the user clicks the button, the corresponding page or window is entered.

[0085] In step S114, the user inputs feedback information. The feedback information can be provided in different ways. For example, the feedback information can be content described in natural language. In one embodiment, when the feedback request information is displayed in text on the interactive interface, the feedback request information prompts the user to input a piece of information described in natural language in text or voice, so as to directly obtain the text content described in natural language, or the text content described in natural language obtained by recognizing voice. For another example, the feedback information can be fed back in the form of scoring, so that the feedback information obtained is a score. In one embodiment, when the feedback request information is displayed on the interactive interface, the feedback request information prompts the user to score the satisfaction with the current resume evaluation information, and provides a score input method, such as inputting in text or numbers or by moving the slider on the drag bar. The user can enter a specific value through the information input interface, or drag the slider on the currently provided drag bar, and identify the specific value by obtaining the position of the slider on the drag bar.

[0086] Step S217: based on the satisfaction level in the feedback information provided by the user, mark the resume evaluation information and store it in a historical recall database.

[0087] It can be seen from the processing process of the above embodiment that when the present invention uses a large language model to evaluate a resume, the resume fragment extracted from the resume is used as the analysis object. When performing an in-depth evaluation of the resume content, it can not only accurately locate the information related to the position from the resume, accurately obtain the matching of the resume and the position, but also improve the processing speed of the large language model. In addition, the present invention can obtain more information that recruiters are concerned about from other dimensions, such as the strengths and weaknesses of job seekers, and provide recruiters with rich evaluation information.

[0088] Embodiment 2

[0089] See also Figure 4 , Figure 4 This is a flow chart of a method for evaluating resumes based on a position ID provided by a user according to Embodiment 2 of the present invention. Figure 3 , no further details will be given here. This embodiment only describes the method flow of the resume evaluation system for screening resumes based on the position ID provided by HR.

[0090] Step S311, obtaining job description information based on the job ID provided by the user.

[0091] Step S312, determining the resumes of job seekers who have submitted to the position and forming a resume database.

[0092] Step S313, determine a resume from the resume database as a target resume.

[0093] Step S314, extracting resume fragments related to the job description information from the target resume.

[0094] Step S315, searching the historical recall database, and determining the resume description information with the highest satisfaction with the job description information as resume reference information.

[0095] Step S316, the job description information, resume snippet and resume reference information are input to the LLM to generate resume evaluation information. In this embodiment, during the training process of the LLM, the sample also includes resume reference information, and the training process is similar to the above training process, which will not be repeated here.

[0096] Step S317, storing the resume evaluation information, for example, using the position ID, resume ID or job seeker name as the file name of the resume evaluation information, and storing it in a database for users to view.

[0097] Step S318, determine whether there are any resumes in the resume database that have not been evaluated. If so, return to step S313; if not, end the evaluation process.

[0098] In this embodiment, when LLM generates resume evaluation information, it also refers to a resume description information in the historical recall database that has the highest satisfaction with the current position description information as resume reference information, thereby helping LLM to better evaluate the current resume and obtain more objective and rational evaluation information.

[0099] Through the processing of this embodiment, when the HR of a recruiting user receives a large number of resumes for a position, he or she can enter the corresponding position name or position ID in the resume evaluation system to obtain the respective evaluation information of all resumes. The recruiting user can complete the preliminary screening of resumes by reading the resume evaluation information. Compared with manual reading and evaluation, this saves the recruiting user a lot of time and energy.

[0100] After completing the resume screening, the recruiting user can also provide feedback on each resume evaluation information. Figure 4The method shown generates evaluation information for all resumes submitted to a position. When storing the evaluation information, an evaluation information table is generated accordingly, and the evaluation information table includes a corresponding resume link column, an evaluation information column, a satisfaction score column, etc. After the recruitment user retrieves the evaluation information table, the specific evaluation information can be viewed through the table, and the corresponding resume can be retrieved by clicking the resume link. The recruitment user can also fill in the score in the satisfaction score column of the corresponding resume evaluation information. The present invention establishes a corresponding historical recall database, which stores job description information, resume description information consisting of a collection of all resume fragments extracted from the resume when generating resume evaluation information, and satisfaction scores. When the user scores the satisfaction of a resume evaluation information in the evaluation information table, the score is obtained and filled in the historical recall database, so that the resume description information with the highest satisfaction with the job description information can be used as resume reference information in the subsequent resume evaluation.

[0101] On the other hand, through each resume evaluation, the positions, resumes, resume evaluation information and user satisfaction are collected as samples and added to the model training set. The positions, resumes and resume evaluation information are used as samples, and user satisfaction is used as sample labels. When the model training conditions are met, the samples in the training sample library are used to train and update the large language model. For example, the large language model is trained and updated regularly. After the scheduled timing cycle is reached, the large language model is trained with the training set samples with newly added samples; for another example, when the new samples reach the preset number, the large language model is trained with the newly added samples or the training set samples with newly added samples, and the originally stored model parameters are updated after the training is completed. By continuously collecting feedback and adjusting the model, the accuracy and efficiency of resume evaluation can be continuously improved.

[0102] Embodiment 3

[0103] See also Figure 5 , Figure 5It is a flow chart of a method for evaluating a resume based on a position ID and a resume ID provided by a user according to Embodiment 3 of the present invention. In this embodiment, steps S411 to S413 and step S415 correspond to steps S111 to S113 and step S114 of Embodiment 1, steps S511 to S513 correspond to steps S211 to S213 of Embodiment 1, and steps S515 to S518 correspond to steps S214 to S217 of Embodiment 1, and thus are not repeated here. This embodiment differs from Embodiment 1 in that a user can input his or her customized evaluation index information through an interactive interface, as shown in step S414. In this embodiment, a user can customize evaluation indexes as needed, such as "education level above a master's degree." In this embodiment, after receiving the evaluation index information, the resume evaluation system uses it as a job description information, and uses the RAG model to extract the resume fragment related to the evaluation index information from the target resume information in step S514. For the purpose of distinction, the resume fragment related to the job description information extracted from the target resume information in step S513 is called the first resume fragment, and the resume fragment related to the evaluation index information is called the second resume fragment. In step S515, both resume fragments and the evaluation index information are input into the LLM to obtain the resume evaluation information.

[0104] In this embodiment, the user can customize the evaluation indicators according to actual needs so that the evaluation results are more in line with actual recruitment needs.

[0105] In another embodiment, after identifying the evaluation index information input by the user, its weight can be set, and the evaluation index information and its corresponding weight are input into the large language model together with the description information of the target information and the resume fragment, and multi-dimensional resume evaluation information is generated after being processed by the large language model. At this time, the large language model evaluates the description information of the target information and the resume fragment according to the evaluation index information and its corresponding weight, thereby obtaining evaluation information that meets the user input requirements.

[0106] When it is recognized that the user input is preference information, as described above, a weight is set for the preference information; the preference information and its corresponding weight, the description information of the target information and the resume fragment are input into the large language model, and multi-dimensional resume evaluation information is generated after being processed by the large language model. At this time, the large language model evaluates the description information of the target information and the resume fragment according to the preference information and its corresponding weight, thereby obtaining evaluation information that meets the user's preferences.

[0107] See also Figure 6 , Figure 6It is a block diagram of the architecture principle of a resume evaluation system according to another embodiment of the present invention. In this embodiment, the architecture of the resume evaluation system includes a session layer, a data processing layer, a large language model layer and a data layer. Among them, the database in the data layer includes various data in the recruitment platform. The large language model layer provides a variety of large language models. In one embodiment, multiple large language models are deployed, wherein the first LLM is used to communicate with the user at the session layer and understand the information input by the user, and the second LLM is used for resume evaluation. In the data processing layer, corresponding processing is performed on the information obtained from the session layer, including obtaining the description information and resume of the target information from the database in the data layer according to the position ID, extracting resume fragments from the resume, etc., and then the information obtained after data processing is input to the second LLM. After being processed by the second LLM, the resume evaluation information is obtained, which can be sent to the user through the session layer or stored in the data layer.

[0108] In this embodiment, the resume evaluation system can interact with the user through an interactive interface, receive the position ID, the resume ID to be evaluated or the resume collection to be evaluated input by the user, and can also receive the evaluation index information, feedback information and preference information input by the user; provide various information to the user, such as resume evaluation information, notification of resume evaluation completion, storage address of resume evaluation information, feedback request information, etc.

[0109] The large language model used in the aforementioned embodiment of the recruitment field obtains resume evaluation information after one processing. However, it can be obtained that the resume evaluation information can be generated by a large language model, or by multiple large language models or multiple calls to a large language model to generate multiple preliminary resume evaluation information, and then the contents of multiple preliminary resume evaluation information are integrated to obtain the final resume evaluation information to adapt to different types of target information and resume information.

[0110] For other application scenarios, such as career counseling and student learning outcome evaluation, the aforementioned processing methods in the recruitment field can be used to evaluate the consultant's resume and the student's academic performance, thereby facilitating the planning of corresponding job directions for the consultant and helping students choose the next course.

[0111] Figure 7 It is a schematic diagram of the hardware structure principle of an electronic device according to an embodiment of the present invention. The electronic device can be implemented as a server or other various terminal devices, such as a desktop personal computer, a tablet computer, a laptop computer, a mobile phone, etc., including a processor 601 and a memory 602. The memory 602 stores a program instruction set. When the processor 601 executes the program instruction set on the memory 602, the aforementioned various resume evaluation methods are implemented.

[0112] Specifically, the processor 601 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiment of the present invention.

[0113] The memory 602 may include a large capacity memory for data or instructions. By way of example and not limitation, the memory 602 may include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive or a combination of two or more of these. In appropriate cases, the memory 602 may include a removable or non-removable (or fixed) medium. In appropriate cases, the memory 602 may be inside or outside the integrated gateway disaster recovery device. In a specific embodiment, the memory 602 is a non-volatile solid-state memory.

[0114] The memory may include a read-only memory (ROM), a random access memory (RAM), a magnetic disk storage medium device, an optical storage medium device, a flash memory device, an electrical, optical or other physical / tangible memory storage device. Therefore, typically, the memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., a memory device) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the resume evaluation method provided by the present invention.

[0115] In one example, the electronic device may further include a communication interface 603 and a bus 604. The processor 601, the memory 602, and the communication interface 603 are connected via the bus 604 and communicate with each other.

[0116] The communication interface 603 is mainly used to implement communication between various modules, devices, units and / or equipment in the embodiment of the present invention.

[0117] Bus 604 includes hardware, software or both, and the components of online data flow billing equipment are coupled to each other. For example, but not limitation, bus may include accelerated graphics port (AGP) or other graphics bus, enhanced industrial standard architecture (EISA) bus, front side bus (FSB), hypertransport (HT) interconnection, industrial standard architecture (ISA) bus, infinite bandwidth interconnection, low pin count (LPC) bus, memory bus, micro channel architecture (MCA) bus, peripheral component interconnection (PCI) bus, PCI-Express (PCI-X) bus, serial advanced technology attachment (SATA) bus, video electronics standard association local (VLB) bus or other suitable bus or two or more of these combinations. In appropriate cases, bus 604 may include one or more buses. Although the embodiment of the present invention describes and shows a specific bus, the present invention considers any suitable bus or interconnection.

[0118] The present invention also provides a computer-readable storage medium having computer program instructions stored thereon, and the computer program instructions can be executed by a processor to implement any of the resume evaluation methods in the aforementioned embodiments. The computer-readable storage medium can be any tangible medium that contains or stores computer-executable instructions for use by or in combination with an instruction execution system, device, and apparatus. The storage medium can be a transient computer-readable storage medium or a non-transient computer-readable storage medium. Non-transient computer-readable storage media may include, but are not limited to, magnetic storage devices, optical storage devices, and / or semiconductor storage devices. Examples of such storage devices include, for example, magnetic disks, optical disks based on CD, DVD, or Blu-ray technology, and persistent solid-state memories such as flash memory, solid-state drives, and the like.

[0119] The present invention also provides a computer program product, which includes a set of computer program instructions, and when the computer program instruction set is executed by a processor, any one of the resume evaluation methods in the above embodiments is implemented. The computer program product includes but is not limited to application installation packages published on websites and application stores, application plug-ins, and small programs that can be run in certain applications.

[0120] It should be clear that the present invention is not limited to the specific configuration and processing described above and shown in the figures. For the sake of simplicity, a detailed description of the known method is omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present invention is not limited to the specific steps described and shown, and those skilled in the art can make various changes, modifications and additions, or change the order between the steps after understanding the spirit of the present invention.

[0121] The above embodiments are only used to illustrate the present invention, but not to limit the present invention. Ordinary technicians in the relevant technical field can make various changes and modifications without departing from the scope of the present invention. Therefore, all equivalent technical solutions should also fall within the scope of the present invention.

Claims

1. A resume evaluation method, characterized in that: include: Acquire description information of the target information based on the target information provided by the first user; Extracting a resume fragment related to the description information of the target information from the resume information of the second user; as well as Inputting the description information of the target information and the resume fragment into a large language model, and generating multi-dimensional resume evaluation information through processing by the large language model; The large language model is fine-tuned and trained to understand the descriptive information and resume information of the target information and generate resume evaluation information that expresses the matching between the resume information and the target information.

2. The resume evaluation method according to claim 1, characterized in that: The step of extracting a resume segment related to the description information of the target information from the resume information of the second user comprises: Semantically grouping the resume information of the second user, and taking each obtained semantic group as a resume segment; Generate a vector for each resume fragment, and multiple resume fragments and their vectors constitute a resume fragment index; Semantically grouping the description information of the target information, and taking each obtained semantic group as a target description sub-information; Generate a vector for each target descriptor information; Calculate the cosine similarity between each resume segment vector and each target descriptor information vector; and Corresponding to each target descriptor information, the resume segment with the maximum cosine similarity with its vector or the cosine similarity greater than a threshold is extracted from the resume information of the second user as the resume segment related to the corresponding target descriptor information.

3. The resume evaluation method according to claim 1, characterized in that: After extracting the resume fragment related to the description information of the target information from the resume information of the second user, the method further includes: Based on the description information of the target information, a historical recall database is retrieved to determine the resume description information with the highest satisfaction with the description information of the target information as the resume reference information; correspondingly, the description information of the target information, the resume fragment and the resume reference information are input into the large language model.

4. The resume evaluation method according to claim 1, characterized in that: The dimensions of the resume evaluation information include: one or more of the second user's strengths, the degree of matching with the target information, the second user's weaknesses and the resume description information.

5. The resume evaluation method according to claim 1, characterized in that: Each dimension information in the resume evaluation information is provided to the first user in the form of descriptive text, rating or structured label.

6. The resume evaluation method according to claim 1, characterized in that: Further including: An interactive interface is provided, through which input information of a first user is received and generated resume evaluation information is output.

7. The resume evaluation method according to claim 6, characterized in that: Furthermore, the generated resume evaluation information is output in the form of text, audio or video.

8. The resume evaluation method according to claim 6, characterized in that: Further including: Analyzing the input information, when the input information includes feedback information on resume evaluation information, further includes: extracting satisfaction information from the feedback information; as well as The resume evaluation information is annotated based on the satisfaction information and stored in a historical recall database.

9. The resume evaluation method according to claim 8, characterized in that: When the input information includes user preference information, the method further includes: Setting a weight for the preference information; Correspondingly, the preference information and its corresponding weight, the description information of the target information and the resume fragment are input into the large language model, and multi-dimensional resume evaluation information is generated after being processed by the large language model.

10. The resume evaluation method according to claim 1, characterized in that: When the input information includes the user's evaluation index information, the method further includes: Setting weights for the evaluation indicator information; Correspondingly, the evaluation index information and its corresponding weight, the description information of the target information and the resume fragment are input into a large language model, and multi-dimensional resume evaluation information is generated after being processed by the large language model; or Extracting a second resume segment related to the evaluation indicator information from the resume information, wherein the resume segment related to the description information of the target information extracted from the resume information is the first resume segment; Correspondingly, the description information of the target information, the evaluation index information, the first resume fragment and the second resume fragment are input into the large language model, and multi-dimensional resume evaluation information is generated after being processed by the large language model.

11. The resume evaluation method according to claim 1, characterized in that: The first user is a recruiting user, the corresponding target information is job information, the corresponding second user is a job seeker, and the corresponding resume is a job application resume; or, the first user is a career planner, the corresponding target information is career information, the corresponding second user is a career consultant, and the corresponding resume is a resume of a career consultant; Or the first user is a course trainer, the corresponding target information is course information, the corresponding second user is a learner, and the corresponding resume is the learner's academic performance information.

12. A resume evaluation device, characterized in that: include: a data acquisition module configured to acquire description information of the target information based on the target information provided by the first user; a data targeting module configured to extract, from the resume information of the second user, a resume fragment related to the description information of the target information; as well as An evaluation module, configured to input the description information of the target information and the resume fragment into a large language model, and generate multi-dimensional resume evaluation information through processing by the large language model; The large language model is fine-tuned and trained to understand the descriptive information and resume information of the target information and generate resume evaluation information that expresses the matching between the resume information and the target information.

13. An electronic device comprising a processor and a memory, characterized in that: The memory stores a computer program instruction set, and when the processor executes the computer program instruction set on the memory, the resume evaluation method described in any one of claims 1-11 is implemented.

14. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program instruction set, and when the computer program instruction set is executed by a processor, the resume evaluation method according to any one of claims 1 to 11 is implemented.

15. A computer program product, characterized in that It includes a computer program instruction set, which, when executed by a processor, implements the resume evaluation method described in any one of claims 1-11.

Images