An online resume registration model and model updating method

Abstract

The present invention discloses an online resume registration model and a method for updating the model. Through the resume detection model, the method of first detecting the title and then detecting the keywords is used for extraction, so that the personal information, learning experience and work experience in each user resume can be more accurately extracted. Using the extracted resume information, multiple company information representing the company's capabilities and company demand information representing the company's recruitment needs, the one-dimensional information is mapped to a two-dimensional space to obtain an intersecting user circle and company circle. The size of the user circle represents the competitiveness of the user resume in multiple user resumes delivered to a company, and the user circle also reflects the matching with the company's demand information. The size of the company circle represents the competitiveness of a company in multiple companies. Afterwards, the image composed of the user circle and the company circle is used as the input of the resume matching network for detecting the delivery company, so that multiple matching user resumes can be delivered more accurately.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to an online resume registration model and a method for updating the model. Background Art

[0002] At present, the model construction of resume matching generally inputs the data of user resume into the model and matches it with the requirements provided by the company. However, in order to prevent modification, the resume is generally submitted in PDF and other formats. Therefore, it is necessary to extract the data of user resume from these data. And because the current resume matching model is equivalent to a judgment mechanism, it is too rigid and cannot flexibly match user resumes with the needs of multiple companies, and find a more suitable company among those companies that can be submitted. Summary of the invention

[0003] The purpose of the present invention is to provide an online resume registration model and a method for updating the model to solve the above-mentioned problems existing in the prior art.

[0004] In a first aspect, an embodiment of the present invention provides an online resume registration model and a method for updating the model, including:

[0005] Obtaining a user resume, multiple company information and corresponding multiple company demand information; the user resume represents an image containing the user's personal information, education experience and work experience; the company information includes the company's location, company assets, and company work time; the company demand information represents the position education, position skills, and position salary;

[0006] Based on the user resume, resume information is obtained through a resume detection model;

[0007] Based on the resume information and the company demand information, the information in the user resume is mapped to a two-dimensional space to obtain a user circle; the user circle represents the level of the user resume in all the submitted resumes; one resume information and one company demand information correspond to one user circle;

[0008] Based on the user circle, company information and corresponding company demand information, the information of the deliverable company is mapped to a two-dimensional space to obtain a company circle; the company circle represents the level of all companies in the database that can be delivered; one user circle corresponds to one company circle;

[0009] The n company information corresponds to the images of n user circles and company circles, and n double circle images are obtained;

[0010] Based on n double circle images, the intersection state of the company circle and the user circle is detected through the resume matching network to obtain the deliverable companies.

[0011] Optionally, mapping the information in the user resume to a two-dimensional space based on the resume information and the company demand information to obtain a user circle includes:

[0012] Build a blank image;

[0013] Get a position in the blank image as the center of the user circle;

[0014] Based on the resume information and the company's demand information, a user radius is obtained; the user radius is used to judge the competitiveness of the user's resume among multiple submitted resumes;

[0015] A user circle is drawn in the blank image according to the user radius and the user circle center.

[0016] Optionally, based on the user circle, company information and company demand information, the information of the deliverable companies is mapped into a two-dimensional space to obtain multiple company circles, including:

[0017] Get the distance length;

[0018] According to the user circle center corresponding to the user circle, a position with a distance length from the user circle center is randomly obtained as the company circle center;

[0019] Based on the company information, a company radius is obtained; the company radius represents the competitiveness of the company corresponding to the company information among multiple companies to which resumes can be submitted;

[0020] A company circle is drawn in the blank image according to the company radius and the company circle center.

[0021] Optionally, obtaining the user radius based on the resume information and the company demand information includes:

[0022] Obtain all resumes in the database, and grade the personal information, learning experience, and work experience information of multiple users respectively to obtain a resume score table; the rows of the resume score table represent the personal information, learning experience, and work experience information, and the columns represent the scores, and the values ​​in the resume score table are the contents of the personal information, learning experience, and work experience information corresponding to the user resumes;

[0023] Matching the resume information with the demand information of multiple companies, if the match fails, setting the user score to 0;

[0024] If the match is successful, the resume information is matched with the resume score table to obtain the user score;

[0025] Obtain multiple resume scores; the resume score table includes scores of all resumes that can be submitted to the company; the resume score table includes user scores;

[0026] Add up all the values ​​in the resume score table to obtain the total resume score;

[0027] The user score is divided by the total resume score and multiplied by the distance length to obtain the user radius.

[0028] Optionally, obtaining a company radius based on the company information includes:

[0029] Obtain company information corresponding to all companies in the database, grade company locations, company assets, and company working hours, and obtain a company score table; the rows of the company score table represent company locations, company assets, and company working hours, and the columns represent scores, and the values ​​in the company score table are the contents of company locations, company assets, and company working hours;

[0030] Multiple company information corresponds to multiple company score tables;

[0031] Find the corresponding score on the company score table according to the company information to obtain the company score;

[0032] Add up all the scores corresponding to the corresponding values ​​in the company score table to obtain the company score;

[0033] Add up all the scores corresponding to the values ​​in the score tables of multiple companies to get the total score of the company;

[0034] Divide the company score by the total company score to obtain the company ratio;

[0035] Multiply the company ratio by the distance length to obtain the company radius.

[0036] Optionally, the resume detection model includes an image detection network and a keyword detection network.

[0037] Optionally, obtaining resume information based on the user resume through a resume detection model includes:

[0038] Input the user resume into an image detection network, segment it according to the font size, and obtain multiple title font positions and corresponding title texts;

[0039] The area between two adjacent title font positions at the vertical coordinates is used as the title content area; p title fonts correspond to p title content areas;

[0040] Inputting the user resume into a keyword detection network, detecting keywords in the user resume, and obtaining multiple keyword positions;

[0041] According to the title content area, multiple keyword positions are screened to extract keywords related to the title text to obtain the title keyword position;

[0042] Locate the title keyword position, extract the sentence containing the keyword, and obtain the title keyword sentence;

[0043] Input the title keyword sentence into a word segmentation network to perform named entity recognition to obtain multiple keyword groups;

[0044] The keyword corresponding to the keyword position is used as the key in the key-value pair, and the keyword group is used as the value in the key-value pair to construct the key-value pair and obtain the resume information.

[0045] Optionally, the training method of the image detection network includes:

[0046] Acquire a training image; the training image is an image containing texts of various font sizes;

[0047] Acquire annotation data; the annotation data includes annotation text and annotation text position; the annotation text represents text in the training image whose font size is larger than the first font size; the annotation text position represents the position where the annotation text is located; the first font size represents a font size whose number of characters is larger than the number of characters of other font sizes;

[0048] Input the training image into an image detection network, predict the text as a title, and obtain the predicted text and the predicted text position;

[0049] The predicted text and the predicted text position are compared with the labeled data to obtain the loss, and the image detection network is backward trained to obtain a trained image detection network.

[0050] Optionally, the method of detecting the intersection of the company circle and the user circle through a resume matching network based on the n double-circle images to obtain the deliverable companies includes:

[0051] The resume matching network includes a first resume matching network and a second resume matching network;

[0052] According to the double circle image, detecting the intersection area of ​​the company circle and the user circle to obtain the intersection area;

[0053] Get the area of ​​the blank image to get the overall area;

[0054] Divide the intersection area by the overall area to obtain a company tendency ratio; the company tendency ratio indicates a matching ratio between user resumes and companies; n double circle images correspond to obtaining n company tendency ratios;

[0055] Input the double circle image into the first resume matching network, perform convolution, extract image features, and obtain the first convolution feature;

[0056] The first convolutional features and company preference ratio are input into the second resume matching network to obtain the delivery companies.

[0057] Optionally, obtaining a delivery feedback value; the delivery feedback value indicates the company number to which the user's resume was successfully delivered;

[0058] According to the delivery feedback value, obtaining delivery company information and delivery company demand information;

[0059] The user resume, the delivery company information and the delivery company demand information are used to train the resume matching network, and the parameters of the resume matching network are updated.

[0060] Compared with the prior art, the embodiments of the present invention achieve the following beneficial effects:

[0061] The embodiment of the present invention also provides an online resume registration model and a method for updating the model.

[0062] In the present invention, through the resume detection model, the method of first detecting the title and then detecting the keywords is used for extraction, so that the personal information, learning experience and work experience in each user resume can be more accurately extracted. Using the extracted resume information, multiple company information representing the company's capabilities and company demand information representing the company's recruitment needs, the one-dimensional information is mapped to a two-dimensional space to obtain intersecting user circles and company circles. The size of the user circle represents the competitiveness of the user resume in multiple user resumes delivered to a company, and the user circle also reflects the matching with the company's demand information. The size of the company circle represents the competitiveness of a company in multiple companies. Afterwards, the image composed of the user circle and the company circle is used as the input of the resume matching network for detecting the delivery company, so that multiple matching user resumes can be delivered more accurately from large to small according to the degree of matching. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 It is a flow chart of an online resume registration model and a method for updating the model provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0064] The present invention will be described in detail below in conjunction with the accompanying drawings.

[0065] Example 1

[0066] like Figure 1 As shown, an embodiment of the present invention provides an online resume registration model and a method for updating the model, the method comprising:

[0067] S101: Obtain a user resume, multiple company information and corresponding multiple company demand information; the user resume represents an image containing information about the user's personal information, education experience and work experience; the company information includes company location, company assets, and company work time; the company demand information represents position education, position skills, and position salary.

[0068] Among them, one company corresponds to one company information and one company demand information.

[0069] The user's personal information includes the user's name, age, and gender. The learning experience includes the school the user graduated from, the corresponding academic stage for each school, and the corresponding academic qualifications. For example, the school the user graduated from was "xx Primary School", and the corresponding academic stage was "Primary School Stage", the school the user graduated from was "xx University", and the corresponding academic stage was "University Stage", and the user's corresponding academic qualification was "University".

[0070] S102: Based on the user resume, resume information is obtained through a resume detection model.

[0071] The resume information refers to the user's personal information, learning experience and work experience information extracted from the user's resume.

[0072] S103: Based on the resume information and the company demand information, the information in the user resume is mapped to a two-dimensional space to obtain a user circle; the user circle represents the level of the user resume in all submitted resumes; one resume information and one company demand information correspond to one user circle.

[0073] S104: Based on the user circle, company information and corresponding company demand information, map the information of the deliverable companies to a two-dimensional space to obtain a company circle; the company circle represents the level of all the deliverable companies in the database; one user circle corresponds to one company circle;

[0074] S105: Obtain n user circle and company circle images corresponding to n company information, and obtain n double circle images.

[0075] Among them, because the user circle is related to the user profile and the company demand information, the company demand information is related to the company information, and the company information is related to the company circle, the double circle image constructed by 1 user profile and 1 company information is special.

[0076] S106: Based on the n double circle images, the intersection state of the company circle and the user circle is detected through the resume matching network to obtain the deliverable companies.

[0077] Optionally, mapping the information in the user resume to a two-dimensional space based on the resume information to obtain a user circle includes:

[0078] Build a blank image.

[0079] In this embodiment, the size of the blank image is 512*512. The coordinate axis is established with the lower left corner of the blank image as the origin, the length as the horizontal coordinate, and the width as the vertical coordinate. The blank image is a binary image, and the value in the blank image is 255.

[0080] Get a position in the blank image as the center of the user circle.

[0081] In this embodiment, the horizontal coordinate of the blank image is 512 / 3, which is rounded down to 170, and the vertical coordinate is 512 / 3, which is rounded down to 170 as the user circle center. In the blank image, the user circle center is marked with a value of 0 to represent black.

[0082] Based on the resume information and the company's demand information, a user radius is obtained; the user radius is used to judge the competitiveness of the user's resume among multiple submitted resumes.

[0083] Among them, one user radius corresponds to one company.

[0084] A user circle is drawn in the blank image according to the user radius and the user circle center.

[0085] In the blank image, a value of 0 is used to represent a black user circle.

[0086] Optionally, based on the user circle, company information and company demand information, the information of the deliverable companies is mapped into a two-dimensional space to obtain multiple company circles, including:

[0087] Get the distance length.

[0088] The distance length is a fixed value.

[0089] In this embodiment, the distance length is 512*2 / 3, which is rounded to 341.

[0090] According to the user circle center corresponding to the user circle, a position with a distance length from the user circle center is randomly obtained as the company circle center.

[0091] Based on the company information, a company radius is obtained; the company radius represents the competitiveness of the company corresponding to the company information among multiple companies to which resumes can be submitted;

[0092] A company circle is drawn in the blank image according to the company radius and the company circle center.

[0093] Optionally, obtaining the user radius based on the resume information and the company demand information includes:

[0094] Obtain all resumes in the database, and grade the personal information, learning experience, and work experience information of multiple users respectively to obtain a resume score table; the rows of the resume score table represent the personal information, learning experience, and work experience information, and the columns represent the scores, and the values ​​in the resume score table are the contents of the personal information, learning experience, and work experience information corresponding to the user resumes;

[0095] Among them, in this embodiment, for example, the academic qualifications in the learning experience, the required salary in the personal information, and the working hours in the work experience are extracted to construct a resume score table. The user's academic qualifications are divided into doctoral, master's, first-class, and second-class academic qualifications, indicating that students with doctoral, master's, first-class, and second-class academic qualifications are available. The score of a doctoral degree is set to 100, the score of a master's degree is set to 90, the score of a first-class is set to 80, the score of a second-class is set to 70, the score of a junior college is set to 60, the score of a high school is set to 50, the score of a junior high school is set to 40, and the score below junior high school is set to 30. The user's required salary is divided into 0-5,000 yuan with a score of 10, 5,000-8,000 yuan with a score of 20, 8,000-10,000 yuan with a score of 30, 10,000-15,000 yuan with a score of 40, 15,000-20,000 yuan with a score of 50, 20,000-30,000 yuan with a score of 60, 30,000-50,000 yuan with a score of 70, 50,000-100,000 yuan with a score of 80, 100,000-200,000 yuan with a score of 90, and more than 200,000 yuan with a score of 100. The longer the working hours, the higher the score.

[0096] Part of the company score table is shown in Table 1 below:

[0097] Table 1

[0098]

[0099]

[0100] The resume information is matched with the demand information of multiple companies. If the match fails, the user score is set to 0.

[0101] Among them, as in this embodiment, if the resume information is "college degree" and the professional qualification in the company's demand information is "master's degree", the matching fails.

[0102] If the match is successful, the resume score table is used to match the resume information with the resume score table to obtain the user score.

[0103] Among them, the resume information shows that the working time is 3 years, the educational background is college degree, the required salary is 5,000-8,000 yuan, and the user score is 30+60+20=110.

[0104] Obtain multiple resume scores; the resume score table includes scores of all resumes that can be submitted to the company; the resume score table includes user scores;

[0105] Add up all the values ​​in the resume score table to obtain the total resume score;

[0106] The user score is divided by the total resume score and multiplied by the distance length to obtain the user radius.

[0107] Among them, the greater the ability of the user on his resume and the lower his own needs, the larger the user radius.

[0108] Optionally, obtaining a company radius based on the company information includes:

[0109] Obtain company information corresponding to all companies in the database, classify company locations, company assets, and company working hours, and obtain a company score table; the rows of the company score table represent company locations, company assets, and company working hours, and the columns represent scores. The values ​​in the company score table are the contents of company locations, company assets, and company working hours. The company location represents the distance of the user from the company.

[0110] Multiple company information corresponds to multiple company score tables.

[0111] The part of the company score table is shown in Table 2 below:

[0112] Table 2

[0113] Company Location Company assets Company working hours 10 More than 6 km Less than 500,000 yuan 5.5-6 hours 20 Within 5-6 km 500,000-1 million yuan 6-6.5 hours 40 Within 4-5 km 1 million-2 million yuan 6.5-7 hours 50 Within 3-4 km 2 million-5 million yuan 7-7.5 hours 60 Within 2-3 km 5 million-10 million yuan 7.5-8 hours 70 Within 1-2 km 10 million-15 million yuan 8-8.5 hours

[0114] Among them, 1 company corresponds to 1 company information and 1 company score table.

[0115] According to the company information, the corresponding score is found on the company score table to obtain the company score.

[0116] Add up all the scores corresponding to the values ​​in the score tables of multiple companies to get the total score of the company;

[0117] Divide the company score by the total company score to obtain the company ratio;

[0118] Multiply the company ratio by the distance length to obtain the company radius.

[0119] Optionally, the resume detection model includes an image detection network and a keyword detection network.

[0120] Optionally, based on the user resume, resume information is obtained through a resume detection model, including:

[0121] The user resume is input into an image detection network and segmented according to font size to obtain multiple title font positions and corresponding title texts.

[0122] The image detection network is a yolo5 model. The image detection network is used to detect fonts that are larger than other fonts, which is equivalent to detecting titles.

[0123] The area between two adjacent title font positions at the vertical coordinates is used as the title content area; p title content areas are obtained corresponding to p title fonts.

[0124] Wherein, p is a natural number greater than 0. When a title font position with the smallest ordinate is detected, an area with a ordinate smaller than the title font position is taken as the title content area.

[0125] Among them, if the title text is "Learning Experience", "Work Experience", "Personal Information", and the vertical coordinates of "Learning Experience" and "Work Experience" are adjacent, then the area between "Learning Experience" and "Work Experience" is set as the title content area.

[0126] The user resume is input into a keyword detection network, key values ​​in the user resume are detected, and multiple keyword positions are obtained.

[0127] Among them, the keyword detection network in this embodiment is a convolutional neural network (CNN). The training data used by the keyword detection network is a plurality of images containing text, and the annotation data is the number of the annotated keyword; the training data is input into the keyword detection network to obtain the predicted keyword number; the predicted keyword number and the annotation data are used to calculate the loss, and the keyword detection network is backward trained to obtain a trained keyword detection network.

[0128] Among them, in this embodiment, the keywords of the user resume are such as "education", "working hours" and "salary".

[0129] The keyword position includes the horizontal coordinate and the vertical coordinate of the center point of the bounding box containing the title text, and the width and height of the bounding box.

[0130] According to the title content area, multiple keyword positions are screened, keywords related to the title text are extracted, and the title keyword positions are obtained.

[0131] If the title text corresponding to the title content area is "Learning Experience", then if the keyword "salary" is detected, it means the salary of the user when he was an intern or part-time worker during the study period, rather than the salary status expected by the company represented by the "salary" in "Working Experience". Therefore, the keyword "salary" in "Learning Experience" is filtered out.

[0132] Wherein, a plurality of keywords are stored under the title text, and matching and screening are performed through the keywords stored under the title text.

[0133] The title keyword position is located, and the sentence containing the keyword is extracted to obtain the title keyword sentence.

[0134] The text before and after the keyword is detected in sequence, and the text between the two "." is stopped, and all the text between the two "." is extracted to form the title keyword sentence.

[0135] The title keyword sentence is input into a word segmentation network to perform named entity recognition to obtain multiple keyword groups.

[0136] Among them, in this embodiment, the word segmentation network is a BiLSTM+CRF model.

[0137] In this embodiment, for example, the keyword phrase corresponding to the detected keyword "education" is "college", the keyword phrase corresponding to the detected keyword "working hours" is "3 years", and the keyword phrase corresponding to the detected keyword "required salary" is "5000-8000 yuan". When a keyword is detected, information about the content of the keyword expressed around the keyword can be obtained.

[0138] The keyword corresponding to the keyword position is used as the key in the key-value pair, and the keyword group is used as the value in the key-value pair to construct the key-value pair and obtain the resume information.

[0139] Optionally, the training method of the image detection network includes:

[0140] A training image is obtained; the training image is an image containing texts of various font sizes.

[0141] Among them, there are texts in font sizes such as "Small Four" and "No. 6".

[0142] Acquire annotation data; the annotation data includes annotation text and annotation text position; the annotation text represents text in a training image whose font size is larger than a first font size; the annotation text position represents the position of the annotation text; the first font size represents a font size whose number of characters is larger than the number of characters of other font sizes.

[0143] In this embodiment, if the entire user resume is entered in the font size of "small four" as the main text size, but the title text is entered in the font size of "large four", "small four" is used as the first font size, and the annotation data is the text in the font size of "large four" and the position of the text;

[0144] The training image is input into an image detection network, predicted as text of the title, and the predicted text and the predicted text position are obtained.

[0145] The predicted text and the predicted text position are compared with the labeled data to obtain the loss, and the image detection network is backward trained to obtain a trained image detection network.

[0146] The loss function adopts a cross entropy loss function. The predicted text and the annotated text are used to obtain the loss, and the annotated text position and the predicted text position are used to obtain the loss.

[0147] Optionally, the method of detecting the intersection of the company circle and the user circle through a resume matching network based on the n double-circle images to obtain the deliverable companies includes:

[0148] The resume matching network includes a first resume matching network and a second resume matching network.

[0149] Among them, the first resume matching network is a convolutional neural network (CNN), and the second resume matching network is a fully-connected neural network (FCNN).

[0150] According to the double circle image, detecting the intersection area of ​​the company circle and the user circle to obtain the intersection area;

[0151] The intersection area is calculated based on the center and radius of the company circle and the center and radius of the user circle in the double circle image.

[0152] Get the area of ​​the blank image to get the overall area.

[0153] The reason why the area of ​​the blank image is used instead of the area of ​​the user circle is that the user circles corresponding to different user profiles change with the company information. So if there are m user circles, there will be n*m ​​double circle images. And the company can find a more matching user profile from multiple user profiles, thus performing a two-way matching.

[0154] Wherein, n and m are both natural numbers greater than 0.

[0155] Divide the intersection area by the overall area to obtain a company tendency ratio; the company tendency ratio indicates a matching ratio between user resumes and companies; n double circle images correspond to obtaining n company tendency ratios;

[0156] Input the double circle image into the first resume matching network, perform convolution, extract image features, and obtain the first convolution feature;

[0157] The first convolutional features and company preference ratio are input into the second resume matching network to obtain the delivery companies.

[0158] Optionally, obtaining a delivery feedback value; the delivery feedback value indicates the company number to which the user's resume was successfully delivered;

[0159] According to the delivery feedback value, obtaining delivery company information and delivery company demand information;

[0160] The user resume, the delivery company information and the delivery company demand information are used to train the resume matching network, and the parameters of the resume matching network are updated.

[0161] The algorithm and display provided herein are not inherently related to any particular computer, virtual system or other device. Various general purpose systems can also be used together with the teachings based on this. According to the above description, it is obvious that the structure required for constructing such systems. In addition, the present invention is not directed to any specific programming language either. It should be understood that various programming languages ​​can be utilized to realize the content of the present invention described herein, and the description of the above specific languages ​​is for disclosing the best mode of the present invention.

[0162] In the description provided herein, a large number of specific details are described. However, it is understood that embodiments of the present invention can be practiced without these specific details. In some instances, well-known methods, structures and techniques are not shown in detail so as not to obscure the understanding of this description.

Claims

1. A method for updating an online resume registration model, characterized in that: include: Obtaining a user resume, multiple company information and corresponding multiple company demand information; the user resume represents an image containing the user's personal information, education experience and work experience; the company information includes the company's location, company assets, and company work time; the company demand information represents the position education, position skills, and position salary; Based on the user resume, the resume information is obtained through a resume detection model; Based on the resume information and company demand information, the information in the user resume is mapped to a two-dimensional space to obtain a user circle, including: Construct a blank image; obtain a position in the blank image as the user circle center; obtain a user radius based on the resume information and the company demand information; the user radius is used to judge the competitiveness of the user resume among multiple resumes submitted; draw a user circle in the blank image according to the user radius and the user circle center; the user circle represents the level of the user resume among all the resumes submitted; one resume information and one company demand information correspond to one user circle; Based on the user circle, company information and corresponding company demand information, the information of the deliverable companies is mapped to a two-dimensional space to obtain a company circle, including: Obtain the distance length; according to the user circle center corresponding to the user circle, randomly obtain a position with a distance length from the user circle center as the company circle center; based on the company information, obtain the company radius; the company radius represents the competitiveness of the company corresponding to the company information among multiple companies that can submit resumes; according to the company radius and the company circle center, draw a company circle in a blank image; the company circle represents the level of all companies that can submit resumes in the database; one user circle corresponds to one company circle; The n company information corresponds to the images of n user circles and company circles, and n double circle images are obtained; Based on n double circle images, the intersection of the company circle and the user circle is detected through the resume matching network to obtain the companies that can be delivered; Obtain a delivery feedback value; the delivery feedback value indicates the company number to which the user's resume was successfully delivered; According to the delivery feedback value, obtaining delivery company information and delivery company demand information; The user resume, the delivery company information and the delivery company demand information are used to train the resume matching network, and the parameters of the resume matching network are updated.

2. The updating method of an online resume registration model according to claim 1, characterized in that: The obtaining of the user radius based on the resume information and the company demand information includes: Obtain all resumes in the database, and grade the personal information, learning experience, and work experience information of multiple users respectively to obtain a resume score table; the rows of the resume score table represent the personal information, learning experience, and work experience information, and the columns represent the scores, and the values ​​in the resume score table are the contents of the personal information, learning experience, and work experience information corresponding to the user resumes; Matching the resume information with the demand information of multiple companies, if the match fails, setting the user score to 0; If the match is successful, the resume information is matched with the resume score table to obtain the user score; Add up all the values ​​in the resume score table to obtain the total resume score; The user score is divided by the total resume score and multiplied by the distance length to obtain the user radius.

3. The method for updating an online resume registration model according to claim 1, characterized in that: The obtaining of the company radius based on the company information includes: Obtain company information corresponding to all companies in the database, grade company locations, company assets, and company working hours, and obtain a company score table; the rows of the company score table represent company locations, company assets, and company working hours, and the columns represent scores, and the values ​​in the company score table are the contents of company locations, company assets, and company working hours; Multiple company information corresponds to multiple company score tables; Find the corresponding score on the company score table according to the company information to obtain the company score; Add up all the scores corresponding to the corresponding values ​​in the company score table to obtain the company score; Add up all the scores corresponding to the values ​​in the score tables of multiple companies to get the total score of the company; Divide the company score by the total company score to obtain the company ratio; Multiply the company ratio by the distance length to obtain the company radius.

4. The method for updating an online resume registration model according to claim 1, characterized in that: The resume detection model includes an image detection network and a keyword detection network.

5. The method for updating an online resume registration model according to claim 4, characterized in that: The obtaining of resume information based on the user resume through a resume detection model includes: Input the user resume into an image detection network, segment it according to the font size, and obtain multiple title font positions and corresponding title texts; The area between two adjacent title font positions at the vertical coordinates is used as the title content area; p title fonts correspond to p title content areas; Inputting the user resume into a keyword detection network, detecting keywords in the user resume, and obtaining multiple keyword positions; According to the title content area, multiple keyword positions are screened to extract keywords related to the title text to obtain the title keyword position; Locate the title keyword position, extract the sentence containing the keyword, and obtain the title keyword sentence; Input the title keyword sentence into a word segmentation network to perform named entity recognition to obtain multiple keyword groups; The keyword corresponding to the keyword position is used as the key in the key-value pair, and the keyword group is used as the value in the key-value pair to construct the key-value pair and obtain the resume information.

6. The method for updating an online resume registration model according to claim 5, characterized in that: The training method of the image detection network comprises: Acquire a training image; the training image is an image containing texts of various font sizes; Acquire annotation data; the annotation data includes annotation text and annotation text position; the annotation text represents text in the training image whose font size is larger than the first font size; the annotation text position represents the position where the annotation text is located; the first font size represents a font size whose number of characters is larger than the number of characters of other font sizes; Input the training image into an image detection network, predict the text as a title, and obtain the predicted text and the predicted text position; The predicted text and the predicted text position are compared with the labeled data to obtain the loss, and the image detection network is backward trained to obtain a trained image detection network.

7. The method for updating an online resume registration model according to claim 1, characterized in that: Based on the n double circle images, the state of intersection between the company circle and the user circle is detected through the resume matching network to obtain the deliverable companies, including: The resume matching network includes a first resume matching network and a second resume matching network; According to the double circle image, detecting the intersection area of ​​the company circle and the user circle to obtain the intersection area; Get the area of ​​the blank image to get the overall area; Divide the intersection area by the overall area to obtain a company tendency ratio; the company tendency ratio indicates a matching ratio between user resumes and companies; n double circle images correspond to obtaining n company tendency ratios; Input the double circle image into the first resume matching network, perform convolution, extract image features, and obtain the first convolution feature; Input the first convolutional feature and the company preference ratio into the second resume matching network to obtain multiple companies that can be submitted; Send user resumes to available companies according to the company preference ratio from large to small.

Images