Resume and post matching method and device based on double-tower model, equipment and storage medium

Abstract

The application discloses a resume and post matching method and device based on a double-tower model, equipment and a storage medium, and relates to the technical field of big data and deep learning. The method comprises the following steps: extracting resume words and post words from a resume text and a post description text, and determining corresponding resume word vectors and post word vectors; extracting post keywords from the post description text, determining a first importance vector of the explicit influence of the post keywords on the resume words, and splicing the resume word vectors and the first importance vector to obtain a resume fusion vector; determining a second importance vector of the implicit influence of an industry on the resume text based on a dynamic knowledge graph of the industry; inputting the resume fusion vector into a resume tower to obtain a resume intermediate vector, and inputting the post word vector into a post tower to obtain a post intermediate vector; and determining a matching result between the resume text and the post description text according to the resume intermediate vector, the post intermediate vector and the second importance vector.

Description

TECHNICAL FIELD

[0001] Embodiments of the present application relate to the technical field of big data and deep learning, in particular to a resume and post matching method and device based on a double tower model, equipment and storage medium. BACKGROUND

[0002] The double tower model is a machine learning model composed of two independent deep neural networks, one tower is responsible for encoding user or query information, and the other tower is responsible for encoding item or document information, and the similarity (such as cosine similarity) between the two is calculated to make recommendations or matching.

[0003] However, although the double tower model performs well in recommendation systems, its accuracy is still difficult to meet the needs when applying this model to post and resume matching. This is mainly because the matching of post and resume involves more complex and multi-dimensional information, and the double tower model may be difficult to fully capture and accurately evaluate these subtle differences, resulting in inaccurate matching results. SUMMARY

[0004] The present application provides a resume and post matching method and device based on a double tower model to improve the accuracy of the matching results.

[0005] According to an aspect of the present application, a resume and post matching method based on a double tower model is provided, the method comprising:

[0006] extracting resume words and post words from resume text and post description text, and determining corresponding resume word vectors and post word vectors;

[0007] extracting post keywords from the post description text, determining a first importance vector of the post keywords on the explicit influence of the resume words, and concatenating the resume word vector and the first importance vector to obtain a resume fusion vector;

[0008] determining a second importance vector of the industry on the implicit influence of the resume text based on a dynamic knowledge graph of the industry;

[0009] inputting the resume fusion vector into a resume tower to obtain a resume intermediate vector, and inputting the post word vector into a post tower to obtain a post intermediate vector;

[0010] determining the matching result between the resume text and the post description text according to the resume intermediate vector, the post intermediate vector and the second importance vector.

[0011] According to another aspect of the present application, a resume and post matching device based on a double tower model is provided, the device comprising:

[0012] a word vector module configured to extract resume words and job description words from the resume text and the job description text respectively, and determine corresponding resume word vectors and job description word vectors;

[0013] a first importance module configured to extract job keywords from the job description text, determine a first importance vector of explicit influence of the job keywords on the resume words, and splice the resume word vectors and the first importance vector to obtain a resume fusion vector;

[0014] a second importance module configured to determine a second importance vector of implicit influence of an industry on the resume text based on a dynamic knowledge graph of the industry;

[0015] an intermediate vector module configured to input the resume fusion vector into a resume tower to obtain a resume intermediate vector, and input the job description word vector into a job tower to obtain a job intermediate vector;

[0016] a matching module configured to determine a matching result between the resume text and the job description text according to the resume intermediate vector, the job intermediate vector, and the second importance vector.

[0017] According to another aspect of the present application, an electronic device is provided, which includes:

[0018] one or more processors;

[0019] a memory configured to store one or more programs;

[0020] When the one or more programs are executed by the one or more processors, the one or more processors implement any one of the resume and job matching methods based on the double-tower model provided by the embodiments of the present application.

[0021] According to another aspect of the present application, a computer readable storage medium is provided, which stores a computer program, and the program is executed by a processor to implement any one of the resume and job matching methods based on the double-tower model provided by the embodiments of the present application.

[0022] According to another aspect of the present application, a computer program product is provided, which includes a computer program, and the computer program is executed by a processor to implement any one of the resume and job matching methods based on the double-tower model provided by the embodiments of the present application.

[0023] The application determines a first importance vector of a post keyword in a post description text which has a dominant influence on a resume word, splices the resume word vector and the first importance vector to obtain a resume fusion vector, determines a second importance vector of an industry which has an implicit influence on the resume text based on a dynamic knowledge graph of the industry, takes the resume fusion vector as an input of a resume tower and the post word vector as an input of a post tower, respectively processes to obtain a resume intermediate vector and a post intermediate vector, and matches the resume text and the post description text in combination with the resume intermediate vector, the post intermediate vector and the second importance vector. The above process realizes the dual goals of precise matching of explicit requirements and intelligent mining of implicit requirements through the cooperation of the staged explicit importance coding and the implicit importance coding, and improves the accuracy of the resume and post matching. BRIEF DESCRIPTION OF DRAWINGS

[0024] Figure 1 FIG. 1 is a flowchart of a resume and post matching method based on a double-tower model according to an embodiment of the application;

[0025] Figure 2a FIG. 2 is a flowchart of a resume and post matching method based on a double-tower model according to an embodiment of the application;

[0026] Figure 2b FIG. 3 is a structural diagram of a double-tower model according to an embodiment of the application;

[0027] Figure 3 FIG. 4 is a structural diagram of a resume and post matching device based on a double-tower model according to an embodiment of the application;

[0028] Figure 4 FIG. 5 is a structural diagram of an electronic device implementing a resume and post matching method based on a double-tower model according to an embodiment of the application. DETAILED DESCRIPTION

[0029] In order to enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should be within the scope of protection of the present application.

[0030] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and in the above drawings are used to distinguish similar objects, and do not necessarily have to describe a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not have to be limited to only those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0031] In addition, it should also be noted that in the technical solutions of the present application, the collection, storage, use, processing, transmission, provision and disclosure of the gear data and related data such as available stiffness involved in the technical solutions of the present application all comply with the relevant legal regulations and do not violate public order and good customs.

[0032] Embodiment one

[0033] Figure 1 A flowchart of a resume and post matching method based on a double-tower model according to Embodiment One of the present application is shown. The present embodiment can be applied to the matching of post requirements of a hiring party and resumes of a job applicant, and can be executed by a resume and post matching device based on a double-tower model. The resume and post matching device based on a double-tower model can be realized in the form of hardware and / or software, and can be configured in a computer device, such as a client or a server. As shown in the figure, the method comprises: Figure 1

[0034] S110, extracting resume words and post words from resume text and post description text, and determining corresponding resume word vectors and post word vectors;

[0035] S120, extracting post keywords from the post description text, determining a first importance vector of the post keywords on the explicit influence of the resume words, and splicing the resume word vector and the first importance vector to obtain a resume fusion vector;

[0036] S130, determining a second importance vector of the industry on the implicit influence of the resume text based on a dynamic knowledge graph of the industry;

[0037] S140, inputting the resume fusion vector into a resume tower to obtain a resume intermediate vector, and inputting the post word vector into a post tower to obtain a post intermediate vector;

[0038] ​S150, determining a matching result between the resume text and the job description text according to the resume intermediate vector, the job intermediate vector and the second importance vector.

[0039] In the embodiments of the present application, the resume text and the job description text can be preprocessed, so that the text information can be effectively understood and processed. The preprocessing can include text standardization and word embedding. The text standardization can include removing stop words, removing punctuation marks and unifying case. Among them, the stop word refers to a word that frequently appears in the text but contributes less to the semantics, such as "of", "is", "in" and the like. By constructing a stop word list, each word in the resume and the job description is traversed, and these stop words are deleted to reduce noise and improve the efficiency of subsequent processing. Punctuation marks usually have no actual meaning in text analysis, so they need to be removed from the text. Regular expressions can be used to match all punctuation marks and replace them with an empty string. For unified case, in order to avoid matching problems caused by different cases, all texts should be converted to lowercase or uppercase. This can be achieved by calling the conversion function of the string.

[0040] In the embodiments of the present application, the resume text and the job description text are obtained respectively. In the case that the resume data includes certificate images, video resumes, portfolio links and other multi-modal data, the multi-modal data can be recognized, for example, image processing to obtain the resume text. The resume text and the job description text are preprocessed respectively to remove stop words, remove punctuation marks and unify case, to obtain preprocessed resume text and job description text, and the preprocessed resume text and job description text are processed to obtain resume words and job words. The pre-trained word embedding model is used to convert the resume words and the job words into vector representations to obtain resume word vectors and job word vectors.

[0041] Moreover, the job keywords can also be extracted from the job description text, that is, the job words are screened to obtain the job keywords. For example, the frequency of the job words in the job description text can be counted, and the job words can be screened according to the frequency to obtain at least one job keyword, and then the vector representation of the corresponding job keyword is obtained. Moreover, the similarity between the resume word vector and the vector representation of the job keyword can be determined to obtain a first importance vector of the explicit influence of the job keyword on the resume word; the first importance vector of the resume word is spliced with the corresponding resume word vector to obtain a resume fusion vector of the resume word, so that the resume fusion vector not only contains the semantic information of the resume word, but also contains the explicit influence of the job keyword on the resume word, so that the subsequent resume and job matching process can focus on the explicit needs of the job, thereby improving the accuracy of the matching result.

[0042] The dynamic knowledge graph of the industry at least includes technical entities of the industry to which the job description text belongs, hierarchical relationships between different technical entities, and can further include attribute data of the technical entities. The dynamic knowledge graph of the industry is updated according to hot news of the industry. Exemplarily, the resume text is matched with the dynamic knowledge graph of the industry to obtain a matched entity node; and the second importance vector of the implicit influence of the industry on the resume text is determined by combining the attribute data of the corresponding entity node with the matched entity node. Since the dynamic knowledge graph of the industry contains complex relationships between different technical entities in the industry, such as industry trends, skill requirements, and inter-enterprise connections; matching the resume text with the dynamic knowledge graph can mine potential demand information that is not explicitly mentioned in the job description text but is closely related to the industry, and calculate the matching degree between the resume text and the potential demand information as the second importance vector of the implicit influence of the industry to which the job demand party belongs on the resume text. That is, the second importance vector is obtained by processing the resume text through the dynamic knowledge graph of the industry, and is used to mine potential demand information that is not explicitly mentioned in the job description text but is related to the industry.

[0043] The resume tower and the job tower can each include an N-layer network, where the i-th layer is an intermediate layer, i and N are natural numbers, and i is less than N. For example, the resume tower and the job tower each have 5 layers, and the 3rd layer is an intermediate layer. The semantic information of the resume word and the resume fusion vector of the explicit influence of the job keyword on the resume word are input into the resume tower to obtain a resume intermediate vector output by the i-th layer in the resume tower. The job word vector is input into the job tower to obtain a job intermediate vector output by the i-th layer in the job tower; the resume intermediate vector, the job intermediate vector, and the second importance vector are processed to obtain a matching result between the resume text and the job description text; exemplarily, the second importance vector and the job intermediate vector can be weighted and fused to generate a fused job intermediate vector; and the matching degree between the resume intermediate vector and the fused job intermediate vector is calculated to obtain the matching result between the resume text and the job description text.

[0044] The embodiment of the application obtains the first importance vector of the explicit influence of the post key word on the resume word by extracting the post key word from the post description text, calculating the similarity between the vector representation of the resume word vector and the post key word, and splicing the first importance vector of the resume word with the corresponding resume word vector to obtain the resume fusion vector of the resume word. The resume fusion vector not only contains the semantic information of the resume word, but also incorporates the explicit influence of the post key word on the resume word. Moreover, the second importance vector of the implicit influence of the industry to which the post demander belongs on the resume text is obtained by processing the resume text through the dynamic knowledge graph of the industry. The resume fusion vector is input into the resume tower, and the post word vector is input into the post tower to obtain the resume intermediate vector and the post intermediate vector, respectively. The matching result between the resume text and the post description text is obtained by processing the resume intermediate vector, the post intermediate vector and the second importance vector, so that the matching result not only considers the explicit demand of the post description text, but also takes into account the implicit demand of the industry to which the post demander belongs, thereby improving the accuracy of the resume and post matching.

[0045] Embodiment two

[0046] Figure 2a is a flowchart of a resume and post matching method based on a double-tower model according to the second embodiment of the application. Referring to Figure 2a On the basis of the technical solutions of the above-mentioned embodiments, the resume and post matching method based on the double-tower model of the present embodiment can include:

[0047] S210, extracting resume words and post words from the resume text and the post description text, respectively, and determining corresponding resume word vectors and post word vectors;

[0048] S220, extracting post key words from the post description text, determining a first importance vector of the explicit influence of the post key words on the resume words, and splicing the resume word vectors and the first importance vector to obtain a resume fusion vector;

[0049] S230, determining a second importance vector of the implicit influence of the industry on the resume text based on the dynamic knowledge graph of the industry;

[0050] S240, inputting the resume fusion vector into the resume tower to obtain a resume intermediate vector, and inputting the post word vector into the post tower to obtain a post intermediate vector;

[0051] S250, inputting the resume intermediate vector and the post intermediate vector into the cross attention layer to obtain the semantic correlation degree between the resume text and the post description text;

[0052] S260, based on the gating mechanism, the second importance vector is taken as an adjustment factor to dynamically adjust the semantic correlation degree, and a fusion correlation degree between the resume text and the post description text is obtained;

[0053] S270, according to the resume intermediate vector, the post intermediate vector and the fusion correlation degree, a matching result between the resume text and the post description text is determined.

[0054] Reference Figure 2b In the processing of the resume intermediate vector, the post intermediate vector and the first importance vector, the resume intermediate vector and the post intermediate vector can be input into the cross attention layer. For example, the resume intermediate vector can be projected into a query matrix, the post intermediate vector can be projected into a key matrix and a value matrix respectively, and the semantic correlation degree between the resume text and the post description text can be calculated according to the query matrix, the key matrix and the value matrix. In addition, the semantic correlation degree and the second importance vector are spliced to obtain a spliced vector, the spliced vector is input into the gating network to obtain the fusion correlation degree between the resume text and the post description text. The gating network can use a multilayer perceptron (MLP), a gated recurrent unit (GRU) or a gating unit in a transformer. The resume intermediate vector, the post intermediate vector and the fusion correlation degree are processed to obtain the matching result between the resume text and the post description text. Through the splicing operation and the gating network, the direct semantic information and the implicit demand of the industry can be effectively fused, and the accuracy and the interpretability of the matching can be improved.

[0055] In an optional embodiment, according to the resume intermediate vector, the post intermediate vector and the fusion correlation degree, a matching result between the resume text and the post description text is determined, including: splicing the resume intermediate vector and the fusion correlation degree to input the next layer of the resume tower to obtain a resume output vector; splicing the post intermediate vector and the fusion correlation degree to input the next layer of the post tower to obtain a post output vector; calculating the similarity between the resume output vector and the post output vector to obtain the matching result between the resume text and the post description text.

[0056] Exemplarily, the resume intermediate vector and the fusion correlation degree are spliced, input into the next layer of the resume tower, and a resume output vector is obtained; the post intermediate vector and the fusion correlation degree are spliced, input into the next layer of the post tower, and a post output vector is obtained; and a similarity between the resume output vector and the post output vector is calculated to obtain a matching result between the resume text and the post description text. By combining the resume intermediate vector and the post intermediate vector with the fusion correlation degree respectively, and through further deepening analysis of the tower layer, the potential connection between the resume and the post can be more comprehensively captured; and by calculating the similarity between the resume output vector and the post output vector, the matching result between the resume text and the post description text can be more accurately obtained.

[0057] Moreover, various similarities such as cosine similarity, Euclidean distance and Manhattan distance can be used to more comprehensively evaluate the similarity between the resume text and the post description text. Exemplarily, the cosine similarity, the Euclidean distance and the Manhattan distance between the resume output vector and the post output vector can be calculated respectively, and each similarity is weighted by using the corresponding weight, wherein the weights of the cosine similarity, the Euclidean distance and the Manhattan distance can be 0.5, 0.3 and 0.2 respectively.

[0058] The embodiment of the present application can dynamically adjust the semantic correlation degree between the resume text and the post description text by introducing the second importance vector based on the gating mechanism as an adjusting factor, thereby further improving the accuracy and flexibility of the matching result; and by splicing the resume intermediate vector and the post intermediate vector with the fusion correlation degree respectively and inputting them into the respective processing towers, the deep features of the resume and the post information can be fully utilized; and by combining the resume intermediate vector and the post intermediate vector with the fusion correlation degree respectively, the matching result between the resume text and the post description text can be more accurately obtained.

[0059] In an optional embodiment, the industry-based dynamic knowledge graph determines a second importance vector of an industry on an implicit influence of the resume text, including: matching resume terms in the resume text with entities in the dynamic knowledge graph to obtain a first entity that is successfully matched, and obtaining a second entity directly dependent on the first entity from the dynamic knowledge graph; and determining the second importance vector of the industry on the implicit influence of the resume text according to demand heat attributes of the first entity and the second entity.

[0060] The dynamic knowledge graph of the industry includes not only technical entities (such as "PyTorch" and "Kubernetes") and technical attributes (such as version number and application scenario) of the technical entities, but also demand heat attributes of the technical entities. The demand heat attribute refers to the degree of demand for the technical entity by the industry or enterprise in the current time period, and is used to measure the market value and application prospect of the technical entity. For example, the demand heat attribute can be determined according to the discussion heat (such as the number of posts and the number of replies) of the technical entity in the community, the solution speed and attention of the technical entity related problems, and the like.

[0061] For example, the resume words in the resume text are matched with the entities in the dynamic knowledge graph to obtain a first entity that is successfully matched. The second entity directly dependent on the first entity is obtained from the dynamic knowledge graph. For example, if the first entity is "PyTorch", the second entity can be "Python" or "deep learning". The demand heat attributes of the first entity and the second entity are obtained respectively. The demand heat attributes of the first entity and the second entity are weighted to obtain a second importance vector of the implicit influence of the industry on the resume text. By matching the words in the resume text with the entities in the dynamic knowledge graph and obtaining the demand heat attributes of the first entity and the second entity, the accuracy of the implicit feature analysis of the industry on the resume text can be improved, and in particular, the relevance and hot time effectiveness of the second importance vector can be enhanced.

[0062] In an optional embodiment, the extracting the post key words from the post description text and determining a first importance vector of the explicit influence of the post key words on the resume words includes: extracting the post key words from the post description text according to the demand field specified by the post demand party to obtain a demand level associated with the post key words; and matching the post key words with the resume words according to the demand level to obtain the first importance vector of the explicit influence of the post key words on the resume words.

[0063] In the embodiments of the present application, the demand field specified by the post demander can be obtained, and the demand level set for each demand field. For example, the post demander can set the fields of specialty, skill, experience, and achievement in a specific field as the highest priority, the fields of specialty, skill, experience, and achievement in a general field as the medium priority, and the fields of non-specialty, non-skill, non-experience, and non-achievement as the low priority. For example, for computer specialty, the words of software, JAVA, test, and algorithm belong to the high priority in the computer or programming field, the words of four levels, six levels, master, and XX University belong to the medium priority, and the word of hard work belongs to the low priority. For example, the post key words are extracted from the post description text according to the specified demand field, and the post key words and the resume words are matched according to the demand level of the post key words. For example, the correlation between the vector representation of the post key words and the vector of the resume words is calculated, and the correlation is corrected by using the weight associated with the corresponding demand level to obtain the first importance vector of the explicit influence of the post key words on the resume words. Moreover, the frequency of the post key words in the post description text can be counted, and the correlation is corrected by combining the frequency, the demand level, and the weight associated with the demand level. By extracting the post key words according to the demand field specified by the post demander and matching the post key words and the resume words according to the corresponding demand level, the explicit influence of the post key words on the resume words can be accurately quantified, the accuracy of the first importance vector is improved, and the accuracy and efficiency of the matching between the resume and the post are improved.

[0064] Embodiment three

[0065] Figure 3 Fig. 1 is a structural schematic diagram of a resume and post matching device based on a double-tower model according to an embodiment of the present application, which can be applied to the case of matching the post demand of a recruiter with the resume of a job seeker. The resume and post matching device based on the double-tower model can be realized in the form of hardware and / or software, and can be configured in a computer device, such as a client or a server. As shown in Fig. 1, the device includes: Figure 3

[0066] A word vector module 310 is configured to extract resume words and post words from a resume text and a post description text respectively, and determine corresponding resume word vectors and post word vectors.

[0067] A first importance module 320 is configured to extract post key words from the post description text, determine a first importance vector of the explicit influence of the post key words on the resume words, and splice the resume word vectors and the first importance vector to obtain a resume fusion vector.

[0068] ​a second importance module 330, configured to determine a second importance vector of an industry on the implicit influence of the resume text on the industry based on a dynamic knowledge graph of the industry;

[0069] an intermediate vector module 340, configured to input the resume fusion vector into a resume tower to obtain a resume intermediate vector, and input a post word vector into a post tower to obtain a post intermediate vector;

[0070] a matching module 350, configured to determine a matching result between the resume text and the post description text according to the resume intermediate vector, the post intermediate vector and the second importance vector.

[0071] In an optional implementation, the matching module 350 includes:

[0072] a semantic association unit, configured to input the resume intermediate vector and the post intermediate vector into a cross-attention layer to obtain a semantic association degree between the resume text and the post description text;

[0073] a semantic adjustment unit, configured to dynamically adjust the semantic association degree by taking the second importance vector as an adjustment factor based on a gating mechanism, to obtain a fusion association degree between the resume text and the post description text;

[0074] a matching unit, configured to determine a matching result between the resume text and the post description text according to the resume intermediate vector, the post intermediate vector and the fusion association degree.

[0075] In an optional implementation, the matching unit is specifically configured to:

[0076] concatenate the resume intermediate vector and the fusion association degree, and input the same into a next layer of the resume tower to obtain a resume output vector;

[0077] concatenate the post intermediate vector and the fusion association degree, and input the same into a next layer of the post tower to obtain a post output vector;

[0078] calculate a similarity between the resume output vector and the post output vector to obtain a matching result between the resume text and the post description text.

[0079] In an optional implementation, the second importance module 330 includes:

[0080] a knowledge graph matching unit, configured to match a resume word in the resume text with an entity in the dynamic knowledge graph to obtain a first entity that is successfully matched, and obtain a second entity directly dependent on the first entity from the dynamic knowledge graph;

[0081] A second importance unit is configured to determine a second importance vector of an industry implicit influence on the resume text according to a demand heat attribute of the first entity and the second entity.

[0082] In an optional implementation, the first importance module 320 includes:

[0083] A keyword extraction unit is configured to extract a job keyword from the job description text according to a demand field specified by a job demand party, to obtain a demand level associated with the job keyword.

[0084] A first importance unit is configured to match the job keyword with the resume word according to the demand level, to obtain a first importance vector of an explicit influence of the job keyword on the resume word.

[0085] The resume and job matching device based on the double-tower model provided in the embodiments of the present application can execute the resume and job matching method based on the double-tower model provided in any embodiment of the present application, and has the corresponding function modules and beneficial effects of executing each resume and job matching method based on the double-tower model.

[0086] According to the embodiments of the present application, the present application further provides an electronic device, a readable storage medium and a computer program product.

[0087] Embodiment Four

[0088] Figure 4 FIG. 4 is a structural schematic diagram of an electronic device 410 for implementing the resume and job matching method based on the double-tower model according to the embodiments of the present application. The electronic device is intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular telephones, smart phones, wearable devices (such as helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be examples only, and are not intended to limit the implementations of the present application described and / or claimed in this document.

[0089] As Figure 4As shown, the electronic device 410 includes at least one processor 411, and a memory, such as a read-only memory (ROM) 412, a random access memory (RAM) 413, and the like, connected to the at least one processor 411 in communication. The memory stores computer programs executable by the at least one processor 411, and the processor 411 can perform various appropriate actions and processes according to the computer programs stored in the read-only memory (ROM) 412 or loaded into the random access memory (RAM) 413 from the storage unit 418. In the RAM 413, various programs and data required for the operation of the electronic device 410 can also be stored. The processor 411, the ROM 412, and the RAM 413 are connected to each other through a bus 414. An input / output (I / O) interface 415 is also connected to the bus 414.

[0090] Various components in the electronic device 410 are connected to the I / O interface 415, including an input unit 416, such as a keyboard, a mouse, and the like, an output unit 417, such as various types of displays, a speaker, and the like, a storage unit 418, such as a magnetic disk, an optical disk, and the like, and a communication unit 419, such as a network card, a modem, a wireless communication transceiver, and the like. The communication unit 419 allows the electronic device 410 to exchange information / data with other devices through a computer network, such as the Internet, and / or various telecommunication networks.

[0091] The processor 411 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the processor 411 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various specialized artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, and the like. The processor 411 performs various methods and processes described above, such as the resume and job matching method based on the double-tower model.

[0092] In some embodiments, the resume and job matching method based on the double-tower model can be implemented as a computer program tangibly embodied in a computer-readable storage medium, such as the storage unit 418. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 410 via the ROM 412 and / or the communication unit 419. When the computer program is loaded into the RAM 413 and executed by the processor 411, one or more steps of the resume and job matching method based on the double-tower model described above can be performed. Alternatively, in other embodiments, the processor 411 can be configured as the resume and job matching method based on the double-tower model by any other appropriate means, such as by means of firmware.

[0093] The various implementations of the systems and techniques described above can be realized in digital electronic circuitry, integrated circuitry, specially designed application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs) computer hardware, firmware, software, and / or combinations thereof. These various implementations can include implementation in one or more computer programs that are executable and / or interpretable on a programmable system including at least one programmable processor, which can be special or general purpose, coupled to receive data and instructions from, and to transmit data and instructions to, a storage system, at least one input device, and at least one output device.

[0094] Computer programs implementing methods of the present application can be written in any combination of one or more programming languages. These computer programs can be provided to a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the computer program running on the

[0095] In the context of the present application, a computer-readable storage medium can be a tangible medium that can contain or store computer programs for use by or in connection with an instruction execution system, apparatus, or device. Computer-readable storage media can include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium can be a machine-readable signal medium. More specific examples of the machine-readable storage medium will include one or more lines of electrical connections, portable computer disks, hard disk drives, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or Flash memory), optical fibers, portable compact disc read-only memories (CD-ROMs), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0096] To provide for interaction with a user, the systems and techniques described here can be implemented on an electronic device having a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the electronic device. Other kinds of devices can be used to provide for interaction with a user as well; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form, including acoustic, speech, or tactile input.

[0097] The systems and techniques described here can be implemented in a computing system that includes a back end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front end component (e.g., a user computer having a graphical user interface or a Web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back end, middleware, or front end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0098] The computing system can include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. A server can be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system, to solve the defects of large management difficulty and weak business scalability in traditional physical host and VPS service.

[0099] It should be understood that various forms of flow shown above can be re-ordered, added to, or deleted from without departing from the spirit and scope of the technology described herein. For example, the steps recited in the application can be executed in parallel, executed in sequence, or executed in a different order, as long as the desired results of the technology described herein are achieved. The present disclosure is not limited in this regard.

[0100] The specific embodiments described above are not intended to be limiting, and persons skilled in the art will appreciate that various modifications, combinations, sub-combinations and alternatives can be made to the specific embodiments without departing from the spirit and scope of the disclosure. Any further modifications, equivalents, and / or alternatives come within the scope of the present disclosure as described in the claims.

Claims

1. A resume and post matching method based on a double-tower model, characterized in that, The method comprises the following steps: extracting resume words and post words from the resume text and the post description text respectively, and determining corresponding resume word vectors and post word vectors; extracting post keywords from the post description text, determining a first importance vector of the explicit influence of the post keywords on the resume words, and splicing the resume word vector and the first importance vector to obtain a resume fusion vector; determining a second importance vector of the implicit influence of the industry on the resume text based on the dynamic knowledge graph of the industry; inputting the resume fusion vector into a resume tower to obtain a resume intermediate vector, and inputting the post word vector into a post tower to obtain a post intermediate vector; determining a matching result between the resume text and the post description text according to the resume intermediate vector, the post intermediate vector and the second importance vector; wherein, determining a matching result between the resume text and the post description text according to the resume intermediate vector, the post intermediate vector and the second importance vector comprises: inputting the resume intermediate vector and the post intermediate vector into a cross attention layer to obtain a semantic correlation degree between the resume text and the post description text; based on the gating mechanism, the second importance vector is used as an adjusting factor to dynamically adjust the semantic correlation degree to obtain a fusion correlation degree between the resume text and the post description text; determining a matching result between the resume text and the post description text according to the resume intermediate vector, the post intermediate vector and the fusion correlation degree.

2. The method of claim 1, wherein, determining a matching result between the resume text and the post description text according to the resume intermediate vector, the post intermediate vector and the fusion correlation degree comprises: splicing the resume intermediate vector and the fusion correlation degree, and inputting the next layer of the resume tower to obtain a resume output vector; splicing the post intermediate vector and the fusion correlation degree, and inputting the next layer of the post tower to obtain a post output vector; calculating the similarity between the resume output vector and the post output vector to obtain a matching result between the resume text and the post description text.

3. The method of claim 1, wherein, determining a second importance vector of the implicit influence of the industry on the resume text based on the dynamic knowledge graph of the industry comprises: matching the resume words in the resume text with the entities in the dynamic knowledge graph to obtain a first entity that matches successfully, and obtaining a second entity directly dependent on the first entity from the dynamic knowledge graph; determining a second importance vector of the implicit influence of the industry on the resume text according to the demand heat attributes of the first entity and the second entity.

4. The method of claim 1, wherein, extracting post keywords from the post description text and determining a first importance vector of the explicit influence of the post keywords on the resume words comprises: extracting post keywords from the post description text according to the demand field specified by the post demand party to obtain the demand level associated with the post keywords; matching the post keywords and the resume words according to the demand level to obtain a first importance vector of the explicit influence of the post keywords on the resume words.

5. A resume and job matching device based on a double tower model, characterized by, ​ a word vector module configured to extract resume words and job description words from a resume text and a job description text respectively, and determine corresponding resume word vectors and job description word vectors; a first importance module configured to extract job keywords from the job description text, determine a first importance vector of explicit influence of the job keywords on the resume words, and splice the resume word vectors and the first importance vector to obtain a resume fusion vector; a second importance module configured to determine a second importance vector of implicit influence of an industry on the resume text based on a dynamic knowledge graph of the industry; an intermediate vector module configured to input the resume fusion vector into a resume tower to obtain a resume intermediate vector, and input the job description word vector into a job tower to obtain a job intermediate vector; a matching module configured to determine a matching result between the resume text and the job description text according to the resume intermediate vector, the job intermediate vector, and the second importance vector; wherein the matching module comprises: a semantic association unit configured to input the resume intermediate vector and the job intermediate vector into a cross-attention layer to obtain a semantic association degree between the resume text and the job description text; a semantic adjustment unit configured to dynamically adjust the semantic association degree based on a gating mechanism, and take the second importance vector as an adjustment factor to obtain a fusion association degree between the resume text and the job description text; a matching unit configured to determine a matching result between the resume text and the job description text according to the resume intermediate vector, the job intermediate vector, and the fusion association degree.

6. An electronic device, comprising: comprise: one or more processors; a memory configured to store one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors implement the method of any one of claims 1-4.

7. A computer-readable storage medium having stored thereon a computer program, characterized in that, The program is executed by the processor to implement the method of any one of claims 1-4.

8. A computer program product comprising a computer program which, when executed by a processor, implements the method of any one of claims 1-4.

Images