Labeling method and system for constructing knowledge graph in recruitment field

Abstract

The present invention relates to the field of online recruitment, and specifically to a labeling method and system for constructing a knowledge graph in the field of recruitment. The method is to obtain a job category knowledge data source constructed by a knowledge graph in the field of recruitment, perform cluster analysis on the job category knowledge data source, use an entity recognition model to extract text features and recognize named entities on the preliminary classification results, output recognized entity results, and perform word segmentation on the preliminary classification results, match the word segmentation results with the knowledge of a knowledge tree, and output knowledge labels; perform similarity calculation and filtration on the entity results and the knowledge labels, and then merge them to obtain a labeling result, which can be sent to an application end and applied to a search recommendation scenario. In the application, the core content of the resume and the position can be identified through the job description and resume description, and the ability, resume, and adaptable position of the talent required for the job recruitment can be determined, so as to achieve accurate matching recommendation of the resume and the position.

Description

Technical Field

[0001] The present invention relates to the field of online recruitment, and in particular to a labeling method and system for constructing a knowledge graph in the field of recruitment. Background Art

[0002] In the context of digital reform, the strength of data application capabilities is also related to the competitiveness of enterprises in the capital market. Especially in the application of job recruitment. Compared with traditional offline recruitment, online recruitment has great advantages in convenience and information transparency.

[0003] Online recruitment websites or APP clients will accumulate a large amount of data resources in the recruitment field. How to use data resources to create more value for enterprises. The application of knowledge graphs is an indispensable part. The application of knowledge graphs is reflected in many aspects, such as: helping enterprises to accurately find the talents they need; helping enterprises to locate the positions they need; helping job seekers find their ideal job opportunities; and providing job seekers with career development roadmaps.

[0004] In the increasingly competitive recruitment industry, it is necessary to screen the complex recruitment information and resume information and generate effective label information, so as to quickly build a knowledge graph, provide clear guidance for the preparation of recruitment requirements and resume information, achieve a high degree of fit between job seekers and companies, help companies improve their core competitiveness, and also provide the necessary conditions for the diversified development of enterprises.

[0005] In the process of building a knowledge graph in the recruitment field, how to label the job category knowledge used in personal resumes or job descriptions and ensure the accuracy of the knowledge labels so that they can be used in search recommendation scenarios after being sent to the application end has become one of the technical problems that need to be urgently solved in the construction of knowledge graphs in the recruitment field. Summary of the invention

[0006] In view of this, the present invention provides a labeling method and system for constructing a knowledge graph in the recruitment field.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] In a first aspect, the present invention provides a labeling method for constructing a knowledge graph in the recruitment field, comprising the following steps:

[0009] Obtaining a job category knowledge data source constructed by a recruitment field knowledge graph, wherein the job category knowledge data source includes job content text data;

[0010] Using the classification model to perform cluster analysis on the inputted job category knowledge data source, and obtaining preliminary classification results of the text data in the corresponding field;

[0011] Using the entity recognition model to perform text feature extraction and named entity recognition on the preliminary classification results, output the recognized entity results, and perform word segmentation on the preliminary classification results, match the word segmentation results with the knowledge of the knowledge tree, and output the knowledge labels;

[0012] The similarity between the entity result and the knowledge label is calculated, entities with similarity lower than a preset threshold are filtered out, the filtered entity result and the knowledge label are combined and output to obtain a labeling result.

[0013] As a further solution of the present invention, obtaining a job category knowledge data source for constructing a recruitment field knowledge graph also includes real-time changes in job category knowledge, and the real-time changes in job category knowledge include online change information, offline change information, and custom sample information.

[0014] As a further solution of the present invention, the position content text data includes position text data of the position description, and the preliminary classification results obtained after cluster analysis using the classification model contain position content classification text information, and the position content classification text information includes resume text classification results, job responsibilities classification results, job requirements classification results, and position information classification results.

[0015] As a further solution of the present invention, the classification model is a language model pre-trained using BERT, which is used to divide the job content into classification modules including job responsibilities, job requirements, resume text, and welfare benefits. The classification model is also used to split the resume work experience in the job text data into personal responsibilities and project responsibilities.

[0016] As a further solution of the present invention, the entity recognition model is a recognition model trained based on a combination of a BERT model, a BiLSTM model and a CRF model. The first layer of the entity recognition model is a BERT model, which is used to obtain word vectors and extract text features of the job category knowledge data source; the second layer of the entity recognition model is a bidirectional LSTM layer, which is used to deeply learn context feature information and perform named entity recognition; the third layer of the entity recognition model is a CRF layer, which is used to process the output sequence of the LSTM layer, obtain the sequence for optimizing the objective function, and output the entity result.

[0017] As a further solution of the present invention, performing word segmentation on the preliminary classification result includes: performing NLP word segmentation on the preliminary classification result to obtain a word segmentation result, and matching the word segmentation result with the knowledge of the knowledge tree.

[0018] As a further solution of the present invention, before outputting the knowledge label, it also includes matching the word segmentation results with the knowledge of the knowledge tree, performing position filtering, and using filtering strategies and filtering screening rules to perform filtering processing to obtain the knowledge label corresponding to the preliminary classification result.

[0019] As a further solution of the present invention, the labeling method for constructing the recruitment field knowledge graph also includes verifying the knowledge labels in an iterative environment. After verifying the knowledge labels, the labeling result information is sent to the application end for application in the search recommendation scenario.

[0020] In a second aspect, in a solution provided by the present invention, a labeling system for constructing a knowledge graph in the recruitment field is provided, the system comprising a data source acquisition module, a data clustering analysis module, an entity recognition module, a knowledge matching module and a labeling processing module;

[0021] The data source acquisition module is used to obtain the job category knowledge data source for constructing the recruitment field knowledge graph; the data clustering analysis module is used to use the classification model to perform cluster analysis on the input job category knowledge data source to obtain the preliminary classification results of the text data in the corresponding field; the entity recognition module is used to use the entity recognition model to perform text feature extraction and named entity recognition on the preliminary classification results, and output the recognized entity results; the knowledge matching module is used to perform word segmentation on the preliminary classification results, match the word segmentation results with the knowledge of the knowledge tree, and output the knowledge labels; the labeling processing module is used to calculate the similarity between the entity results and the knowledge labels, filter out entities with similarities lower than a preset threshold, merge the filtered entity results with the knowledge labels and output them to obtain the labeling results.

[0022] In a third aspect, in another solution provided by the present invention, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of a labeling method for constructing a knowledge graph in the recruitment field when loading and executing the computer program.

[0023] In a fourth aspect, in another solution provided by the present invention, a storage medium is provided, which stores a computer program, and when the computer program is loaded and executed by a processor, the steps of the labeling method for constructing the recruitment field knowledge graph are implemented.

[0024] The technical solution provided by the present invention has the following beneficial effects:

[0025] The labeling method, system, device and medium for constructing a knowledge graph in the field of recruitment provided by the present invention obtain a job category knowledge data source constructed by a knowledge graph in the field of recruitment, perform cluster analysis on the job category knowledge data source, use an entity recognition model to extract text features and recognize named entities on the preliminary classification results, output recognized entity results, and perform word segmentation on the preliminary classification results, match the word segmentation results with the knowledge of a knowledge tree, and output knowledge labels; calculate similarity between the entity results and the knowledge labels, filter and merge them to obtain labeling results, which can be sent to an application end and applied to a search recommendation scenario. In the application, the core content of the resume and the position can be identified through the job description and resume description, and the capabilities, resumes, and suitable positions of the talents required for the job recruitment can be determined, so as to achieve accurate matching recommendations for resumes and positions.

[0026] These and other aspects of the present invention will be more clearly understood in the following description of the embodiments. It should be understood that the above general description and the following detailed description are only exemplary and explanatory and cannot limit the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments or the prior art description. Obviously, the drawings in the following description are only some embodiments of the present invention. In the drawings:

[0028] Figure 1 A flowchart of a labeling method for constructing a knowledge graph in the recruitment field according to an embodiment of the present invention.

[0029] Figure 2 This is a linking flowchart in a labeling method for constructing a knowledge graph in the recruitment field according to an embodiment of the present invention.

[0030] Figure 3 This is a structural diagram of an entity recognition model in a labeling method for constructing a recruitment field knowledge graph according to an embodiment of the present invention.

[0031] Figure 4 A flowchart of entity recognition in a labeling method for constructing a knowledge graph in the recruitment field according to an embodiment of the present invention.

[0032] Figure 5 A flowchart of knowledge labeling in a labeling method for constructing a knowledge graph in the recruitment field according to an embodiment of the present invention.

[0033] Figure 6 A system block diagram of a labeling system constructed by a recruitment field knowledge graph according to an embodiment of the present invention.

[0034] Figure 7A structural block diagram of a computer device in a labeling method for constructing a recruitment field knowledge graph according to an embodiment of the present invention. DETAILED DESCRIPTION

[0035] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0036] In some of the processes described in the specification and claims of the present invention and the above-mentioned figures, multiple operations that appear in a specific order are included, but it should be clearly understood that these operations may not be executed in the order in which they appear in this article or executed in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this article are used to distinguish different messages, devices, modules, etc., do not represent the order of precedence, and do not limit the "first" and "second" to be different types.

[0037] The following will be combined with the drawings in the exemplary embodiments of the present invention to clearly and completely describe the technical solutions in the exemplary embodiments of the present invention. Obviously, the exemplary embodiments described are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0038] Since it is impossible to label the job category knowledge used in personal resumes or job descriptions during the process of building a knowledge graph in the recruitment field, and it is also impossible to ensure the accuracy of the knowledge labels, it is impossible to make accurate matching recommendations for resumes and positions in the search and recommendation scenarios on the application side.

[0039] In response to the above problems, the labeling method, system, device and medium for constructing a knowledge graph in the recruitment field provided by the present invention can obtain labeling results through cluster analysis, named entity recognition, knowledge matching and similarity calculation of the job category knowledge data source, which can be sent to the application end and applied to the search recommendation scenario.

[0040] Specifically, the embodiments of the present application are further described below in conjunction with the accompanying drawings.

[0041] See also Figure 1 As shown, an embodiment of the present invention provides a labeling method for constructing a knowledge graph in the recruitment field, including the following steps:

[0042] S10: Obtain a job category knowledge data source constructed by a recruitment field knowledge graph, wherein the job category knowledge data source includes position content text data.

[0043] In this embodiment, the position content text data includes position text data of the position description, and the preliminary classification results obtained after cluster analysis using the classification model contain position content classification text information, and the position content classification text information includes resume text classification results, job responsibilities classification results, job requirements classification results, and position information classification results.

[0044] See also Figure 2 As shown, Figure 2 The linking flow chart of the labeling method for constructing the recruitment domain knowledge graph of the present invention. In this embodiment, when obtaining the job category knowledge data source for constructing the recruitment domain knowledge graph, it also includes real-time changes in job category knowledge, and the real-time changes in job category knowledge include online change information, offline change information, and custom sample information.

[0045] In this embodiment, the job category knowledge can be applied to a specific JD / CV (CV: curriculum vitae resume (book), personal resume; JD: job description work description). The JD / CV is graphed through the present invention. After obtaining the job category knowledge data source, automatic linking can be performed through business dimension tables and rule checks to split the job category knowledge data source into position detail data and resume detail data. Then, based on expert metadata, word segmentation services, models and rules, the position detail data and resume detail data are labeled to obtain the iterative linking result Hbase library (rt) and the online linking result Hbase library (dp), and the message of the online linking result Hbase library (dp) is notified to Solr (Solr backend service) and FeatureServer (feature server).

[0046] Among them, when the JD / CV that needs to be marked is clearly identified, circle the JD / CV or other information that needs to be marked; because the data information changes in real time, monitor the data changes of JD / CV in this step to ensure that the marking result matches the data information.

[0047] The content to be labeled is sent to the automated linking service layer to identify the core content, call the word segmentation service, and match based on rules and models, and use the strategy model to output the data knowledge system.

[0048] Based on the above content, the knowledge label coverage / accuracy can be verified in an iterative environment. After verifying the knowledge, all result information is sent to the application end. It is currently mainly used in search recommendation scenarios.

[0049] S20: Perform cluster analysis on the input job category knowledge data source using the classification model to obtain preliminary classification results of the text data in the corresponding field.

[0050] Among them, the classification model is a language model pre-trained using BERT, which is used to divide the job content into classification modules including job responsibilities, job requirements, resume text, and welfare benefits. The classification model is also used to split the resume work experience in the job text data into personal responsibilities and project responsibilities.

[0051] In this embodiment, when identifying the core content, the service layer will first divide the marked data into modules. It mainly uses the BERT pre-trained language model + RNN model. The job content is divided into modules such as job responsibilities, job requirements, and company benefits; the resume work experience is divided into personal responsibilities and project responsibilities.

[0052] In this embodiment, when the word segmentation service is called, the core content to be recognized is input and the content is segmented. The word segmentation service is currently developed based on the log-linear model and combined with the recruitment application scenario.

[0053] S30: Using the entity recognition model to perform text feature extraction and named entity recognition on the preliminary classification results, and outputting the recognized entity results.

[0054] In this embodiment, when matching based on rules and models, the rules include longest substring matching, contextual response, matching the core expression of the text (such as removing stop words), etc. The model is mainly based on BERT, BiLSTM and CRF (such as Figure 3 Entity recognition is performed (as shown), the similarity between the entity recognition result and the rule output content is calculated, and the strategy model is used to output the data knowledge system.

[0055] In this embodiment, see Figure 4 and Figure 5 As shown, the entity recognition model is a recognition model trained based on the combination of the BERT model, the BiLSTM model and the CRF model. The first layer of the entity recognition model is the BERT model, which is used to obtain word vectors and extract text features of the job category knowledge data source; the second layer of the entity recognition model is a bidirectional LSTM layer, which is used to deeply learn context feature information and perform named entity recognition; the third layer of the entity recognition model is a CRF layer, which is used to process the output sequence of the LSTM layer, obtain the sequence for optimizing the objective function, and output the entity result.

[0056] S40: Segment the preliminary classification results, match the segmentation results with the knowledge in the knowledge tree, and output knowledge tags.

[0057] The segmentation of the preliminary classification results includes:

[0058] S31: Perform NLP word segmentation on the preliminary classification results to obtain word segmentation results, and match the word segmentation results with the knowledge of the knowledge tree.

[0059] S32: Before outputting the knowledge label, it also includes matching the word segmentation result with the knowledge of the knowledge tree, performing position filtering, and performing filtering processing using filtering strategies and filtering screening rules to obtain the knowledge label corresponding to the preliminary classification result.

[0060] S50: Calculate the similarity between the entity result and the knowledge label, filter out entities whose similarity is lower than a preset threshold, merge the filtered entity result and the knowledge label, and output them to obtain a labeling result.

[0061] In this embodiment, the labeling method for constructing the recruitment field knowledge graph also includes verifying the knowledge labels in an iterative environment. After verifying the knowledge labels, the labeling result information is sent to the application end for application in the search recommendation scenario.

[0062] When applied, by identifying the core content of the resume and position through the job description and resume description (work experience, project experience, etc.), you will know what kind of person is needed for the position, what abilities are required, and what the resume will be suitable for. The labeling method constructed by the recruitment field knowledge graph of the present invention can be applied to languages ​​including but not limited to Uyghur, Iranian, and English. The labeling method constructed by the recruitment field knowledge graph of the present invention can be applied to customer service directions including but not limited to travel customer service, logistics customer service, front desk customer service, etc. The labeling method constructed by the recruitment field knowledge graph of the present invention can also be applied to all-night shifts, morning and evening shifts, etc. Based on these labels, accurate matching recommendations can be made for resumes and positions.

[0063] It should be understood that, although described in a certain order, these steps are not necessarily performed in sequence in the above order. Unless there is clear explanation in this article, the execution of these steps does not have strict order restriction, and these steps can be performed in other orders. Moreover, a part of the steps of the present embodiment may include a plurality of steps or a plurality of stages, and these steps or stages are not necessarily performed at the same time, but can be performed at different times, and the execution order of these steps or stages is not necessarily performed in sequence, but can be performed in turn or alternately with at least a part of the steps or stages in other steps or other steps.

[0064] See also Figure 6As shown, an embodiment of the present invention provides a labeling system for constructing a knowledge graph in the recruitment field, the system comprising a data source acquisition module 100, a data clustering analysis module 200, an entity recognition module 300, a knowledge matching module 400 and a labeling processing module 500.

[0065] The data source acquisition module 100 is used to acquire the job category knowledge data source for constructing the recruitment field knowledge graph; the data clustering analysis module 200 is used to perform cluster analysis on the input job category knowledge data source using a classification model to obtain a preliminary classification result of the text data in the corresponding field; the entity recognition module 300 is used to perform text feature extraction and named entity recognition on the preliminary classification result using an entity recognition model, and output the recognized entity result; the knowledge matching module 400 is used to perform word segmentation on the preliminary classification result, match the word segmentation result with the knowledge of the knowledge tree, and output the knowledge label; the labeling processing module 500 is used to calculate the similarity between the entity result and the knowledge label, filter out entities with a similarity lower than a preset threshold, merge the filtered entity result with the knowledge label, and output it to obtain a labeling result.

[0066] In the embodiments of the present invention, see Figure 2 and Figure 4 As shown, the position content text data in the data source acquisition module 100 includes the position text data of the position description, and the preliminary classification results obtained after cluster analysis using the classification model include position content classification text information, and the position content classification text information includes resume text classification results, job responsibility classification results, job requirement classification results, and position information classification results.

[0067] The classification model in the data clustering analysis module 200 is a language model pre-trained using BERT, which is used to divide the job content into classification modules including job responsibilities, job requirements, resume text, and welfare benefits. The classification model is also used to split the resume work experience in the job text data into personal responsibilities and project responsibilities.

[0068] In some embodiments, the knowledge matching module 400 in the labeling system constructed by the recruitment field knowledge graph inputs the recognized core content and segments the content when calling the word segmentation service. The word segmentation service is currently developed based on the log-linear model combined with the recruitment application scenario.

[0069] Among them, the entity recognition model is a recognition model trained based on the combination of the BERT model, the BiLSTM model and the CRF model. The first layer of the entity recognition model is the BERT model, which is used to obtain word vectors and extract text features of the job category knowledge data source; the second layer of the entity recognition model is a bidirectional LSTM layer, which is used to deeply learn context feature information and perform named entity recognition; the third layer of the entity recognition model is a CRF layer, which is used to process the output sequence of the LSTM layer, obtain the sequence for optimizing the objective function, and output the entity result.

[0070] In some embodiments, the knowledge matching module 400 performs word segmentation on the preliminary classification result, including: performing NLP word segmentation on the preliminary classification result to obtain a word segmentation result, and matching the word segmentation result with the knowledge of the knowledge tree. Before outputting the knowledge label, it also includes matching the word segmentation result with the knowledge of the knowledge tree, performing position filtering, and filtering processing using filtering strategies and filtering screening rules to obtain a knowledge label corresponding to the preliminary classification result.

[0071] Therefore, when using the labeling system constructed by the recruitment field knowledge graph to label knowledge, the core content of the resume and the position can be identified through the job description, resume description (work experience, project experience, etc.), and the position can be known as to what kind of person is needed and what abilities are required, and what kind of resume can be suitable for what position. The labeling method constructed by the recruitment field knowledge graph of the present invention can realize accurate matching recommendation of resumes and positions.

[0072] In one embodiment, see Figure 7 As shown, in an embodiment of the present invention, a computer device 1000 is also provided, including at least one processor 1002, and a memory 1001 communicatively connected to the at least one processor 1002, wherein the memory 1001 stores instructions executable by the at least one processor 1002, and the instructions are executed by the at least one processor 1002 so that the at least one processor 1002 executes the labeling method for constructing the recruitment field knowledge graph, and the processor 1002 implements the steps in the above-mentioned method embodiments when executing the instructions:

[0073] Obtaining a job category knowledge data source constructed by a recruitment field knowledge graph, wherein the job category knowledge data source includes job content text data;

[0074] Using the classification model to perform cluster analysis on the inputted job category knowledge data source, and obtaining preliminary classification results of the text data in the corresponding field;

[0075] Using the entity recognition model to perform text feature extraction and named entity recognition on the preliminary classification results, output the recognized entity results, and perform word segmentation on the preliminary classification results, match the word segmentation results with the knowledge of the knowledge tree, and output the knowledge labels;

[0076] The similarity between the entity result and the knowledge label is calculated, entities with similarity lower than a preset threshold are filtered out, the filtered entity result and the knowledge label are combined and output to obtain a labeling result.

[0077] In an exemplary embodiment of the present invention, a computer-readable storage medium is also provided, on which a program product capable of implementing the above method of this specification is stored. In some possible embodiments, various aspects of the present invention can also be implemented in the form of a program product, which includes a program code. When the program product is run on a terminal device, the program code is used to enable the terminal device to execute the labeling method for constructing a recruitment field knowledge graph in various exemplary embodiments of the present invention described in the above "Exemplary Method" section of this specification. The labeling method for constructing a recruitment field knowledge graph includes:

[0078] Obtaining a job category knowledge data source constructed by a recruitment field knowledge graph, wherein the job category knowledge data source includes job content text data;

[0079] Using the classification model to perform cluster analysis on the inputted job category knowledge data source, and obtaining preliminary classification results of the text data in the corresponding field;

[0080] Using the entity recognition model to perform text feature extraction and named entity recognition on the preliminary classification results, output the recognized entity results, and perform word segmentation on the preliminary classification results, match the word segmentation results with the knowledge of the knowledge tree, and output the knowledge labels;

[0081] The similarity between the entity result and the knowledge label is calculated, entities with similarity lower than a preset threshold are filtered out, the filtered entity result and the knowledge label are combined and output to obtain a labeling result.

[0082] In an exemplary embodiment of the present invention, a program product for implementing the above method according to an embodiment of the present invention is described, which can adopt a portable compact disk read-only memory (CD-ROM) and include program code, and can be run on a terminal device, such as a personal computer. However, the program product of the present invention is not limited thereto, and in this document, a readable storage medium can be any tangible medium containing or storing a program, which can be used by or in combination with an instruction execution system, an apparatus or a device.

[0083] The program product may use any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0084] Computer readable signal media may include data signals propagated in baseband or as part of a carrier wave, in which readable program code is carried. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Readable signal media may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0085] The program code embodied on the readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination of the foregoing.

[0086] Program code for performing the operations of the present invention may be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving a remote computing device, the remote computing device may be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., through the Internet using an Internet service provider).

[0087] A person skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory.

[0088] In summary, the labeling method, system, device and medium for constructing a knowledge graph in the field of recruitment provided by the present invention obtains a job category knowledge data source constructed by a knowledge graph in the field of recruitment, performs cluster analysis on the job category knowledge data source, and then uses an entity recognition model to extract text features and recognize named entities on the preliminary classification results, outputs the recognized entity results, and performs word segmentation on the preliminary classification results, matches the word segmentation results with the knowledge of the knowledge tree, and outputs knowledge labels; calculates the similarity of the entity results and the knowledge labels, and then merges them after filtering to obtain a labeling result, which can be sent to an application end and applied to a search recommendation scenario. In the application, the core content of the resume and the position can be identified through the job description and resume description, and the capabilities, resumes, and adaptable positions of the talents required for the job recruitment can be determined, so as to achieve accurate matching recommendations for resumes and positions.

[0089] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A labeling method for constructing a knowledge graph in the recruitment field, characterized in that: The following steps are involved: S10: Obtaining a job category knowledge data source constructed by a recruitment field knowledge graph, wherein the job category knowledge data source includes job content text data; S20: using the classification model to perform cluster analysis on the inputted job category knowledge data source, and obtaining a preliminary classification result of the text data in the corresponding field; The position content text data includes the position text data of the position description. The preliminary classification results obtained after cluster analysis using the classification model include the position content classification text information, and the position content classification text information includes the resume text classification results, the job responsibility classification results, the job requirement classification results, and the position information classification results. First, the marked data is divided into modules. The classification model is a model that combines the BERT pre-trained language model + RNN model. The position content is divided into job responsibilities, job requirements, and company welfare and benefits modules; the resume work experience is split into personal responsibilities and project responsibilities. The classification model is also used to split the resume work experience in the position text data into personal responsibilities and project responsibilities. S30: Using the entity recognition model to perform text feature extraction and named entity recognition on the preliminary classification results, and output the recognized entity results; S40: Segment the preliminary classification results, match the segmentation results with the knowledge in the knowledge tree, and output knowledge labels; S50: Calculate the similarity between the entity result and the knowledge label, filter out entities whose similarity is lower than a preset threshold, merge the filtered entity result and the knowledge label and output them to obtain a labeling result.

2. The labeling method for constructing a recruitment field knowledge graph according to claim 1, characterized in that: In step S10, the job category knowledge data source constructed by the recruitment field knowledge graph is obtained, and the real-time changes of the job category knowledge are also included. The real-time changes of the job category knowledge include online change information, offline change information and custom sample information.

3. The labeling method for constructing a recruitment field knowledge graph according to claim 1, characterized in that: The entity recognition model is a recognition model trained based on a combination of a BERT model, a BiLSTM model, and a CRF model. The first layer of the entity recognition model is a BERT model, which is used to obtain word vectors and extract text features of job category knowledge data sources. The second layer of the entity recognition model is a bidirectional LSTM layer, which is used for deep learning of context feature information and named entity recognition; the third layer of the entity recognition model is a CRF layer, which is used to process the output sequence of the LSTM layer, obtain the sequence for optimizing the objective function, and output the entity result.

4. The labeling method for constructing a recruitment field knowledge graph according to claim 2, characterized in that: The word segmentation of the preliminary classification results in step S30 includes: S31: Perform NLP word segmentation on the preliminary classification results to obtain word segmentation results, and match the word segmentation results with the knowledge of the knowledge tree; S32: Before outputting the knowledge label, it also includes matching the word segmentation result with the knowledge of the knowledge tree, performing position filtering, and performing filtering processing using filtering strategies and filtering screening rules to obtain the knowledge label corresponding to the preliminary classification result.

5. The labeling method for constructing a recruitment field knowledge graph according to claim 2, characterized in that: The labeling method for constructing the recruitment field knowledge graph also includes verifying the knowledge labels in an iterative environment. After verifying the knowledge labels, the labeling result information is sent to the application end for application in the search recommendation scenario.

6. A labeling system for building a knowledge graph in the recruitment field, characterized in that: The labeling system for constructing the recruitment field knowledge graph is used to execute the labeling method for constructing the recruitment field knowledge graph according to any one of claims 1 to 5; the system comprises: The data source acquisition module is used to obtain the job category knowledge data source for the construction of the recruitment field knowledge graph; A data clustering analysis module is used to perform clustering analysis on the inputted job category knowledge data source using a classification model to obtain a preliminary classification result of the text data in the corresponding field; The position content text data includes the position text data of the position description. The preliminary classification results obtained after cluster analysis using the classification model include the position content classification text information, and the position content classification text information includes the resume text classification results, the job responsibility classification results, the job requirement classification results, and the position information classification results. First, the marked data is divided into modules. The classification model is a model that combines the BERT pre-trained language model + RNN model. The position content is divided into job responsibilities, job requirements, and company welfare and benefits modules; the resume work experience is split into personal responsibilities and project responsibilities. The classification model is also used to split the resume work experience in the position text data into personal responsibilities and project responsibilities. An entity recognition module is used to perform text feature extraction and named entity recognition on the preliminary classification results using an entity recognition model, and output recognized entity results; A knowledge matching module is used to segment the preliminary classification results, match the segmentation results with the knowledge of the knowledge tree, and output knowledge labels; The labeling processing module is used to calculate the similarity between the entity results and the knowledge labels, filter out entities whose similarity is lower than a preset threshold, merge the filtered entity results and the knowledge labels, and output them to obtain the labeling results.

7. A computer device, characterized in that: The computer device includes multiple computer devices, each of which includes a memory, a processor, and a computer program stored in the memory and executable on the processor. It is characterized in that the processors of the multiple computer devices jointly implement the steps of the labeling method for constructing a knowledge graph in the recruitment field as described in any one of claims 1 to 5 when executing the computer program.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: The computer program stored in the storage medium, when executed by a processor, implements the steps of the labeling method for constructing a recruitment field knowledge graph as described in any one of claims 1 to 5.

Images