Knowledge graph-based occupational planning auxiliary system and method

Abstract

The invention relates to the field of intelligent occupational planning, and provides an occupational planning auxiliary system and method based on a knowledge graph, and the method comprises the steps: firstly constructing an occupational knowledge graph, then extracting a first alternative occupational entity, and carrying out the semantic analysis of the first alternative occupational entity, so as to obtain a semantic feature of the first alternative occupational entity; meanwhile, obtaining personal information, work experience, skill specialty and job hunting intention input by the user to construct a user portrait, and performing semantic analysis on the user portrait to obtain semantic features of the user portrait; and matching the obtained semantic features of the first alternative occupational entity with the semantic features of the user portrait so as to judge whether the first alternative occupational entity is returned or not, namely whether occupational information related to the first alternative occupational entity is matched with the user or not. Therefore, suitable occupational information can be screened for the user from a more comprehensive and objective angle, and the problems that suggestions are not comprehensive and are greatly influenced by artificial factors in traditional occupational planning services can be effectively solved.

Description

Technical Field

[0001] The present application relates to the field of intelligent career planning, and more specifically, to a career planning assistance system and method based on knowledge graph. Background Art

[0002] In today's increasingly competitive and rapidly changing job market, traditional career planning services often expose many limitations. On the one hand, such services are usually based on static information. For example, they may only refer to some relatively fixed industry job description materials in the past years, or outdated career development case collections, etc., and the data sources they can rely on are very limited, limited to data collected from certain specific channels. On the other hand, traditional career planning services often rely on manual work to provide career advice. Different consultants have different personal experiences, growth backgrounds, and areas of expertise, which makes the advice they give greatly affected by their own experience or preferences. This may easily make the career advice given not fit the user's actual situation, and it is difficult to truly help users move forward smoothly on their career path.

[0003] Therefore, a career planning assistance solution based on knowledge graph is needed. Summary of the invention

[0004] In response to the shortcomings of the prior art, the present application provides a career planning assistance system and method based on a knowledge graph.

[0005] A career planning auxiliary method based on knowledge graph, comprising:

[0006] Obtain personal information, work experience, skills, and job-seeking intentions input by users;

[0007] Performing semantic analysis on the personal information, work experience, skills, and job-seeking intentions input by the user to obtain a user portrait semantic feature vector;

[0008] Constructing a professional knowledge graph, wherein nodes in the professional knowledge graph represent professional entities, and edges in the professional knowledge graph represent relationships between professional entities;

[0009] Extracting a first candidate occupation entity from the occupation knowledge graph;

[0010] Performing semantic analysis on the first candidate occupation entity to obtain a first candidate occupation entity semantic analysis vector;

[0011] Performing a dot multiplication of the user portrait semantic feature vector and the first candidate occupation entity semantic analysis vector to obtain a user portrait-occupation entity semantic matching feature vector;

[0012] According to the information in the user portrait-occupation entity semantic matching feature vector, it is determined whether the first candidate occupation entity needs to be recommended to the user.

[0013] A career planning auxiliary system based on knowledge graph, comprising:

[0014] The user input data collection module is used to obtain the user's personal information, work experience, skills, and job-seeking intentions;

[0015] A user input data analysis module is used to perform semantic analysis on the personal information, work experience, skills, and job-seeking intentions input by the user to obtain a user portrait semantic feature vector;

[0016] A professional knowledge graph construction module is used to construct a professional knowledge graph, wherein the nodes in the professional knowledge graph represent professional entities, and the edges in the professional knowledge graph represent the relationships between professional entities;

[0017] An alternative occupation entity extraction module, used to extract a first alternative occupation entity from the occupation knowledge graph;

[0018] A candidate occupation entity semantic analysis module, used for performing semantic analysis on the first candidate occupation entity to obtain a first candidate occupation entity semantic analysis vector;

[0019] A user portrait-occupation entity semantic matching module, used for performing a dot multiplication of the user portrait semantic feature vector and the first candidate occupation entity semantic analysis vector to obtain a user portrait-occupation entity semantic matching feature vector;

[0020] The recommendation result generation module is used to determine whether the first candidate occupation entity needs to be recommended to the user based on the information in the user portrait-occupation entity semantic matching feature vector.

[0021] This application has significant technical effects due to the adoption of the above technical solutions:

[0022] The career planning assistance system and method based on knowledge graph provided by the present application first constructs a career knowledge graph, then extracts the first candidate career entity and performs semantic analysis on it to obtain the semantic features of the first candidate career entity, and at the same time obtains the personal information, work experience, skills, and job-seeking intentions input by the user to construct a user portrait, and performs semantic analysis on the user portrait to obtain the user portrait semantic features, and then matches the obtained semantic features of the first candidate career entity with the semantic features of the user portrait to determine whether to return the first candidate career entity, that is, whether the career information related to the first candidate career entity matches the user. In this way, suitable career information can be screened for users from a more comprehensive and objective perspective, which can effectively solve the problem that the suggestions in traditional career planning services are not comprehensive and are greatly affected by human factors. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0024] Figure 1 It is a flowchart of a career planning assistance method based on knowledge graph according to an embodiment of the present application.

[0025] Figure 2 This is a data flow diagram of a career planning assistance method based on a knowledge graph according to an embodiment of the present application.

[0026] Figure 3 This is a flowchart of step S120 in the career planning assistance method based on knowledge graph according to an embodiment of the present application.

[0027] Figure 4 This is a flowchart of step S170 in the career planning assistance method based on knowledge graph according to an embodiment of the present application.

[0028] Figure 5 It is a system block diagram of a career planning assistance system based on a knowledge graph according to an embodiment of the present application. DETAILED DESCRIPTION

[0029] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0030] Based on the problems in the above-mentioned background technology, this application provides a career planning assistance method based on knowledge graph. Figure 1 It is a flowchart of a career planning assistance method based on knowledge graph according to an embodiment of the present application. Figure 2 Schematic diagram of data flow of a career planning assistance method based on knowledge graph according to an embodiment of the present application. Figure 1 and Figure 2 As shown, according to the embodiment of the present application, the career planning assistance method based on the knowledge graph includes: S110, obtaining personal information, work experience, skills, and job-seeking intentions input by the user; S120, performing semantic analysis on the personal information, work experience, skills, and job-seeking intentions input by the user to obtain a user portrait semantic feature vector; S130, constructing a career knowledge graph, wherein the nodes in the career knowledge graph represent career entities, and the edges in the career knowledge graph represent the relationships between career entities; S140, extracting a first candidate career entity from the career knowledge graph; S150, performing semantic analysis on the first candidate career entity to obtain a first candidate career entity semantic analysis vector; S160, performing dot multiplication of the user portrait semantic feature vector and the first candidate career entity semantic analysis vector to obtain a user portrait-career entity semantic matching feature vector; S170, determining whether the first candidate career entity needs to be recommended to the user based on the information in the user portrait-career entity semantic matching feature vector.

[0031] In step S110, the personal information, work experience, skills, and job-seeking intentions input by the user are obtained. It should be understood that the personal information input by the user includes information such as age, gender, and educational background. These seemingly basic information plays an important role in career planning. For example, different age groups may be suitable for different career development stages. Young people may prefer positions that accumulate experience and learn and grow quickly, while people approaching retirement age may pay more attention to job stability and ease; in terms of gender, although gender equality in employment is currently advocated, some industries or positions still have certain gender preferences in practice, such as the proportion of men in front-line construction positions in the construction engineering field, while the proportion of women in the kindergarten teacher industry is high. The level of education determines the entry threshold of the profession to a certain extent. For example, scientific research positions usually require a higher degree of education; the adaptability of the major studied is more directly related to the profession, and computer majors have a natural advantage in Internet technology-related positions. The work experience entered by the user can clarify the specific job title and job description and working hours that have been worked. If the information entered by the user shows that the user frequently changes jobs and the working hours are short, it implies that the user's career planning is unclear or it is difficult to adapt to the work environment, and it is necessary to help sort out the career goals in a targeted manner. The skill characteristics input by the user directly determine which specific professional positions the user can be competent for. When planning a career, it is necessary to match the skill level with the corresponding technical difficulty and requirements of the position, so that the skills can be fully utilized and it is also conducive to the user to continue to improve their skills in the position. For example, for programmers, mastering programming languages ​​(Python, Java, etc.) and development frameworks (Spring, Django, etc.) are important professional skills; for designers, proficiency in using design software (Adobe series software, etc.) is a key skill. In addition to these professional skills, soft skills such as communication skills, teamwork skills, leadership, and time management skills are also important considerations. People with excellent communication skills can be competent for positions that require frequent external communication and coordination, such as business development specialists; users with strong leadership are more suitable for planning development in management positions, leading the team to complete projects, etc. Rich soft skills can broaden the dimensions of career choices and make career planning more diversified. The job search intentions input by users specifically include the type of position expected by the user, the expected salary range, and career development goals.In detail, for example, whether the user wants to engage in technical research and development, marketing, administrative management or other types of positions, this clarifies the general direction of career planning and is an important basis for subsequent screening and recommendation of specific occupations. It directly narrows the scope of recommendation and focuses on relevant occupations that meet the user's expected job type. The salary expectation range reflects the user's expectation of the value of their own labor. The salary levels corresponding to different industries, regions and job levels vary greatly. Understanding salary expectations can screen out occupations that meet the user's economic needs and avoid recommending some positions that are suitable in other aspects but too low to meet the user's requirements. It ensures that the recommended occupations can be accepted by the user in terms of economic returns. At the same time, career development goals affect the design of the entire career planning path. It is necessary to recommend occupations that can help users gradually achieve these career development goals and promote the user's career growth at a reasonable pace. In general, these input data outline the user's overall picture from different dimensions. Only by fully mastering these detailed information can the system accurately locate the user's position in the career market, know the user's strengths, weaknesses, expectations and constraints, and then provide a solid foundation for subsequent recommendation of suitable occupations and planning reasonable career development paths, avoiding blindly recommending occupational information that does not meet the user's actual situation.

[0032] In step S120, semantic analysis is performed on the personal information, work experience, skills, and job-seeking intentions input by the user to obtain a user portrait semantic feature vector. Specifically, Figure 3 FIG. 1 is a flowchart of step S120 in the career planning auxiliary method based on knowledge graph according to an embodiment of the present application. Figure 3 As shown, the step S120 includes: S121, organizing and preprocessing the personal information, work experience, skills, and job-seeking intentions input by the user to obtain user portrait data; S122, segmenting the user portrait data and passing it through a user portrait data processor to obtain the user portrait semantic feature vector.

[0033] In step S121, the personal information, work experience, skills, and job-seeking intentions input by the user are sorted and preprocessed to obtain user portrait data. It should be understood that the personal information, work experience, skills, and job-seeking intentions input by the user often have various formats. For example, when describing work experience, some people may describe it in detail in a chronological order, job responsibilities, etc., while others may simply list the names of the units where they have worked; in terms of skills, different users may use different ways of expressing the skills they have mastered, some using full names, some using abbreviations, etc. Through sorting and preprocessing, these data in different formats can be standardized and standardized to meet the standard format required by the subsequent processing flow, so that the computer system can accurately identify and process this information. Moreover, in the content input by the user, there may be some irrelevant information or information that is not really helpful for career planning, such as mentioning some overly trivial hobbies and interests that are not very relevant to the profession when describing personal information (such as liking to collect a certain type of niche stamps, etc.), or adding some non-core information such as workplace environment descriptions to the work experience. Sorting and preprocessing can filter and eliminate this noise information, extract data that is truly valuable for career planning, and prevent irrelevant information from interfering with subsequent analysis and recommendation processes.

[0034] In step S122, the user portrait data is segmented and then passed through a user portrait data processor to obtain the user portrait semantic feature vector. In particular, in the present application, the user portrait data processor specifically refers to a converter-based semantic analyzer. It should be understood that there may be long-distance semantic associations between different parts of the user portrait data, such as the major studied in personal information and the expected position type in the job-seeking intention, which may be separated by many sentences but are actually closely related. The converter-based semantic analyzer can easily capture this long-distance dependency with its multi-head attention mechanism. It can assign different attention weights to words in different positions. No matter how far apart these words are in the text, as long as they are semantically related, they can be effectively paid attention to, unlike traditional models, where information transmission is prone to attenuation and loss as the distance increases, which helps to fully and completely understand the complex semantic information contained in the user portrait data. Moreover, the length of the user portrait data of different users after processing and sorting is often inconsistent. Some users may elaborate on their rich work experience and skills in detail, with long text content, while some users are relatively concise. The converter-based semantic analyzer does not need to perform complex length adjustments (such as padding or truncation operations) like some traditional models when processing user portrait data after word segmentation of different lengths. It can flexibly adapt to various input lengths, directly process the input text, and effectively extract the semantic features therein, ensuring that high-quality semantic features can be extracted in a unified manner regardless of the length of the user portrait data, improving the overall robustness and versatility of the system. Based on the rich semantic information captured, especially about the changes and needs in the user's career development process, a development path that is more in line with their personal characteristics and career expectations can be planned for the user.

[0035] In step S130, a professional knowledge graph is constructed, wherein the nodes in the professional knowledge graph represent professional entities, and the edges in the professional knowledge graph represent the relationships between professional entities. It should be understood that the professional entity first covers various specific professional names, such as professional titles in different fields and positions such as software engineers, marketing specialists, human resources managers, doctors, and teachers. It clarifies various specific jobs in the social division of labor, which have their own unique job responsibilities, working environments, and career development paths. It is one of the core elements of the entire professional knowledge graph construction. In addition, it also includes information related to skills, educational backgrounds, and industry dynamics required for different professions. The relationship between professional entities includes skill dependency, career promotion, coordination and cooperation, and skill transfer. Considering that traditional career planning often relies on fragmented and local information, the construction of a professional knowledge graph can integrate information on occupations, skills, educational backgrounds, and industry dynamics to form an organic whole. This can provide users with references from multiple dimensions and avoid unreasonable planning caused by considering only a single factor. In an embodiment of the present application, a possible implementation method for constructing a professional knowledge graph may be as follows: To construct a knowledge graph, the data source must first be determined, and data must be collected from multiple channels such as authoritative industry reports, job information posted on recruitment websites, professional books, academic papers, and job descriptions within the company. Then, natural language processing technology and data mining tools are used to extract and clean the collected raw data, extract relevant entity information about occupations, skills, educational backgrounds, industry trends, etc., and remove duplicate, erroneous, and incomplete data to ensure quality. Then, different occupational entities and the relationships between them are accurately identified through machine learning algorithms, manual annotation, etc., such as combining machine learning with predefined occupational classification dictionaries. The machine learning classification model is used to judge the occupational reference in the text, and the relationship type between occupational entities is determined according to the text context and industry knowledge. Then, tools such as graph databases are used to take the identified occupational entities as nodes and the labeled relationships as edges to construct an occupational knowledge graph and present it with the help of visualization tools for easy viewing and operation. Finally, new data should be collected regularly, and the previous data extraction, cleaning, entity recognition and relationship labeling processes should be repeated. New occupations, changing skill requirements, updated industry trends and other information should be integrated into the knowledge graph, and unreasonable or inaccurate areas should be adjusted and optimized according to application feedback to ensure that it can accurately reflect the actual situation in the occupational field to serve the auxiliary work of career planning.

[0036] In step S140, the first candidate occupation entity is extracted from the occupation knowledge graph. Accordingly, considering that the occupation knowledge graph information is complex, it is unrealistic to directly conduct in-depth analysis of all occupation entities at the same time in terms of computing resources and time costs. By extracting the first candidate occupation entity, only the subsequent semantic analysis and other processing are performed on this part of the screened occupations, which reduces unnecessary data processing, allows the system to quickly focus on key occupation options, speeds up the information processing speed of the entire career planning assistance, and can provide users with corresponding career planning suggestions in a more timely manner.

[0037] In step S150, the first candidate occupation entity is semantically analyzed to obtain the first candidate occupation entity semantic analysis vector. Specifically, in the embodiment of the present application, the step S150 includes: after the first candidate occupation entity is segmented, it is passed through the occupation entity semantic analyzer to obtain the first candidate occupation entity semantic analysis vector. It should be understood that the first candidate occupation entity itself is various occupation-related information described in the form of natural language, such as occupation names such as "software engineer" and "marketing specialist" and the skills, industry dynamics and other information associated therewith. It is difficult for the computer system to directly perform in-depth understanding and effective computational analysis on these textual information. In order to convert these natural languages ​​into data forms that can be recognized and processed by computers, the first candidate occupation entity can be semantically analyzed. Semantic analysis can integrate these different aspects of information scattered in the occupation entity and present them in a unified vector representation. In this way, in the subsequent analysis, a certain attribute of the occupation will not be viewed in isolation, but the overall semantics formed by comprehensively considering various factors will help to more comprehensively and accurately measure the adaptability between the occupation and the user, and avoid recommending unreasonable occupations due to focusing only on a single dimension. Specifically, the method of semantic analysis is to segment the first candidate occupation entity into words and input them into the occupation entity semantic analyzer for processing. In particular, the occupation entity semantic analyzer described in this application specifically refers to a converter-based semantic analyzer. The converter-based semantic analyzer uses a multi-head attention mechanism to capture the semantic concepts represented by different words in the first candidate occupation entity and the relationship between them. For example, for the occupation entity "data analyst", it can identify the core semantic concept of "data analysis", and at the same time capture the associated skill concepts such as "data collection" and "data visualization" and the inherent connection between educational background concepts such as "statistical background", sort out a complete semantic network around the occupation, and clearly show how the various elements of the occupation are interdependent and related.

[0038] In step S160, the user portrait semantic feature vector and the first alternative occupation entity semantic analysis vector are point-multiplied to obtain a user portrait-occupation entity semantic matching feature vector. It should be understood that the user portrait semantic feature vector reflects the comprehensive semantic features of the user in terms of personal information, work experience, skills, job-seeking intentions, etc., while the first alternative occupation entity semantic analysis vector reflects the multi-dimensional semantic information related to each alternative occupation. Fusion of these two vectors in the same semantic space can correlate the characteristics of the user with the characteristics of the occupation. For example, the user portrait indicates that the user has strong data analysis skills, and in a certain alternative occupation entity, the data analysis-related work content is also emphasized. Through fusion, the semantic correlation between these two aspects can be reflected, which is convenient for subsequent more intuitive and accurate analysis of the degree of matching between the two.

[0039] In step S170, based on the information in the user portrait-occupation entity semantic matching feature vector, it is determined whether the first candidate occupation entity needs to be recommended to the user. Specifically, Figure 4 The flowchart of step S170 in the career planning auxiliary method based on knowledge graph according to the embodiment of the present application. The step S170 includes: S171, performing an intrinsic connection pattern analysis based on the target parameter space node decomposition on the user portrait-occupation entity semantic matching feature vector to obtain an optimized user portrait-occupation entity semantic matching feature vector; S172, passing the optimized user portrait-occupation entity semantic matching feature vector through a recommendation classifier to obtain a classification result, and the classification result is used to indicate whether the first candidate occupation entity needs to be recommended to the user.

[0040] In step S171, the user portrait-occupation entity semantic matching feature vector is subjected to an internal connection pattern analysis based on the node decomposition of the target parameter space to obtain an optimized user portrait-occupation entity semantic matching feature vector. In particular, there is a semantic logical connection between the data points within the user portrait semantic feature vector and the first candidate occupation entity semantic encoding vector itself, and the matching feature vector formed after the dot product needs to be further optimized to highlight these internal connections. For example, the correlation between a certain skill specialty of a user and the skills required by the occupation entity may exist in a certain data relationship in the vector, but if it is not optimized, these implicit semantic connections may not be clear enough to fully provide a comprehensive and in-depth reference for the recommendation classifier, affecting the classifier's accurate judgment of the degree of matching between the two, and is not conducive to accurately recommending occupations. Due to inaccurate and unstable recommendation results, users may miss the career choice that is truly suitable for them, and cannot enter the career field that can give full play to their own advantages and meet their own development expectations, hindering the further improvement of personal professional ability and the good expansion of career growth paths. Based on this, in the technical solution of the present application, it is necessary to perform an intrinsic connection pattern analysis on the user portrait-occupation entity semantic matching feature vector based on the target parameter space node decomposition to obtain an optimized user portrait-occupation entity semantic matching feature vector.

[0041] Specifically, in an embodiment of the present application, an intrinsic connection pattern analysis is performed on the user portrait-occupation entity semantic matching feature vector based on the target parameter space node decomposition to obtain an optimized user portrait-occupation entity semantic matching feature vector, including: extracting the target occupation matching recommendation classification parameter matrix of the recommendation classifier; performing node-based decomposition on the target occupation matching recommendation classification parameter matrix in units of row vectors to obtain a set of target occupation matching recommendation classification parameter node encoding vectors; using each target occupation matching recommendation classification parameter node encoding vector in the set of target occupation matching recommendation classification parameter node encoding vectors as the wandering topological space, and performing topological space constraints on the user portrait-occupation entity semantic matching feature vectors to obtain a set of constrained user portrait-occupation entity semantic matching feature vectors; calculating the positional mean vector of the set of constrained user portrait-occupation entity semantic matching feature vectors to obtain the optimized user portrait-occupation entity semantic matching feature vector.

[0042] More specifically, in an embodiment of the present application, each target occupation matching recommendation classification parameter node encoding vector in the set of target occupation matching recommendation classification parameter node encoding vectors is used as a walking topological space, and the user portrait-occupation entity semantic matching feature vector is subjected to topological space constraints to obtain a constrained user portrait-occupation entity semantic matching feature vector set, including: multiplying the user portrait-occupation entity semantic matching feature vector and the transposed vector of the target occupation matching recommendation classification parameter node encoding vector, and calculating the natural exponential function value of the multiplication result to obtain the occupation matching recommendation classification weighted index response weight; calculating the user portrait-occupation entity semantic matching feature vector and the transposed vector of the target occupation matching recommendation classification parameter node encoding vector. The Euclidean distance between the user portrait-occupation entity semantic matching feature vector and the target occupation matching recommended classification parameter node coding vector is used to obtain the occupation matching recommended classification node coding distance value; the occupation matching recommended classification node coding distance value is multiplied by the user portrait-occupation entity semantic matching feature vector, and the natural exponential function value is calculated for each eigenvalue of the vector after the dot product to obtain the occupation matching recommended classification distance guidance index feature vector; the occupation matching recommended classification weighted index response weight is multiplied by the occupation matching recommended classification distance guidance index feature vector to obtain the constrained user portrait-occupation entity semantic matching feature vector.

[0043] In the embodiment of the present application, specifically, the user portrait-occupation entity semantic matching feature vector is subjected to an intrinsic connection pattern analysis based on the target parameter space node decomposition to obtain an optimized user portrait-occupation entity semantic matching feature vector, including: processing the user portrait-occupation entity semantic matching feature vector with the following optimization formula to obtain the optimized user portrait-occupation entity semantic matching feature vector; wherein the optimization formula is:

[0044] W=[w1,w2,...,w i ,...,w n ] T

[0045]

[0046]

[0047] Among them, W represents the target occupation matching recommendation classification parameter matrix, w1, w2, w i 、w n represents the first, second, i-th, and n-th target occupation matching recommendation classification parameter node encoding vectors of the set of target occupation matching recommendation classification parameter node encoding vectors, T represents the transpose of the vector, exp represents the natural exponential function, v1 represents the user portrait-occupation entity semantic matching feature vector, represents matrix multiplication, ⊙ represents point multiplication by position, D(v1,wi ) represents the calculation of vector v1 and vector w i The Euclidean distance between 1-i represents the i-th constrained user portrait-occupation entity semantic matching feature vector of the set of constrained user portrait-occupation entity semantic matching feature vectors, n represents the total number of sets of constrained user portrait-occupation entity semantic matching feature vectors, V i ' represents the optimized user portrait-occupation entity semantic matching feature vector.

[0048] That is, in response to the above technical problems, in the technical solution of the present application, the intrinsic connection pattern analysis of the user portrait-occupation entity semantic matching feature vector is performed based on the node decomposition of the target parameter space. The process first extracts the key parameters used for decision-making from the trained recommendation classifier. These parameters constitute a matrix form in a high-dimensional space, and each row represents the weight or influencing factor on a different dimension. The target occupation matching recommendation classification parameter matrix can provide insight into the location and shape of the model decision boundary, and then infer which input features are most critical to the prediction results.

[0049] Next, the target occupation matching recommendation classification parameter matrix is ​​node-decomposed in units of row vectors to obtain a set of node encoding vectors of target occupation matching recommendation classification parameters. Here, each row vector is a node in graph theory, which means that each set of parameters is now regarded as an entity with potential connectivity. This transformation allows the application of methods from graph theory and network science to explore the interactions between features. The node encoding vector not only carries information about the original parameters, but also implies knowledge about the topological structure of the entire system. Node decomposition further reveals the intrinsic connection pattern or structure of the data, so that the optimized feature vector can better adapt to new task requirements.

[0050] Then, each target occupation matching recommended classification parameter node encoding vector in the set of target occupation matching recommended classification parameter node encoding vectors is used as the walking topological space, and the user portrait-occupation entity semantic matching feature vector is respectively subjected to topological space constraints to obtain a set of constrained user portrait-occupation entity semantic matching feature vectors. "Walking" in the topological space defined by the node encoding vector is actually simulating an exploratory process, the purpose of which is to find those feature transformations that can best maintain the characteristics of the original data structure. Each step determines the position of the next step based on the probability distribution of the current state. The topological space constraints ensure that even in different contexts, the feature representation still retains certain invariances. At the same time, it can also promote cross-domain transfer learning because it emphasizes the universal relationship between features rather than the details of a specific field. In this way, the reconstruction of the feature space is achieved, making the optimized feature vector more compact and having better generalization ability.

[0051] Finally, the positional mean vector of the set of the constrained user portrait-occupation entity semantic matching feature vectors is calculated to obtain the optimized user portrait-occupation entity semantic matching feature vector. Calculating the mean vector is a statistical aggregation method used to integrate the optimal solution from multiple perspectives. The idea behind this step is to reduce the deviation caused by a single estimate by fusing the information provided by different sample points. The averaging process is equivalent to performing a soft voting, which enhances the expressiveness of common features and makes the optimized feature vector more stable and reliable.

[0052] In step S172, the optimized user portrait-occupation entity semantic matching feature vector is passed through a recommendation classifier to obtain a classification result, and the classification result is used to indicate whether the first candidate occupation entity needs to be recommended to the user. It should be understood that the recommendation classifier is essentially a machine learning model, which is trained based on a large amount of existing annotated data (usually the matching of users with different occupations in history and the corresponding recommendation feedback data, etc.), and learns the mapping relationship between the feature vector (here, the optimized user portrait-occupation entity semantic matching feature vector) and the recommendation result (whether the occupation is recommended to the user). The most direct role of the classification result is to clearly give a conclusion on whether the first candidate occupation entity needs to be recommended to the user, which makes the entire knowledge graph-based career planning auxiliary process have a clear output, and the user can intuitively know which occupations are suitable for themselves after comprehensive consideration by the system, and which ones do not meet the requirements and are not recommended, avoiding the confusion of users when facing many alternative occupations, providing users with a clear reference for career selection, and helping users focus on the career direction that is truly worthy of attention and consideration.

[0053] In another embodiment of the present application, considering that the development trends and job requirements of different industries are also constantly evolving. Therefore, in order to ensure that career planning suggestions can keep pace with the times and accurately reflect the latest market conditions, the present application also introduces a dynamic data update mechanism. This mechanism aims to continuously track, collect and analyze the latest industry reports, changes in labor market demand, and policy and regulatory updates from multiple authoritative channels, including industry statistics released by the government, research results of professional consulting agencies, job posting trends on recruitment websites, and discussion heat on career development on social media. The collected data will undergo a series of preprocessing steps, such as data cleaning, deduplication, formatting conversion, etc., to ensure its quality and consistency. Subsequently, these data are deeply mined through natural language processing technology and machine learning algorithms to extract valuable features and patterns, such as growth points of emerging industries, transfer directions of skill requirements, and employment hotspots in specific regions or fields. In addition, time series analysis will be used to predict development trends in the future to provide users with forward-looking career planning guidance. The dynamically acquired information is not only used for direct reference to users, but more importantly, it will be regularly fed back to the occupational knowledge map as an important basis for adjusting and enriching the map structure. When it is detected that a new professional role has appeared in a certain industry or the skills required for an existing profession have changed significantly, the corresponding nodes can be added to the graph in time or the relationship between the existing nodes can be modified, so that the professional knowledge graph is always kept up to date and better serves the user's career development needs. Based on the dynamically updated professional knowledge graph, the recommendation engine of this application can more keenly capture the subtle changes in the user's field or field of interest, thereby achieving more accurate and personalized career path planning. If it is found that there is a potential mismatch between the user's skill combination and the current market demand, early warning can be given and targeted learning suggestions can be given; if certain emerging professions show good development prospects, such opportunities can be pushed to users with relevant backgrounds at an appropriate time. In summary, by establishing a complete dynamic data update mechanism, this application can not only grasp the pulse of the entire employment market at the macro level, but also provide each user with practical and visionary career planning services at the micro level. This will undoubtedly greatly improve the quality and efficiency of traditional career planning services and add more possibilities to users' careers.

[0054] In summary, the knowledge graph-based career planning assistance method based on the embodiment of the present application is explained, which first constructs a career knowledge graph, then extracts the first candidate career entity and performs semantic analysis on it to obtain the semantic features of the first candidate career entity, and at the same time obtains the personal information, work experience, skills, and job-seeking intentions input by the user to construct a user portrait, and performs semantic analysis on the user portrait to obtain the user portrait semantic features, and then matches the obtained semantic features of the first candidate career entity with the user portrait semantic features to determine whether to return the first candidate career entity, that is, whether the career information related to the first candidate career entity matches the user. In this way, suitable career information can be screened for users from a more comprehensive and objective perspective, which can effectively solve the problem that the suggestions in traditional career planning services are not comprehensive and are greatly affected by human factors.

[0055] Figure 5 : is a system block diagram of a career planning assistance system based on a knowledge graph according to an embodiment of the present application. Figure 5 As shown, according to the embodiment of the present application, the career planning assistance system 100 based on the knowledge graph includes: a user input data acquisition module 110, which is used to obtain the personal information, work experience, skills, and job-seeking intentions input by the user; a user input data analysis module 120, which is used to perform semantic analysis on the personal information, work experience, skills, and job-seeking intentions input by the user to obtain a user portrait semantic feature vector; a career knowledge graph construction module 130, which is used to construct a career knowledge graph, wherein the nodes in the career knowledge graph represent career entities, and the edges in the career knowledge graph represent the relationship between career entities; an alternative career entity extraction module 140, used to extract the first candidate occupation entity from the occupation knowledge graph; the alternative occupation entity semantic analysis module 150, used to perform semantic analysis on the first candidate occupation entity to obtain a first candidate occupation entity semantic analysis vector; the user portrait-occupation entity semantic matching module 160, used to perform dot multiplication of the user portrait semantic feature vector and the first candidate occupation entity semantic analysis vector to obtain a user portrait-occupation entity semantic matching feature vector; the recommendation result generation module 170, used to decide whether the first candidate occupation entity needs to be recommended to the user based on the information in the user portrait-occupation entity semantic matching feature vector.

[0056] Here, those skilled in the art can understand that the specific functions and operations of each unit and module in the above-mentioned career planning assistance system 100 based on the knowledge graph have been referred to above. Figures 1 to 4 It has been introduced in detail in the description of the knowledge graph-based career planning auxiliary method, and therefore, its repeated description will be omitted.

[0057] In summary, the career planning assistance system 100 based on the knowledge graph according to the embodiment of the present application is explained, which first constructs a career knowledge graph, then extracts the first candidate career entity and performs semantic analysis on it to obtain the semantic features of the first candidate career entity, and at the same time obtains the personal information, work experience, skills, and job-seeking intentions input by the user to construct a user portrait, and performs semantic analysis on the user portrait to obtain the user portrait semantic features, and then matches the obtained semantic features of the first candidate career entity with the semantic features of the user portrait to determine whether to return the first candidate career entity, that is, whether the career information related to the first candidate career entity matches the user. In this way, suitable career information can be screened for users from a more comprehensive and objective perspective, which can effectively solve the problem that the suggestions in traditional career planning services are not comprehensive and are greatly affected by human factors.

[0058] The above is only a preferred embodiment of the present application and does not constitute any form of limitation to the present application. Although the present application has been disclosed as a preferred embodiment as above, it is not intended to limit the present application. Any technician familiar with the profession can make some changes or modifications to equivalent embodiments of the technical contents disclosed above without departing from the scope of the technical solution of the present application. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present application without departing from the content of the technical solution of the present application still falls within the scope of the technical solution of the present application.

Claims

1. A career planning auxiliary method based on knowledge graph, characterized in that: include: Obtain personal information, work experience, skills, and job-seeking intentions input by users; Performing semantic analysis on the personal information, work experience, skills, and job-seeking intentions input by the user to obtain a user portrait semantic feature vector; Constructing a professional knowledge graph, wherein nodes in the professional knowledge graph represent professional entities, and edges in the professional knowledge graph represent relationships between professional entities; Extracting a first candidate occupation entity from the occupation knowledge graph; Performing semantic analysis on the first candidate occupation entity to obtain a first candidate occupation entity semantic analysis vector; Performing a dot multiplication of the user portrait semantic feature vector and the first candidate occupation entity semantic analysis vector to obtain a user portrait-occupation entity semantic matching feature vector; According to the information in the user portrait-occupation entity semantic matching feature vector, it is determined whether the first candidate occupation entity needs to be recommended to the user.

2. The career planning auxiliary method based on knowledge graph according to claim 1 is characterized in that: The personal information, work experience, skills, and job-seeking intentions input by the user are semantically analyzed to obtain a user portrait semantic feature vector, including: Organizing and preprocessing the personal information, work experience, skills, and job-seeking intentions input by the user to obtain user portrait data; The user portrait data is segmented and passed through a user portrait data processor to obtain the user portrait semantic feature vector.

3. The career planning auxiliary method based on knowledge graph according to claim 2 is characterized in that: Performing semantic analysis on the first candidate occupation entity to obtain a first candidate occupation entity semantic analysis vector includes: segmenting the first candidate occupation entity and passing it through an occupation entity semantic analyzer to obtain the first candidate occupation entity semantic analysis vector.

4. The career planning auxiliary method based on knowledge graph according to claim 3 is characterized in that: The user portrait data processor and the occupational entity semantic analyzer are converter-based semantic analyzers.

5. The career planning auxiliary method based on knowledge graph according to claim 4 is characterized in that: Determining whether to recommend the first candidate occupation entity to the user according to information in the user portrait-occupation entity semantic matching feature vector includes: Performing an intrinsic connection pattern analysis on the user portrait-occupation entity semantic matching feature vector based on target parameter space node decomposition to obtain an optimized user portrait-occupation entity semantic matching feature vector; The optimized user portrait-occupation entity semantic matching feature vector is passed through a recommendation classifier to obtain a classification result, and the classification result is used to indicate whether the first candidate occupation entity needs to be recommended to the user.

6. The career planning auxiliary method based on knowledge graph according to claim 5 is characterized in that: The user portrait-occupation entity semantic matching feature vector is subjected to an intrinsic connection pattern analysis based on the target parameter space node decomposition to obtain an optimized user portrait-occupation entity semantic matching feature vector, including: Extracting a target occupation matching recommendation classification parameter matrix of the recommendation classifier; Decomposing the target occupation matching recommendation classification parameter matrix into nodes in units of row vectors to obtain a set of target occupation matching recommendation classification parameter node encoding vectors; Taking each target occupation matching recommendation classification parameter node encoding vector in the set of target occupation matching recommendation classification parameter node encoding vectors as the walking topological space, topological space constraints are respectively performed on the user portrait-occupation entity semantic matching feature vectors to obtain a set of constrained user portrait-occupation entity semantic matching feature vectors; The positional mean vector of the set of the constrained user portrait-occupation entity semantic matching feature vectors is calculated to obtain the optimized user portrait-occupation entity semantic matching feature vector.

7. The career planning auxiliary method based on knowledge graph according to claim 6 is characterized in that: Taking each target occupation matching recommendation classification parameter node encoding vector in the set of target occupation matching recommendation classification parameter node encoding vectors as the walking topological space, topological space constraints are respectively performed on the user portrait-occupation entity semantic matching feature vectors to obtain a set of constrained user portrait-occupation entity semantic matching feature vectors, including: After multiplying the user portrait-occupation entity semantic matching feature vector and the transposed vector of the target occupation matching recommendation classification parameter node encoding vector, the natural exponential function value of the multiplication result is calculated to obtain the occupation matching recommendation classification weighted index response weight; Calculate the Euclidean distance between the user portrait-occupation entity semantic matching feature vector and the target occupation matching recommendation classification parameter node encoding vector to obtain an occupation matching recommendation classification node encoding distance value; Perform a dot multiplication of the occupation matching recommendation classification node encoding distance value and the user portrait-occupation entity semantic matching feature vector, and calculate a natural exponential function value for each eigenvalue of the vector after the dot multiplication to obtain an occupation matching recommendation classification distance guidance index feature vector; The occupation matching recommendation classification weighted index response weight and the occupation matching recommendation classification distance guidance index feature vector are point-multiplied to obtain the constrained user portrait-occupation entity semantic matching feature vector.

8. A career planning assistance system based on knowledge graph, characterized in that: include: The user input data collection module is used to obtain the user's personal information, work experience, skills, and job-seeking intentions; A user input data analysis module is used to perform semantic analysis on the personal information, work experience, skills, and job-seeking intentions input by the user to obtain a user portrait semantic feature vector; A professional knowledge graph construction module is used to construct a professional knowledge graph, wherein the nodes in the professional knowledge graph represent professional entities, and the edges in the professional knowledge graph represent the relationships between professional entities; An alternative occupation entity extraction module, used to extract a first alternative occupation entity from the occupation knowledge graph; A candidate occupation entity semantic analysis module, used for performing semantic analysis on the first candidate occupation entity to obtain a first candidate occupation entity semantic analysis vector; A user portrait-occupation entity semantic matching module, used for performing a dot multiplication of the user portrait semantic feature vector and the first candidate occupation entity semantic analysis vector to obtain a user portrait-occupation entity semantic matching feature vector; The recommendation result generation module is used to determine whether the first candidate occupation entity needs to be recommended to the user based on the information in the user portrait-occupation entity semantic matching feature vector.

9. The career planning assistance system based on knowledge graph according to claim 8, characterized in that: The candidate occupation entity semantic analysis module is used to: segment the first candidate occupation entity and pass it through an occupation entity semantic analyzer to obtain the first candidate occupation entity semantic analysis vector.

10. The career planning assistance system based on knowledge graph according to claim 9 is characterized in that: The recommendation result generating module comprises: A user portrait-occupation entity semantic matching feature optimization unit, used for performing an intrinsic connection pattern analysis on the user portrait-occupation entity semantic matching feature vector based on a target parameter space node decomposition to obtain an optimized user portrait-occupation entity semantic matching feature vector; The user portrait-occupation entity semantic matching feature parsing unit is used to pass the optimized user portrait-occupation entity semantic matching feature vector through a recommendation classifier to obtain a classification result, and the classification result is used to indicate whether the first candidate occupation entity needs to be recommended to the user.

Images