Human resource matching system and storage medium

Abstract

Provided is a human resource matching system that achieves more appropriate matching between human resources and projects. The human resource matching system comprises a values extracting means, a characteristics extracting means, a calculating means, a correspondence means, and an interaction means. The values extracting means extracts values information from work history data of a participant. The characteristics extracting means extracts project information indicating characteristics of a project. The calculating means calculates a degree of fit between the values information and the project information. The correspondence means generates a matching result between the participant and the project on the basis of the degree of fit. On the basis of the matching result, the interaction means confirms a degree of interest of the participant via an interactive interface that employs a machine learning model. The correspondence means determines the matching result on the basis of weighting information between the skills of the participant and the values information.

Description

Talent matching system and storage medium

[0001] The present invention relates to a talent matching system and a storage medium.

[0002] There are technologies related to talent matching for projects.

[0003] For example, Patent Document 1 describes matching a project with a team composed of a plurality of people. The information processing apparatus of Patent Document 1 comprises a requirement information acquisition unit, a team characteristic information acquisition unit, and a matching unit. The requirement information acquisition unit acquires requirement information indicating requirements for executing a project. The team characteristic information acquisition unit acquires, for each team, team characteristic information indicating respective characteristics of one or more teams. The matching unit performs matching processing between the project and the team by referring to the requirement information and the team characteristic information.

[0004] Japanese Unexamined Patent Application Publication No. 2022-189593

[0005] Talent matching is often performed based on objective indicators such as candidates' skills, experience, and qualifications, and document information such as resumes and skill sheets for such skills is databased and managed. Further, a necessary skill set is defined based on the requirement definition of the project, and talents are selected through matching with the skill set. In such a process, the candidate's values and motivation are often individually assessed through qualitative evaluation methods such as interviews and aptitude tests.

[0006] However, relying solely on skills and experience cannot sufficiently evaluate the suitability of a candidate for a project, making it difficult to maintain the motivation of project participants and achieve long-term results. In addition, there is no means for quantitatively grasping an individual's potential interests and values, and it is impossible to provide opportunities for challenging new fields and support appropriate career development. Furthermore, mismatches between the requirements of a project and an individual's orientation can lead to problems such as mid-term withdrawal and decreased productivity.

[0007] A main object of the present invention is to provide a talent matching system and a storage medium that contribute to realizing more appropriate matching between talents and projects.

[0008] According to a first aspect of the present invention, a human resource matching system is provided which comprises: a value extraction means for extracting value information from a participant's work history data; a characteristic extraction means for extracting project information indicating the characteristics of a project; a calculation means for calculating the degree of fit between the value information and the project information; a response means for generating a matching result between the participant and the project based on the degree of fit; and a dialogue means for confirming the participant's level of interest through an interactive interface using a machine learning model based on the matching result, wherein the response means determines the matching result based on weighting information between the participant's skills and the value information.

[0009] A computer-readable storage medium is provided that stores a program which causes a computer to perform a values ​​extraction step of extracting values ​​information from a participant's work history data, a characteristics extraction step of extracting project information indicating the characteristics of a project, a calculation step of calculating the degree of fit between the values ​​information and the project information, a correspondence step of generating a matching result between the participant and the project based on the degree of fit, and a dialogue step of confirming the degree of interest of the participant through an interactive interface using a machine learning model based on the matching result, wherein the correspondence step determines the matching result based on weighting information between the participant's skills and the values ​​information.

[0010] From each perspective of the present invention, a personnel matching system and a storage medium are provided that contribute to achieving a more appropriate match between personnel and projects. However, the effects of the present invention are not limited to those described above. The present invention may produce other effects in lieu of or in conjunction with the effects described above.

[0011] Figure 1 is a diagram illustrating the outline of one embodiment. Figure 2 is a flowchart showing the operation of one embodiment. Figure 3 is a diagram conceptually illustrating the system of the present disclosure. Figure 4 is a diagram showing an example of the overall configuration of the system of the present disclosure. Figure 5 is a diagram showing an example of the functional configuration in the system of the present disclosure. Figure 6 is a diagram showing an example of the processing configuration of the information processing device of the present disclosure. Figure 7 is a flowchart showing the processing procedure of the system of the present disclosure. Figure 8 is a diagram showing an example of the data structure of the system of the present disclosure. Figure 9 is a diagram illustrating the processing procedure of the system of the present disclosure. Figure 10 is a diagram illustrating the analysis process by the system of the present disclosure. Figure 11 is a diagram illustrating the processing algorithm of the matching engine of the present disclosure. Figure 12 is a sequence diagram showing the processing procedure of the system of the present disclosure. Figure 13 is a diagram showing an example of the display screen of the present disclosure. Figure 14 is a diagram showing an example of the display screen of the present disclosure. Figure 15 is a diagram conceptually illustrating the system of the present disclosure. Figure 16 is a diagram showing an example of the functional configuration in the system of the present disclosure. Figure 17 is a flowchart showing the processing procedure of the system of the present disclosure. Figure 18 is a diagram showing an example of the display screen of the present disclosure. Figure 19 shows an example of the hardware configuration of the information processing device according to this disclosure.

[0012] First, an overview of one embodiment will be described. The reference numerals in the drawings attached to this overview are provided for convenience as examples to aid understanding, and this overview is not intended to be limiting in any way. Furthermore, unless otherwise specified, the blocks shown in each drawing represent functional units, not hardware units. The connecting lines between blocks in each drawing include both bidirectional and unidirectional lines. Unidirectional arrows schematically indicate the flow of the main signal (data) and do not exclude bidirectional flow. In this specification and in the drawings, elements that can be similarly described are given the same reference numerals to avoid redundant explanation.

[0013] A talent matching system according to one embodiment comprises a values ​​extraction means 11, a characteristics extraction means 12, a calculation means 13, a response means 14, and a dialogue means 15 (see Figure 1). The values ​​extraction means 11 extracts values ​​information from the participant's work history data (step S1 in Figure 2). The characteristics extraction means 12 extracts project information that indicates the characteristics of the project (step S2). The calculation means 13 calculates the degree of fit between the values ​​information and the project information (step S3). The response means 14 generates a matching result between the participant and the project based on the degree of fit (step S4). The dialogue means 15 confirms the participant's level of interest through an interactive interface using a machine learning model based on the matching result (step S5). The response means 14 determines the matching result based on weighted information between the participant's skills and values ​​information (step S6).

[0014] The above talent matching system generates matching results based on the values ​​of participants (project candidates) and project information. Furthermore, the talent matching system uses the generated matching results as a basis to engage in dialogue with participants and obtain information about their interests and priorities. The talent matching system then determines the final matching result by reflecting (weighting) the acquired information (level of interest) in the value information, etc. Through the operation of this talent matching system, a more appropriate match between participants (talent) and projects is achieved.

[0015] Specific embodiments will be described in more detail below with reference to the drawings.

[0016] [First Embodiment] First, in order to facilitate understanding of the embodiments disclosed herein, the prerequisites for requiring the system according to the embodiments disclosed herein will be explained. More specifically, the prerequisites for requiring the system disclosed herein will be explained, such as business operations within a company.

[0017] In the corporate environment targeted by the system disclosed in this application, various projects are underway in parallel, and each employee is expected to be assigned to an appropriate project according to their skill set and experience. In traditional personnel placement processes, project managers and human resources departments have taken the lead in selecting personnel based on documented information such as resumes and skill sheets. These documents are stored in a database, and matching is performed with the necessary skill sets for the project requirements.

[0018] Furthermore, various types of data are generated within a company through daily operations. For example, the daily work report system allows employees to record their daily work content, results, and challenges. Internal communication tools facilitate discussions and knowledge sharing regarding projects, accumulating vast amounts of text data. In addition, knowledge management systems and document management systems store past project materials and technical documents.

[0019] From a human resource development perspective, employee skill acquisition and qualification history are managed through training and learning management systems. Information on career aspirations (career path orientation) and desired future job roles is also collected during regular goal setting and performance review meetings. Furthermore, participation in internal innovation programs and suggestion systems provides insights into employees' areas of spontaneous interest and areas where they demonstrate creativity.

[0020] However, these abundant data sources are fragmented, and cross-sectional analysis and integrated utilization are not being adequately performed. In particular, information on employees' latent values ​​and motivations remains at the level of qualitative evaluation and is not accumulated or utilized as quantitative data. Furthermore, while technical requirements for project information are clearly defined, the multifaceted characteristics of the project, such as its social significance and market positioning, are not managed in a structured manner.

[0021] In this corporate environment, the system disclosed in this application aims to integrate and analyze diverse data sources that were previously fragmented, and to find the compatibility between employees' values ​​and motivations and the multifaceted characteristics of projects. This will enable talent placement that goes beyond mere skill matching, resulting in more sustainable motivation and higher performance. Furthermore, it is expected that employees will have more opportunities to participate in projects that align with their values, leading to improved career satisfaction and the realization of their potential.

[0022] <Outline of this Embodiment> Next, based on the background described above, an overview of this embodiment will be explained.

[0023] Figure 3 is a diagram illustrating the concept of a matching system based on values ​​and motivations as disclosed in this application.

[0024] The system disclosed in this application consists of four main parts.

[0025] The rectangle located in the upper left of Figure 3 represents project information, which includes information such as technical requirements, project objectives, and social impact. This project information is sent to a characteristics extraction process, which is represented as a circle in Figure 3. In the characteristics extraction process, the essential characteristics and requirements of the project are extracted as structured data.

[0026] The rectangle located in the lower left of Figure 3 represents employee information, and this rectangle aggregates data such as work logs, chat history, and activity records. This information is input into the values ​​estimation process. The values ​​estimation process is also represented as a circle, similar to the characteristic extraction process. In the values ​​estimation process, individual preferences, potential skills, and values ​​are extracted from the employee's behavior patterns and statements.

[0027] The results obtained from the characteristic extraction process and the values ​​estimation process are sent to the matching engine, which is shown as a diamond in the center of Figure 3. The matching engine compares project characteristics and employee values ​​in multiple dimensions and calculates the degree of fit. The results from the matching engine are then passed on to the AI ​​(Artificial Intelligence) dialogue system, which is shown as a rounded rectangle on the right.

[0028] The AI ​​dialogue system uses initial suitability scores received from the matching engine to assess an employee's level of interest in a project through dialogue. This dialogue process utilizes a large-scale language model to extract more detailed information about the employee's values ​​and motivations from their responses.

[0029] Furthermore, new insights gained through dialogue with employees are fed back into the matching engine, as indicated by the dotted arrows, contributing to the optimization of matching accuracy. Through this feedback loop, the system can more accurately capture employees' true interests and values, and gradually improve the accuracy of matching them with projects.

[0030] The system disclosed in this application features a focus not only on skills and experience, but also on employees' intrinsic values ​​and motivations, which are then matched with project characteristics to achieve long-term motivation maintenance and high performance. Furthermore, the system disclosed in this application incorporates a gradual accuracy improvement mechanism through AI dialogue, enabling matching that takes into account individuals' latent interests and preferences that cannot be fully grasped by data alone.

[0031] <Overall Configuration> The system disclosed in this application includes an information processing device 10 and a terminal 20, as shown in Figure 4. The information processing device 10 is a server on the cloud. Users access the information processing device 10 using a terminal 20 such as a personal computer or smartphone.

[0032] Next, the functional configuration of the embodiment disclosed in this application will be described.

[0033] <Device Configuration> The device configuration according to the embodiment disclosed herein will be described below.

[0034] Figure 5 is a block diagram showing an example of the functional configuration of the system disclosed in this application. The system is broadly composed of five functional blocks, each responsible for a specific process.

[0035] The input processing unit 101 is a function (module) for inputting data into the system. The input processing unit 101 consists of three subcomponents. The upper section houses a project information acquisition function (project information acquisition unit). The project information acquisition function acquires data from project-related documents such as requirements definition documents and goal setting documents. The middle section houses an employee information acquisition function (employee information acquisition unit), which collects employee behavior data such as work logs and chat history. The input processing unit 101 may also include a data preprocessing function (data preprocessing unit). This data preprocessing function may perform preprocessing such as noise reduction, normalization, and structuring of the acquired data, and convert it into a format suitable for subsequent analysis.

[0036] The analysis processing unit 102 is a function (module) that performs analysis of acquired data. The upper natural language processing function (natural language processing unit) performs language processing such as text data analysis, sentiment analysis, and topic extraction. The middle characteristic extraction function (characteristic extraction unit) grasps the characteristics of the project from multiple perspectives by performing analysis of technical requirements and extraction of objectives (objective analysis) and evaluation of marketability (market value evaluation) from project information. The lower value estimation function (value estimation unit) performs a process to vectorize employee values ​​by recognizing patterns from employee data (employee information) and analyzing trends.

[0037] The matching processing unit 103 is a function (module) that performs matching processing based on the analysis results. The upper similarity calculation function (similarity calculation unit) quantifies the similarity between project characteristics and employee values ​​using methods such as vector distance, cosine similarity, and Mahalanobis distance. The middle weighting parameter function (weighting parameter unit) manages the weighting for values ​​and skills, and adjusts the matching priority. The lower overall score calculation function (overall score calculation unit) integrates multidimensional evaluation indicators and derives the final matching score by applying threshold processing.

[0038] The dialogue processing unit 104 is a function (module) that improves matching accuracy through dialogue with employees. The upper large-scale language model function (large-scale language model unit) performs natural language dialogue processing such as question generation and answer analysis. The middle interactive confirmation function (interactive confirmation unit) confirms the employee's level of interest, adjusts the details of their values, and listens to their specific wishes. The lower parameter adjustment function (parameter adjustment unit) performs feedback processing to update the weighting parameters of the matching processing unit 103 based on the information obtained through the dialogue.

[0039] The dialogue processing in the dialogue processing unit 104 is implemented using a machine learning model, but the implementation of this dialogue processing is not limited to large-scale language models. For example, a variety of machine learning methods and artificial intelligence technologies can be applied to the dialogue processing, such as conventional natural language processing techniques, machine learning algorithms such as decision trees and random forests, or rule-based expert systems. A hybrid approach combining multiple technologies is also effective, and the optimal technology selection may be made according to the implementation environment and requirements while ensuring the quality and efficiency of the dialogue with participants.

[0040] The output processing unit 105 is a function (module) responsible for the final output of the system. Four sub-components are horizontally arranged in the output processing unit 105, and functions such as generation of matching results (matching result generation unit), visualization of results (visualization unit), provision of recommended training information (training recommendation unit), and management of feedback information (feedback management unit) are implemented from the left.

[0041] The information processing apparatus 10 is configured to include the input processing unit 101, the analysis processing unit 102, the matching processing unit 103, the dialogue processing unit 104, and the output processing unit 105 described above.

[0042] As shown in FIG. 6, the information processing apparatus 10 is configured to include a communication control unit 100, an input processing unit 101, an analysis processing unit 102, a matching processing unit 103, a dialogue processing unit 104, an output processing unit 105, and a storage unit 106.

[0043] The communication control unit 100 is means for controlling communication with other apparatuses. For example, the communication control unit 100 receives data (packets) from the terminal 20. The communication control unit 100 also transmits data to the terminal 20. The communication control unit 100 delivers data received from other apparatuses to other processing modules. The communication control unit 100 transmits data acquired from other processing modules to other apparatuses. In this way, other processing modules transmit and receive data to and from other apparatuses via the communication control unit 100. The communication control unit 100 includes a function as a receiving unit that receives data from other apparatuses and a function as a transmitting unit that transmits data to other apparatuses.

[0044] The storage unit 106 is means for storing information necessary for the operation of the information processing apparatus 10.

[0045] <Processing Flow> A processing flow according to an embodiment of the present disclosure will be described.

[0046] FIG. 7 is a flowchart showing a processing flow of the system of the present disclosure.

[0047] Processing of the system disclosed in the present application starts from the topmost "start" node and branches into two parallel processes. Project information processing is performed on the left side, and employee information processing is executed on the right side.

[0048] In project information processing, "project information acquisition" is performed (step S01). Here, necessary data is collected from requirement definition documents, project plans, and the like.

[0049] Subsequently, the process proceeds to the step of "feature extraction processing", and features are extracted from the acquired project information based on three axes: technical aspect, social aspect, and business aspect (step S02). As a result, the multifaceted characteristics of the project are expressed as structured data.

[0050] In the employee information processing on the right side, data such as business logs, chat histories, and activity records are input in step S03 of "employee information input".

[0051] Subsequently, the process proceeds to step S04 of "value estimation processing", and a skill vector and a value vector are generated from the input data. Note that the skill vector is a vector composed of elements obtained by digitizing each skill possessed by an employee. Further, the value vector is a vector composed of elements obtained by digitizing the employee's values for work. Through this processing, not only the employee's technical ability but also information about the employee's intrinsic values and motivations are quantified.

[0052] The results of the left and right parallel processes are aggregated into the central "matching calculation" node. In this node, matching calculation is executed (step S05). More specifically, in this node, similarity calculation between the extracted project characteristics and the employee's values and skill vectors is performed, and appropriate weighting is applied. A "first-stage matching result" is generated based on the calculation result (step S06), and an initial matching score is calculated.

[0053] The initial matching results are sent to step S07, "Interest Confirmation via AI Dialogue." In this step, an interactive dialogue is conducted with the employee using a dialogue system based on a large-scale language model. Through this dialogue, the employee's level of interest in the matching results, specific expectations, and concerns are confirmed.

[0054] The result of the dialogue is sent to a decision point called "Result Confirmed?". At this decision point, it is evaluated whether the matching result has reached sufficient accuracy and it is determined whether the result has been confirmed (step S08). If the result of the decision is "Yes", the process proceeds to the "Final Matching Result" node, which is step S09 on the right. At this node, follow-up processes such as training recommendations and feedback collection are performed based on the confirmed matching result.

[0055] On the other hand, if the result of the judgment is "No," the process proceeds to step S10 on the left, the "Matching Parameter Adjustment" node. In this node, the weighting parameters and thresholds for the similarity calculation are adjusted based on the information obtained from the dialogue. The adjusted parameters are then fed back to the "Matching Calculation" node, and an improved matching result is generated.

[0056] Furthermore, the loop of dialogue, judgment, and parameter adjustment is repeatedly executed until the matching accuracy reaches a satisfactory level. In other words, in the system disclosed in this application, the dialogue and adjustment loop is executed multiple times.

[0057] For example, the criteria for determining a "satisfactory level" in the matching process may be a combination of several conditions, such as the following:

[0058] A matching is judged to be at a "satisfactory level" when any of the following conditions are met: (1) convergence of the matching score: the change in the score between two consecutive dialogues falls below a threshold ε (e.g., 0.05); (2) explicit expression of satisfaction by the participant: statements such as "I am satisfied with this matching result" are confirmed; (3) reduction of information entropy: the amount of information obtained from the new dialogue (amount of entropy reduction) falls below a threshold; or (4) qualitative satisfaction of the dialogue: sufficient information is obtained for all important categories (technological, social, career, etc.).

[0059] As a control condition for the loop, the maximum number of iterations can be set to five. Even if it is determined that further dialogue is necessary, a mechanism may be introduced to first finalize a provisional result and then propose an additional dialogue session at a later date. Furthermore, as a convergence condition, a process may be implemented to terminate the loop if the fluctuation range of the matching score is less than 0.03 for three consecutive times, or if the participant's response is 85% or more similar to the previous response (measured by cosine similarity between semantic vectors of the text).

[0060] To handle situations where a participant wishes to end the conversation midway, three options may be presented at the end of each conversation step: "Continue," "Save and resume later," and "Confirm current results." If "Save" is selected, the conversation content up to the present and the intermediate matching results are saved, and a function is provided to resume from that state. If "Confirm current results" is selected, the matching process is completed based only on the collected information, and the confidence level of the result (e.g., "70% confidence") may also be displayed. This improves the usability of the system while flexibly accommodating participants' time constraints.

[0061] This iterative optimization process enables highly accurate matching that incorporates employee feedback and opinions, rather than simply a one-off matching process.

[0062] Thus, the system disclosed in this application employs a structure that combines a data-driven approach with interactive human interaction to achieve flexible and highly accurate talent matching that takes values ​​and motivations into consideration.

[0063] The above is an overview of the embodiment disclosed in this application. Furthermore, the operation of the embodiment disclosed in this application will be described.

[0064] <Data Structure> First, the structure of the data handled in the embodiments disclosed herein will be explained.

[0065] Figure 8 is a diagram illustrating the concept of data structure. Figure 8 consists of three main areas: the values ​​data model on the left, the project characteristics model on the right, and the correspondence between these two models in the center. The specific methods for calculating this data will be described later, but here we will explain the data structure.

[0066] The values ​​data model on the left represents information extracted from employee work logs and chat history as a multidimensional vector. The top section shows an overview of the multidimensional vector representation, describing features such as information extracted from work logs, skill elements, values ​​elements, and vectors that take into account changes over time (deterioration over time). The model is mainly structured around three main axes, capturing employee characteristics from the perspectives of "technical interests," "social values," and "career orientation."

[0067] The "Technical Interests" axis includes elements such as an inclination towards technical fields (backend / frontend, etc.) and an inclination towards technological challenges. The "Social Values" axis evaluates the degree to which an employee values ​​social contribution, user experience, and human-centered approaches. The "Career Aspirations" axis expresses elements such as career path priorities and the balance between long-term growth and short-term results. These three axes are interrelated and comprehensively form an employee's values ​​profile.

[0068] The project characteristics model on the right also employs a three-axis representation structure. The top section indicates that the model is constructed based on information extracted from project documents, such as requirements specifications and objectives, and consists of a hierarchical categorization and weighted evaluation metrics.

[0069] The three axes of project characteristics are "technical requirements," "social impact," and "marketability." The "technical requirements" axis evaluates the technology stack required for the project, its technical challenge, and its innovativeness. The "social impact" axis measures the project's social contribution, public benefit, and impact on users. The "marketability" axis evaluates aspects such as business impact, growth market, and strategic importance. These axes are interrelated and represent the overall characteristics of the project.

[0070] The central section shows the correspondence between the values ​​data model and the project characteristics model. Here, the degree of fit between the two models is calculated using methods such as cosine similarity, weighted evaluation, and multidimensional matching. The calculation results are quantified as a "matching score" and expressed as a value in the range of 0.0 to 1.0. In addition to the overall score, partial scores for each axis are also calculated, enabling a multifaceted evaluation.

[0071] Furthermore, if there are differences in the axis configuration between the values ​​data model and the project characteristics model, a simple similarity comparison may not be possible. To address such challenges, a correspondence matrix may be defined. Specifically, a 3x3 matrix is ​​constructed that represents the degree of association between the three axes of the values ​​model (technical interest, social values, career orientation) and the three axes of the project characteristics model (technical requirements, social impact, marketability). For example, the degree of association between "technical interest" and "technical requirements" may be quantified as 0.9 (strong), and the degree of association between "technical interest" and "social impact" as 0.3 (weak).

[0072] As a method for calculating similarities between models with different dimensional structures, multidimensional similarity calculation based on tensor products is introduced. First, each axis of both models is represented as a normalized basis vector, and a common high-dimensional space is constructed by their tensor product. Next, by performing distance calculations (such as the Mahalanobis distance) within this high-dimensional space, the similarity between vectors with different dimensional structures can be calculated in a consistent manner.

[0073] Furthermore, to ensure semantic consistency, methods for calculating semantic similarity across axes using word embedding models such as Word2Vec may also be employed. For example, "long-term growth," a sub-element of "career orientation," and "growth market," a sub-element of "marketability," are semantically close in the word embedding space, and therefore will be assigned high values ​​in the correspondence matrix. This enables appropriate matching between semantically related concepts, even if they are expressed using superficially different terminology.

[0074] Furthermore, the definition of values ​​in this embodiment can also be understood as a synthesis of multiple axes defined in R-CAP (Resource-based Career Assessment Program) based on data acquired from users. Specifically, by comprehensively analyzing information obtained from each of the following axes—GIAL (General Interest and Activity List), BSI (Basic Skills Inventory), CIS (Career Interest Survey), WVI (Work Values ​​Inventory), CD (Career Drivers), CI (Career Implications), and JFI (Job Feature Inventory)—a more comprehensive definition of "values" can be achieved.

[0075] Each of these axes of R-CAP can also be said to be interrelated with the three main axes defined in this embodiment (technological interests, social values, and career aspirations). For example, WVI (Work Values ​​Inventory) is strongly associated with social values, CIS (Career Interest Survey) with technological interests, and CD (Career Drivers) with career aspirations. In implementing the system, it is possible to define a mapping function that converts the data obtained from each axis of R-CAP into the three main axes, and if there are existing R-CAP evaluation results, it is also possible to construct value vectors using them as initial values.

[0076] Furthermore, a key feature of this system's value extraction process is its ability to estimate values ​​not only from explicit evaluations like R-CAP, but also from text data naturally generated during daily work. By combining both approaches, a more precise matching can be achieved that considers both implicit value elements that cannot be captured by formal evaluations and systematically organized value elements.

[0077] This data structure allows the system disclosed in this application to go beyond simple existing skill matching and quantitatively evaluate the compatibility between employees' intrinsic values ​​and motivations and the multifaceted characteristics of a project. Furthermore, since this data structure is continuously optimized through adjustment and learning via dialogue processing, the matching accuracy improves over time.

[0078] <Details of Employee Data Processing Method> Next, we will explain in more detail the details of how employee information is processed. Figure 9 is a diagram showing an example of the employee information processing flow. Figure 9 represents a series of processing steps that generate value vectors from various employee data sources.

[0079] The "Input Data" block located in the upper left of Figure 9 shows the information sources that serve as the starting point for processing. This block includes text data generated from employees' daily work activities, such as work reports, chat logs, and activity records. This data is a valuable source of information that indirectly reflects employees' interests and values. As a concrete example of input data, the dotted box on the left shows an example of a work report entry that reads, "We worked on improving the UI design and were able to enhance user usability. In the future, we would like to try expanding functionality using AI technology."

[0080] The input data is sent along the arrow to the "Natural Language Processing" block in the upper center. The input data acquired by the input processing unit 101 is sent to the natural language processing function of the analysis processing unit 102. In this block, natural language processing techniques such as feature word extraction, sentiment analysis, and topic analysis are applied. Feature word extraction identifies words related to technical terms and values ​​from the text. Sentiment analysis detects emotional elements and attitudes from the text. Topic analysis extracts the theme and subject of the entire text. As an example of the results of these processes, extracted features such as "UI Design: Technical Keywords," "User Experience: Social Aspects," and "AI Technology: Technical Keywords + Interests" are shown in the dotted line frame in the lower left.

[0081] The results of natural language processing are sent in two directions: one to the "Values ​​Categories" block in the upper right, and the other to the "Machine Learning Model" block in the center. The Values ​​Categories block defines the three main axes of values ​​that the system disclosed in this application analyzes: "Technological Interests," "Social Values," and "Career Aspirations." These categories are a predefined classification system of values, providing a framework for understanding employee characteristics from multiple perspectives.

[0082] The central "Machine Learning Model" block receives the results of natural language processing and definitions of value categories as input. The "Machine Learning Model" block is part of the value estimation function. Within this block, processes such as feature association, value vector generation, and time-series change considerations are performed. Feature association analyzes the relationship between extracted feature words and sentiment expressions and value categories. Value vector generation constructs multidimensional numerical vectors based on the associated features. Time-series change considerations evaluate the temporal changes and consistency of values ​​by comparing them with past data.

[0083] As an example of the results of these machine learning processes, a specific numerical example of a values ​​vector is shown in the dotted box in the center right. For example, a multidimensional vector is generated in which multiple numerical values ​​are represented for each category, such as "Technical Interest: [0.8, 0.7, 0.9]". These numerical values ​​represent the employee's tendency or strength towards each value element.

[0084] Ultimately, the output from the machine learning model is aggregated in the "Processing Results" block located at the bottom of Figure 9. This block generates two main deliverables: "Value Vectors" and "Confidence Score."

[0085] The values ​​vector is a data structure that represents an employee's values ​​profile in a multidimensional manner and serves as input to the subsequent matching process (matching processing unit 103). The confidence score is an indicator of the reliability and accuracy of the generated vector and fluctuates depending on the quality and quantity of the data.

[0086] In generating the values ​​vector, each of the axes—"technological interest," "social values," and "career orientation"—is broken down into multiple subcategories. For example, in the three-dimensional vector "Technological Interest: [0.8, 0.7, 0.9]," the first element (0.8) represents "interest in new technologies," the second element (0.7) represents "orientation towards technological depth," and the third element (0.9) represents "interest in the diversity of technological fields."

[0087] The conversion from text data to value vectors can begin by analyzing the frequency and context of characteristic words using a pre-constructed dictionary of technical terms and value expressions. Next, a machine learning model (a fine-tuning model of a pre-trained language model such as BERT) is used to obtain the semantic representation of the sentence, and a matrix transformation is applied to map it to the value space. For example, the sentence "I am interested in AI" might be assigned a high score of around 0.8 for "interest in new technologies" based on the co-occurrence of the technical word "AI" and the interest expression.

[0088] The exponential moving average (EMA) is applied to account for time-series changes. The past vector value V_old and the newly calculated vector value V_new are updated using the formula V_updated = α * V_new + (1-α) * V_old. Here, α is the time decay coefficient (e.g., 0.2), and more recent data has a greater impact. Furthermore, if data is available for more than three months, a trend detection algorithm is applied to detect upward and downward trends.

[0089] This processing flow enables the conversion of text-based qualitative data about employees into quantitative value vectors using machine learning. The generated value vectors quantify employees' intrinsic motivations and orientations, allowing for an objective and multifaceted evaluation of the compatibility between individual potential characteristics and project characteristics, which could not be captured by existing systems. Furthermore, by considering changes over time, dynamic matching that reflects the development and changes in values ​​is achieved.

[0090] Thus, the analysis processing unit 102 includes a value extraction means that extracts value information from the participants' work history data, and a characteristic extraction means that extracts project information that indicates the characteristics of the project. More specifically, the value extraction means generates skill vectors and value vectors from the participants' chat logs and work history. The characteristic extraction means expresses the project information along three axes: technical, social, and industry.

[0091] <Processing Method for Project Data> Next, we will explain in more detail the processing of project information. Figure 10 is a diagram illustrating the project characteristics analysis process. Figure 10 shows a series of processing steps from various documents related to the project to the generation of characteristic vectors.

[0092] The "Input Data Sources" block located in the upper left of Figure 10 shows the documents that serve as the starting point for the analysis. The main sources of information are official documents created during the development project, such as project plans, requirements specifications, and technical specifications. These documents contain a wide range of information, including project objectives, technical requirements, development methodologies, schedules, and resource allocation.

[0093] Within the dotted box in the lower left, "AI-powered medical diagnostic support" is shown as a specific project example, illustrating its technology stack with machine learning (deep learning) and cloud infrastructure, and its business objectives such as improving the accuracy of medical diagnoses and streamlining doctors' work.

[0094] Information from the input data source is sent along the arrow to the "Extracted Information" block at the top center. The information from the input data source is sent to the characteristic extraction function of the analysis processing unit 102. In this block, information necessary for characteristic analysis is extracted from the project documents using natural language processing and structured data analysis. Specifically, information such as the technology stack (language used, framework, infrastructure environment, etc.), development methodology (agile, waterfall, etc.), and business objectives (cost reduction, market expansion, improved customer experience, etc.) is extracted in a structured format.

[0095] The extracted information is classified according to the framework shown in the "Evaluation Axis" block in the upper right. For example, the evaluation axis consists of three main categories: "Technical Aspects," "Social Aspects," and "Industry Aspects," which provide fundamental perspectives for evaluating a project from multiple angles. The technical aspect evaluates the project's technical characteristics, difficulty, and innovativeness; the social aspect evaluates its impact on users, social value, and ethical aspects; and the industry aspect evaluates its market positioning, competitive landscape, and business impact.

[0096] Both the extracted information and evaluation axes are entered into the "Hierarchical Evaluation" block in the central section. In this block, a detailed evaluation of project characteristics is conducted hierarchically. For example, in the technical aspects, "technological advancement" is weighted at 0.4 and "difficulty of learning" at 0.3. Similarly, in the social aspects, "social impact" and "improvement of user experience" are each weighted at 0.5. This hierarchical structure and weighting enable a multifaceted and granular evaluation of the project.

[0097] The results of the hierarchical structure evaluation are quantified in the "Scoring Results" block in the right-center of Figure 10. A score is calculated for each evaluation axis on a scale from 0 to 1, with Figure 10 showing examples such as a technical score of 0.75, a social score of 0.82, and an industry score of 0.68. These scores are the result of weighting and aggregating the scores of individual evaluation items, and quantitatively represent the characteristics of the project.

[0098] Finally, the information from the hierarchical structure evaluation and scoring results is integrated into the "Project Characteristics Vector" block located at the bottom of Figure 10. Here, a multidimensional vector of the form [0.75, 0.82, 0.68, 0.70, 0.88, 0.65, ...] is generated. Each dimension of this vector represents the score of the evaluation axis and its sub-items, providing a comprehensive and precise representation of the project's characteristics.

[0099] The project characteristics vector is used to calculate the similarity with employee values ​​vectors and serves as the foundational data for the matching process. Its hierarchical and multifaceted evaluation structure allows for a more comprehensive understanding of project characteristics, going beyond simple technical requirements matching. Furthermore, the adoption of a unified numerical vector format enables automated matching calculations and optimization by computer. Thus, the process disclosed in this application systematizes the data processing flow, starting with information extraction from project documents, progressing through hierarchical evaluation, and transforming the data into a quantitative vector representation.

[0100] <Details of the Matching Process Operation> Next, we will explain the details of the operation of the matching process (matching processing unit 103). Figure 11 is a diagram illustrating the processing algorithm of the matching engine. This algorithm takes employee value vectors and project characteristic vectors as input and numerically evaluates the optimal combination of the two.

[0101] On the left side of Figure 11, two sets of input data are arranged. The upper "Values ​​Vector" is data generated from employee information and consists of multidimensional vectors such as technical interest [0.8, 0.7, 0.9], social values ​​[0.7, 0.8, 0.5], and career orientation [0.6, 0.8, 0.7]. The lower "Project Characteristics" is data generated from project information and includes elements (vectors) such as technical requirements [0.7, 0.9, 0.8], social impact [0.8, 0.7, 0.6], and marketability [0.5, 0.6, 0.8]. These vector representations make it possible to quantitatively handle qualitative concepts such as human values ​​and project characteristics.

[0102] The central "matching process" section consists of three processing steps.

[0103] In the initial "similarity calculation," the similarity between the two input vectors is measured. Specifically, mathematical methods such as cosine similarity and Euclidean distance are applied. The "similarity calculation" is implemented by the similarity calculation function of the matching processing unit 103.

[0104] The cosine similarity is calculated using equation (1) below, and the Euclidean distance is calculated using equation (2).

[0105] ... (1)

[0106] ... (2)

[0107] For example, when calculating the cosine similarity between the technical interest vector [0.8, 0.7, 0.9] and the technical requirements vector [0.7, 0.9, 0.8], the numerator is 0.8 × 0.7 + 0.7 × 0.9 + 0.9 × 0.8, The formula has as the denominator, and as a result, a high similarity value is obtained.

[0108] The next step, "Skill Suitability," evaluates the compatibility between an employee's skill vector and the project's requirements. This evaluation includes not only checking the degree of agreement with the requirements but also assessing the skill level. For example, if the project requires Java (registered trademark, hereinafter the same) programming skill level 4, it is determined whether the employee's Java skill level meets this requirement.

[0109] In the third step, "weighting calculation," the similarity score and skill suitability score obtained in the previous two steps are integrated. The "weighting calculation" is performed by the weighting parameter function of the matching processing unit 103. Here, the formula "α × value similarity + β × skill suitability" is used, and α and β are shown as adjustable parameters in the "adjustable parameters" section in the upper right. For example, in Figure 11, the value weight (α) is set to 0.7 and the skill weight (β) is set to 0.3, indicating that a matching policy is adopted in which value similarity takes precedence over skill suitability.

[0110] Furthermore, the initial values ​​of the weighting parameters α (values ​​weight) and β (skills weight) may be set by combining three elements: (1) automatic adjustment based on project attributes, (2) basic settings based on organizational policies, and (3) analysis of past success patterns. For example, initial values ​​such as α = 0.6 and β = 0.4 for research and development projects, and α = 0.7 and β = 0.3 for customer service projects may be set according to the project type. In addition, if there is a policy of "emphasizing values" as an overall organizational policy, constraints such as setting the lower limit of α to 0.5 can also be applied.

[0111] A hierarchical Bayesian model can be used to adjust parameters based on information obtained from the dialogue. Information extracted from participants' responses (e.g., "I value social impact") is updated using Bayesian inference to update the probability distribution. Specifically, the prior distribution P(α) before the dialogue is updated to a posterior distribution P(α|D) conditioned on the dialogue information D. For example, if the statement "I value social impact" is confirmed, the weight related to social values ​​is adjusted to increase statistically significantly.

[0112] The adjustment of the technical skills weight from 0.7 to 0.5 corresponds to the participant's statement emphasizing social values, "I am attracted to the social impact of using data in the medical field." Based on the information extracted from this statement, "Emphasis on social value: 0.8 (high)," the balance of the relative importance of values ​​and skills is updated. The adjustment amount is calculated by multiplying the difference between the intensity of the statement (0.8) and the current weight (0.7) by the learning rate of 0.5, resulting in 0.7 - 0.5 * (0.8 - 0.7) = 0.5.

[0113] The "Feedback" section in the right-center of Figure 11 represents the mechanism for adjusting parameters based on information obtained from the dialogue system. For example, if information is obtained through dialogue with an employee indicating a "high interest in technical challenges," the weight given to technical aspects will be increased. This feedback loop enables adaptive matching, gradually improving accuracy from the initial settings.

[0114] The "Matching Results" section at the bottom center of Figure 11 shows the output of the process. The matching results are generated by the overall score calculation function of the matching processing unit 103. In Figure 11, the overall matching score is calculated to be 0.82, with a breakdown of a values ​​score of 0.85 and a skills score of 0.75. This indicates a case where the similarity in values ​​is high, but the fit in terms of skills is somewhat low.

[0115] The "Output Utilization" section in the lower right of Figure 11 shows how the calculated matching results can be used. The radar chart display visualizes the multifaceted aspects of the matching, and specific learning content is suggested to supplement the skills lacking in the training recommendation generation. In other words, the output processing unit 105 functions as a training recommendation means that recommends training courses necessary for the participant based on the matching results. The training recommendation means uses a machine learning model to identify training to recommend to the participant and provides the participant with information on the identified training.

[0116] The supplementary explanation in the lower left of Figure 11 includes technical details such as axis-specific calculations in similarity calculations, consideration of importance between dimensions, and normalization processing, as well as explanations of thresholding processes such as essential requirement checks and minimum goodness-of-fit settings.

[0117] The matching engine employs a hybrid approach that combines mathematical similarity calculations with machine learning-based weighting optimization, and further incorporates human intent through dialogue. This enables sophisticated matching that goes beyond simple keyword matching, taking into account values ​​and motivations.

[0118] Thus, the matching processing unit 103 includes a function as a calculation means for calculating the degree of fit between value information and project information, and a function as a response means for generating matching results between participants and projects based on the degree of fit. Furthermore, the response means determines (finalizes) the matching result based on weighting information between the participant's skills and value information.

[0119] <Details of Dialogue Processing Operation> The detailed operation of the dialogue processing unit 104 and each function included in the dialogue processing unit 104 (large-scale language model function, interactive confirmation function, parameter adjustment function) will be explained. Figure 12 is a sequence diagram that visualizes the dialogue process between the participant and the system in chronological order. Three main actors are arranged from left to right: the participant, the dialogue processing unit 104 in the center, and the matching engine on the right. The flow of information and processing exchanged between them is represented by arrows.

[0120] The detailed operation of the dialogue processing unit 104 begins with presenting the first-stage matching results obtained from the matching engine to the participant (step S11). Based on this presentation, the dialogue processing unit 104 uses a large-scale language model to generate questions that elicit the participant's values ​​and interests (step S12). For example, questions about values ​​are asked to the participant, and the participant answers the questions (steps S13, S14).

[0121] Specific examples of such dialogues include a system asking, "Your skill set excels in cloud infrastructure construction and distributed system design, but which technical element of this project interests you most?" and a participant responding, "I'm interested in implementing a microservices architecture." Another example is a question asking, "Which aspect of this project—its social significance or its technical challenges—doesn't appeal to you more?" and a participant responding, "I'm more drawn to the social impact of data utilization in the medical field than to the technical challenges."

[0122] The dialogue processing unit 104 extracts numerical vector information such as "microservices: 0.8", "social impact: 0.9", and "utilization of medical data: 0.7" from these responses and sends it to the matching engine (step S15). The matching engine adjusts the weighting, which was originally "technical skills: 0.7" and "social values: 0.3", to "technical skills: 0.5" and "social values: 0.5". In other words, the matching engine dynamically adjusts the weighting of values ​​and skills based on the dialogue analysis (step S16). The updated matching results are sent back to the dialogue processing unit 104 (step S17).

[0123] Upon receiving the updated matching results, the dialogue processing unit 104 generates follow-up questions such as, "Do you have experience with medical data anonymization technology? Or do you see this as an opportunity to acquire new skills?" Based on the generated follow-up questions, the dialogue processing unit 104 asks the participant questions (step S18). For example, the participant might respond, "I don't have direct experience, but I am interested in privacy protection technology and would like to learn about it on this occasion" (step S19). Upon receiving this response, the dialogue processing unit 104 extracts information such as "Skill gap: Medical data anonymization" and "Motivation to learn: 0.8" and sends it back to the matching engine.

[0124] The dialogue processing unit 104 saves the dialogue history with the participant (step S20).

[0125] The above type of dialogue is usually repeated three to five times, with the questions becoming more specific and individualized each time. For example, questions about work preferences and roles may be included, such as, "How many days a week would you prefer to work remotely?" or "In terms of your role on the team, would you prefer a position closer to a technical lead or project management role?"

[0126] Text analysis utilizes natural language processing techniques such as sentiment analysis and intent recognition, and the information extracted from responses is mapped onto a multidimensional vector space. For example, the space is composed of axes such as "social impact," "technological challenges," "autonomy," and "team collaboration," and the responses are analyzed to determine which area they fall into. The analysis results are reflected in the matching score, improving the accuracy of the final project recommendation.

[0127] The specific text analysis method involves a two-stage process. In the first stage, a prompt engineering method based on a predefined analysis framework is used. For example, an instruction such as "Score the user's responses on three axes—technological interest, social values, and career aspirations—using a numerical value from 0 to 1, and output the results in JSON format" is given, resulting in a structured output.

[0128] In the second stage, post-processing is applied to normalize and validate the output of the large-scale language model. Specifically, this involves (1) syntactic validation of the output JSON format data, (2) detection and correction of extreme values ​​(reliability validation of values ​​above 0.95 or below 0.05), (3) consistency validation with past dialogue history (generating confirmation questions if there are sudden changes in values), and (4) quantification of uncertainty (assigning confidence scores to answers with low confidence in the model).

[0129] To address the ambiguity of natural language, dialogue designs that incorporate redundancy by asking the same information from different angles may be employed. For example, a direct question such as "Do you value technological challenges or social value more?" could be combined with an indirect question such as "What is most important to you when choosing a job?" to verify the consistency of the answer. Additionally, confirming questions such as "From your answer, I understand that the value of XX is important to you; is that correct?" can be periodically inserted into the dialogue to reduce the risk of misunderstanding.

[0130] The data collected throughout the dialogue will be structured and used to improve future matching algorithms and analyze the time-series changes in participants' values. As a result, more accurate and sustainable project matching will be achieved, taking into account not only skills but also values ​​and aspirations.

[0131] Thus, the dialogue processing unit 104 functions as a dialogue means to confirm the degree of interest of participants through an interactive interface using a large-scale language model, based on the matching results. Specifically, the dialogue means collects information about projects that the participants are interested in and information about matching results that the participants are interested in, through the interactive interface. Furthermore, the dialogue means confirms the degree of interest of the participants in stages and updates weighting information (weights for values, etc.) based on the confirmation results. In addition, the values ​​information includes information about the participants' desired work style, and the matching results may be determined based on the participants' desired work style.

[0132] <User Interface Example 1: Matching Results Screen> In the embodiment disclosed herein, the user interface output by the output processing unit 105 will be described. Figure 13 shows an example of a matching results visualization screen output by the output processing unit 105. This screen is designed to allow users to grasp the suitability of participants and projects from multiple perspectives and intuitively.

[0133] Although omitted from the illustration due to space limitations, a colored title bar (for example, a blue title bar) may be placed at the top of the screen shown in Figure 13, displaying "Project Matching Results" and the username on the right. This display allows participants to immediately recognize that the information is their own matching information.

[0134] In the upper left of Figure 13, an octagonal radar chart is displayed. This radar chart visualizes the degree of matching across eight evaluation axes—technical fit (P1 in Figure 13), industry fit (P2), work style (P3), growth opportunities (P4), team fit (P5), social values ​​(P6), career orientation (P7), and compensation (P8)—as filled areas (for example, blue areas). The system is designed so that users can intuitively understand which aspects show a high degree of fit and which aspects have room for improvement simply by looking at the shape of the polygon.

[0135] To the right of the radar chart, the project name "Cloud-Native Medical System Development" and the overall matching score "82%" are displayed in bold, and below that, individual fit scores such as technology stack fit, social value fit, work style fit, and career growth opportunities are shown numerically (the individual fit scores may also be displayed with a color-coded background). As a result, participants can grasp not only the overall evaluation but also the detailed fit score for each evaluation axis.

[0136] In the center of Figure 13 is a matching parameter adjustment section, displaying two sliders: skill emphasis and team size preference. Participants can adjust these sliders to customize their preferences, such as whether to prioritize technical skills or values, and whether to prefer a small or large team. The system is designed to reflect the adjustment results in real time on the radar chart and the fit score.

[0137] In the upper right of Figure 13, there is a section for recommended training courses, displaying three courses: Introduction to Microservices Architecture, Healthcare Data Privacy Protection Techniques, and Cloud-Native Security. Each course also includes its level and duration, making it easier for participants to create a learning plan to bridge their skill gaps.

[0138] In the middle right of Figure 13, there is a section for feedback from project veterans, which includes firsthand accounts from two individuals with different roles, such as a senior engineer and a project manager. This feedback allows participants to gain a concrete understanding not only of the technical aspects of the project but also of the actual work environment and cultural aspects. Thus, the output processing unit 105 functions as a feedback display means, displaying feedback from individuals with experience in similar projects.

[0139] At the bottom of Figure 13, the time-series changes in the degree of fit are displayed as a line graph, showing the trends in technical fit and values ​​fit over a nine-month period. This graph allows participants to understand how changes in their skills and values ​​are affecting the matching results, providing them with information to consider the direction of their career development.

[0140] The user interface shown in Figure 13 features a design that goes beyond simply presenting information in numerical and textual formats, making extensive use of visual elements such as radar charts and line graphs to facilitate intuitive understanding. Furthermore, in addition to static information presentation, the inclusion of a parameter adjustment function via sliders enables interactive functionality, allowing participants to actively change conditions and explore optimal matching. As a result, participants can reaffirm their own values ​​and skill balance, leading to more satisfying project selections and future career planning.

[0141] Thus, the output processing unit 105 is equipped with a matching result display means that displays the matching results. The matching result display means quantifies and displays the degree of matching between the participants and domain knowledge related to the project (for example, information obtained from requirements specifications and project plans).

[0142] <User Interface Example 2: Dialogue Results Screen> As another example of a user interface, we will also explain the dialogue screen. Figure 14 shows an example of the UI (User Interface) of the AI ​​dialogue optimization screen. This screen is an interface for extracting values ​​and interest in projects through dialogue between participants and the AI ​​system during the matching process, and for adjusting matching parameters in real time.

[0143] A colored title bar (for example, a blue title bar) may also be placed at the top of the screen in Figure 14. This title bar may display, for example, "Optimization Screen via AI Dialogue" in bold. The username (for example, "Taro Yamada") may be displayed to the right of the title bar so that participants can recognize that it is their own dialogue session.

[0144] In the summary area directly below the title bar, the current matching score is prominently displayed as "82%", and below that, key matching items such as "Cloud Technology: 90%", "Interest in the Medical Field: 85%", and "Team Collaboration Style: 80%" are shown as color-coded tags. Note that the color coding of matching items is omitted in Figure 14. This display allows participants to grasp the matching status at a glance at the start of the conversation.

[0145] In the upper right corner of the screen is a "Values ​​Keywords" section, where keywords such as "technological innovation," "social impact," "autonomy," and "learning orientation" are displayed as colorful, rounded tags. These tags visualize the participants' values ​​and interests extracted through dialogue, and a system is in place to dynamically update them as the dialogue progresses.

[0146] The majority of the screen is occupied by a dialogue interface. This interface displays questions from the AI ​​in a speech bubble on the left (for example, a gray speech bubble) and the participant's answers in a speech bubble on the right (for example, a blue speech bubble), in chronological order. Specifically, the AI ​​asks, "What aspect of this project interests you the most?" and the participant replies, "I'm interested in contributing to society through the use of medical data and gaining practical experience with microservice architecture." The AI ​​then follows up on the initial answer with a more in-depth question, such as, "It seems you prioritize social impact over technical challenges. Is that understanding correct?" and the participant replies, "Yes, technology is important, but ultimately I find it rewarding to create practicality in the medical field and social value."

[0147] At the bottom of the dialogue area, a rounded rectangle (for example, a white rounded rectangle) is placed as an input field, displaying the placeholder text "Enter your message..." and a send button (for example, a blue send button) on the far right. This display allows participants to continue their conversation with the AI ​​in a natural way.

[0148] A radar chart titled "Real-time Fit" is located in the center right of the screen. This chart visualizes the fit across eight axes: technical fit (P1), industry fit (P2), work style (P3), growth opportunities (P4), team fit (P5), social values ​​(P6), career orientation (P7), and compensation (P8).

[0149] A distinctive feature is the display of two types of polygons: dotted and solid. The dotted line represents the initial fit, while the solid line represents the updated fit after the dialogue. A particularly significant improvement is observed on the "social values" axis (P6), indicating that the alignment between participants' values ​​and the project's characteristics improved through the dialogue.

[0150] At the bottom of the screen in Figure 14, color-coded buttons are arranged as dialogue options, such as "I want to know more," "Ask a question from a different perspective," "Clear history," and "Confirm matching." These buttons allow participants to control the direction of the dialogue themselves.

[0151] In the lower right corner of Figure 14, there is a "Parameter Change History" section, which displays a bulleted list of parameter adjustments resulting from the dialogue, such as "Weight of social values ​​+15%", "Weight of technical skills -5%", and "Aptitude for the medical field +10%". This display allows participants to transparently see how their responses are reflected in the matching algorithm.

[0152] At the bottom of Figure 14, under the heading "AI Analysis Memo," a summary analysis is displayed, stating: "There is a strong interest in social impact, and a tendency to prioritize value realization over technical aspects. Due to a strong desire to contribute in the medical field, recommendations for related training will be strengthened. Specific questions regarding microservices will be asked in a follow-up." This analysis presents participants with trends extracted from the entire dialogue and a plan for future action.

[0153] The dialogue screen in Figure 14 is characterized not only by being a simple chat interface, but also by the fact that the degree of fit and parameters change in real time as the dialogue progresses, and the results are visually represented. Participants simply express their values ​​and interests through natural conversation, and the system extracts these values ​​as structured information and reflects it in the matching algorithm. Through this approach, the system can grasp subtle nuances and latent interests that could not be captured by existing fixed questionnaires, resulting in more accurate matching and improved participant satisfaction.

[0154] <Effects of the First Embodiment> The effects of the first embodiment will now be explained. Firstly, compared to existing personnel placement methods based solely on skill matching, the system disclosed in this application can achieve more appropriate matching by multidimensionally analyzing employees' values ​​and motivations and project characteristics. As a result, it is expected that employee motivation will be maintained and work performance will improve.

[0155] Furthermore, the system disclosed in this application utilizes an AI dialogue system to more accurately grasp employees' potential interests and values. Information obtained through direct dialogue can reflect elements that cannot be captured by static data analysis.

[0156] Furthermore, the system disclosed in this application can continuously learn and optimize through a feedback loop in the matching process. By adjusting parameters based on employee responses and dialogue content, the matching accuracy improves over time.

[0157] For employees, increased opportunities to participate in projects that align with their values ​​and career aspirations lead to improved job satisfaction and the realization of their potential. Matching employees with shared values ​​also contributes to higher long-term employee retention.

[0158] On the other hand, for companies, leveraging employees' intrinsic motivation in personnel allocation can be expected to improve the probability of project success and increase overall organizational productivity. Furthermore, objective matching based on data enhances the transparency and fairness of personnel allocation.

[0159] Thus, the system according to the embodiment of the present application integrates previously separate personnel data and project data, and evaluates the compatibility between the two from the perspective of shared values, thereby enabling more sustainable and effective human resource utilization.

[0160] [Modification] A modification of the first embodiment will be described. First, the challenges related to human resource development and skill development within companies will be explained as a prerequisite for the need to modify the first embodiment.

[0161] In the corporate environment targeted by the system disclosed in this application, various projects are running in parallel, and gaps frequently occur between employees' existing skills and the skills required for their assigned tasks. Traditional talent development processes have included regular training programs and self-development support. However, these do not always match actual work needs or individual learning styles, resulting in the challenge that what has been learned is not fully utilized in practical work.

[0162] Furthermore, various data regarding employee skills are generated within companies through daily operations. For example, an employee's actual skill level can be indirectly grasped from their work performance and the quality evaluation of project deliverables. In the internal evaluation system, strengths and areas for improvement are pointed out through feedback from superiors and colleagues, and text-based evaluation comments are accumulated. In addition, the training management system stores past training history and completion test results, and records of officially certified skills are managed.

[0163] From a talent allocation perspective, the typical process involves project managers defining the necessary skill sets based on task requirements and then searching for suitable personnel. However, when a gap exists between the existing skills of the personnel and the task requirements, there are often only two options: either rely on external recruitment because "no suitable person is available," or be forced to assign personnel who lack sufficient skills. In particular, many cases were seen where enthusiastic and capable individuals missed out on growth opportunities due to a lack of current skills.

[0164] In this type of business environment, the system disclosed in this application aims to accurately identify the gap between task requirements and the existing skills of personnel, and to provide personalized training plans to efficiently bridge that gap. As a result, it goes beyond mere "matching of personnel and tasks" to realize a new approach of "creating matches through talent development." Furthermore, it is expected that employees will have more learning opportunities directly related to their work, leading to accelerated career growth and improved motivation.

[0165] <Functionality of this modified example> The modified example according to the first embodiment provides personalized learning opportunities based on matching results by adding a training recommendation function to the dialogue processing unit 104. The dialogue processing unit 104 performs a comparative analysis of value data extracted from conversations with employees and project characteristics to identify skill gaps. For example, from an employee's statement that they are "interested in microservice architecture" in relation to a medical system development project, the dialogue processing unit 104 calculates the difference between the current skill level and the project requirements.

[0166] The dialogue processing unit 104 searches the company's training database for training courses that address the identified skill gaps and generates an optimal training plan that takes into account the employee's learning style and time constraints. The recommended training is presented in a format that includes specific content, difficulty level, and duration, such as "Microservice Architecture Practice Course (Intermediate Level, 4 Weeks)."

[0167] When recommending training, the evaluation considers not only how to fill skill gaps but also how the training aligns with the employee's long-term career aspirations. For example, if a value of "emphasizing social impact" is detected, the dialogue processing unit 104 will include in the recommendation, in addition to technical training, content that matches this value, such as "examples of social contribution through the use of medical data."

[0168] Employee responses to recommended training programs are collected through dialogue, and recommendations are further optimized through questions such as, "Are you interested in this training?" and "Do you prefer online learning or in-person training?". This response data is stored as learning history and used as foundational data for future skill assessments and career development plans.

[0169] After the training is completed, an evaluation process is proposed to verify the acquired skills, and a mechanism is in place to reflect the acquisition of new skills in updating the values ​​vector and skills vector. Through this continuous learning cycle, dynamic matching is achieved that responds to employee growth and the evolution of project requirements.

[0170] <Effects of this Modified Form> One effect of the modified form according to the first embodiment is that it contributes to both the improvement of employee skills and the improvement of the company's project success rate. By providing appropriate training recommendations for the skill gaps identified in the matching process, employees can obtain opportunities to improve their skills in a way that aligns with their own values ​​and interests. This leads to intrinsic motivation that goes beyond mere skill acquisition, resulting in improved learning efficiency and knowledge retention rates.

[0171] Furthermore, considering the alignment between employee values ​​data and training content is expected to increase training participation and completion rates. Compared to traditional, uniform training, providing learning experiences that match individual interests and preferences can encourage a more proactive approach to skill development.

[0172] For companies, this allows them to systematically develop personnel with the necessary skill sets for projects, reducing external recruitment costs and optimizing the use of internal talent. It also makes it easier to secure technical capabilities through internal talent development, particularly in highly specialized technical fields and emerging sectors.

[0173] Furthermore, the use of training data and its effectiveness measurement results to update value vectors and skill vectors contributes to improving the accuracy of the matching engine itself. This enables matching that reflects dynamic elements such as changes in values ​​due to skill acquisition and the discovery of new interests during the learning process, realizing a more long-term perspective on talent utilization.

[0174] For the organization as a whole, visualizing and systematically managing the gap between individual employee growth and project requirements makes it easier to develop strategic talent development plans. This promotes proactive skill development that looks ahead to future business developments and technological trends, supporting the organization's competitiveness and sustainable growth.

[0175] [Second Embodiment] The second embodiment adds a function to the values ​​and motivation-based matching system shown in the first embodiment, allowing employees and managers to dynamically adjust the weighting of values ​​and skills. Figure 15 is a diagram illustrating the concept of this extended system.

[0176] In the first embodiment of the matching engine, the value weight (α) and skill weight (β) were treated as pre-set fixed values. However, in the second embodiment, a slider interface is introduced that allows users to interactively adjust these parameters. This interface enables flexible matching according to organizational needs and individual circumstances.

[0177] <Changes in Device Configuration> Figure 16 is a block diagram showing the changes in the device configuration in the second embodiment. The system (information processing device 10) according to the second embodiment is based on the functional blocks of the first embodiment, with the following extended functions added.

[0178] Specifically, a "user-adjustable parameter management function" has been added to the matching processing unit 103. This component manages the weighting parameters set by the user using sliders and is responsible for appropriately reflecting them in the overall score calculation function. Adjustable parameters include the emphasis on values, the emphasis on skills, the weighting of specific categories (technical / social / industry aspects, etc.), and the setting of thresholds for essential requirements.

[0179] The output processing unit 105 includes a "parameter adjustment UI component." This component provides a visual slider interface and accepts intuitive user operation. The parameter adjustment UI component also implements a mechanism that updates the matching results in real time in response to parameter changes. The parameter adjustment UI component provides a slider that allows adjustment of weighting information.

[0180] Furthermore, a new "authority management function" will be added to control the entire system. The authority management function (authority management department) controls the range of adjustable parameters according to the role of the user, such as project managers, HR departments, and general employees. The authority management function enables flexible customization while maintaining consistency with organizational policies.

[0181] <Changes to the Processing Flow> Figure 17 is a flowchart showing the changes to the processing flow in the second embodiment. Based on the flow of the first embodiment, the following processing steps have been added or modified.

[0182] Step S21, "Parameter Initialization," is added before the matching calculation step. In this step, the initial values ​​of the weighting parameters are set based on the organization's default values ​​or the user's past configuration history.

[0183] After the generation of the first-stage matching results, a new step S22, "Parameter Adjustment Interface Display," is added. In this step, a slider UI is presented to the user, giving them the opportunity to adjust the weighting.

[0184] If the user adjusts the parameters (step S23, Yes branch), step S24 of "Parameter adjustment by user" is executed, and the set values ​​are reflected in the system. Subsequently, step S25 of "Matching recalculation" is immediately executed based on the adjusted parameters, and the results are updated in real time.

[0185] Once the user finishes adjusting the parameters (step S23, No branch), step S26, "Save parameter settings," is executed before the final matching result is confirmed, and the user's adjustments are saved as history. As a result, personalized initial settings will be applied during the next matching.

[0186] <User Interface: Adjustable Slider Screen> Figure 18 shows an example of a user interface equipped with an adjustable slider, which is a core function of the second embodiment. This screen is an extension of the matching results screen of the first embodiment.

[0187] In the lower center of the screen in Figure 18, a large "Matching Parameter Adjustment" section is prominently displayed. This section contains multiple sliders arranged vertically, each corresponding to the adjustment of a different parameter.

[0188] The slider at the top is responsible for adjusting the weighting of "values ​​vs. skills," and users can adjust the balance between the two elements by moving the slider left or right. Initially, it is set to the median (values ​​50%: skills 50%), but it can be moved towards "values-focused" or "skills-focused" depending on the organization's needs. In response to the user's slider operation, a numerical value (e.g., values ​​70%: skills 30%) is dynamically displayed on the right.

[0189] The second slider adjusts the weighting between "technical aspects vs. social aspects," allowing users to set whether to prioritize the technical aspects or the social impact of a project. For example, it's possible to prioritize technical aspects in research and development projects and social aspects in customer service development projects.

[0190] The third slider is responsible for balancing "long-term development vs. immediate contribution," allowing users to set whether to prioritize a person's long-term growth potential or their immediate ability to perform tasks. This adjustment enables different approaches, such as prioritizing immediate contribution in urgent projects and emphasizing development when aiming to strengthen the organization over the long term.

[0191] On the right side of the screen in Figure 18, the "Real-time Matching Score" is displayed as a preview of the adjustment results. Every time the user changes a parameter, the Real-time Matching Score is instantly recalculated, and the radar chart and overall score on the left side of the screen, as well as the score breakdown for each category, are updated. As a result, users can immediately see how their adjustments affect the matching results.

[0192] Furthermore, to visualize score fluctuations, the values ​​before and after adjustment are displayed along with the amount of change, such as "82% → 86%". Items that have changed particularly significantly are highlighted, allowing users to intuitively understand the areas where the parameter adjustments have a significant impact.

[0193] On the right side of the screen in Figure 18, there is a "Adjustment History" section where past parameter setting patterns are saved. By selecting named setting presets such as "Settings for Project A," "Training-Focused Settings," and "Technology-Specific Settings," users can easily recall frequently used parameter sets.

[0194] Additionally, operation buttons such as "Save," "Reset," and "Restore to Defaults" are provided, offering the ability to save the current settings as a named preset or revert to the initial state.

[0195] <Usage Scenarios> The adjustable slider function according to the second embodiment can be used in a variety of scenarios, such as the following.

[0196] Project managers can adjust parameters according to the nature of the project. For example, a project manager can prioritize "interest in technological challenges" in an innovative research and development project, and increase the weight of "social values" in a customer-focused service development project.

[0197] The HR department can manage default settings in line with the organization's overall policies. For example, by adjusting the balance between "long-term talent development" and "utilization of immediately productive employees" according to the organization's growth phase, strategic talent allocation can be achieved.

[0198] Regular employees can adjust parameters to match their career aspirations. For example, if a regular employee wishes to challenge themselves in a new technological field, they can lower the weight of "skill matching" and increase the weight of "growth opportunities," allowing them to receive matching that takes into account not only their current skill set but also their potential for growth.

[0199] In inter-organizational personnel exchanges, parameter adjustments may be made to reflect differences in departmental culture and evaluation criteria. For example, settings tailored to departmental characteristics may be selected, such as emphasizing technical expertise in the research and development department and communication skills in the sales department.

[0200] <Technical Implementation Details> The slider's internal processing converts the user's actions into a numerical value in the range of 0.0 to 1.0, and applies this value as a weighting coefficient to the matching algorithm. For example, if the "Values ​​vs. Skills" slider is at position 0.7, a weight of 0.7 is assigned to value similarity and a weight of 0.3 is assigned to skill suitability.

[0201] Parameter changes are immediately sent to the matching engine, and recalculations are performed in the background. During this process, instead of reanalyzing all the data, an optimization process is applied that only modifies the weight coefficients of the already calculated similarity matrix, resulting in real-time response.

[0202] Adjustment history is stored in conjunction with user profiles and serves as material for learning each user's preferences and tendencies. Furthermore, statistical analysis of these adjustment patterns can reveal changes in values ​​and important factors within the organization.

[0203] <Effects of the Second Embodiment> The adjustable slider function of the second embodiment can be expected to have the following effects.

[0204] Firstly, the transparency and controllability of matching results will improve. Users will have a clear understanding of the criteria used for matching and will be able to adjust the criteria as needed, leading to greater satisfaction with the results and a greater sense of ownership.

[0205] Secondly, it enables flexible matching that meets the needs of both organizations and individuals. By adjusting parameters according to project characteristics, organizational circumstances, and individual career aspirations, it becomes possible to apply diverse, non-uniform matching criteria.

[0206] Thirdly, the matching process offers learning benefits. By experiencing the relationship between parameter adjustments and their results, users can learn how their values ​​and skills are evaluated and what types of projects they are most compatible with.

[0207] Fourthly, dynamic matching becomes possible in response to the passage of time and changes in circumstances. By adjusting parameters according to each stage of a career and the phase of an organization, continuous matching optimization that responds to growth and change is achieved, rather than static matching.

[0208] Finally, it becomes easier to visualize and implement the organization's talent strategy. Since the talent utilization policies set forth by management and the HR department can be implemented as specific parameter settings, it becomes possible to bridge the gap between abstract policies and actual talent allocation.

[0209] Thus, the second embodiment achieves more adaptive and personalized talent matching by adding a user-driven adjustment function to the matching system of the first embodiment.

[0210] Next, we will describe the hardware of each device that makes up the information processing system. Figure 19 is a diagram showing an example of the hardware configuration of the information processing device 10.

[0211] The information processing device 10 can be configured using a so-called computer and has the configuration illustrated in Figure 19. For example, the information processing device 10 includes a processor 311, memory 312, input / output interface 313, and communication interface 314, etc. The components of the processor 311, etc. are connected by an internal bus or the like and are configured to communicate with each other.

[0212] However, the configuration shown in Figure 19 is not intended to limit the hardware configuration of the information processing device 10. The information processing device 10 may include hardware not shown, and it may not have to include the input / output interface 313 if necessary. Furthermore, the number of processors 311 etc. included in the information processing device 10 is not limited to the example shown in Figure 19; for example, multiple processors 311 may be included in the information processing device 10.

[0213] The processor 311 is a programmable device such as a CPU (Central Processing Unit), MPU (Micro Processing Unit), DSP (Digital Signal Processor), TPU (Tensor Processing Unit), or GPU (Graphics Processing Unit). Alternatively, the processor 311 may be a device such as an FPGA (Field Programmable Gate Array) or ASIC (Application Specific Integrated Circuit). The processor 311 executes various programs, including an operating system (OS).

[0214] Memory 312 can be RAM (Random Access Memory), ROM (Read Only Memory), HDD (Hard Disk Drive), SSD (Solid State Drive), etc. Memory 312 stores the OS program, application programs, and various data.

[0215] The input / output interface 313 is an interface for a display device or input device (not shown). The display device is, for example, a liquid crystal display. The input device is, for example, a device that accepts user input such as a keyboard or mouse.

[0216] The communication interface 314 is a circuit, module, etc., that communicates with other devices. For example, the communication interface 314 may include a NIC (Network Interface Card).

[0217] The functions of the information processing device 10 are realized by various processing modules. These processing modules are realized, for example, by the processor 311 executing a program stored in the memory 312. The program can also be recorded on a computer-readable storage medium. The storage medium can be a non-transitory medium such as a semiconductor memory, hard disk, magnetic recording medium, or optical recording medium. In other words, the present invention can also be embodied as a computer program product. Furthermore, the program can be downloaded via a network or updated using the storage medium on which the program is stored. Moreover, the processing module may be realized by a semiconductor chip.

[0218] The terminal 20 can also be configured to include a computer, similar to the information processing device 10, and its basic hardware configuration is no different from that of the information processing device 10, so its explanation will be omitted.

[0219] The information processing device 10 is equipped with a computer, and its functions can be realized by having the computer execute a program. Furthermore, the information processing device 10 executes control methods and information processing methods based on the program.

[0220] [Modification] Note that the configuration and operation of the information processing system described in the above embodiment are illustrative examples and are not intended to limit the system configuration.

[0221] The information processing device 10 may be a server on the cloud. Alternatively, the information processing device 10 may be an on-premise server managed and operated by a company or other organization within its own facilities. Alternatively, the functions of the information processing device 10 (human resource matching system) may be implemented by multiple devices.

[0222] In the flowcharts (sequence diagrams) used in the above description, multiple processes are shown in order, but the execution order of the processes performed in the embodiment is not limited to the order in which they are shown. In the embodiment, the order of the illustrated processes can be changed to the extent that it does not impede the content, for example, by executing each process in parallel.

[0223] The embodiments described above are explained in detail to facilitate understanding of the disclosure, and it is not intended that all the configurations described above are necessary. Furthermore, when multiple embodiments are described, each embodiment may be used individually or in combination. For example, it is possible to replace parts of the configuration of one embodiment with those of another embodiment, or to add configurations from other embodiments to the configuration of one embodiment. In addition, it is possible to add, delete, or replace parts of the configuration of one embodiment with those of another.

[0224] As described above, the industrial applicability of the present invention is clear, and it is particularly suitable for systems that match employees with projects.

[0225] Some or all of the above embodiments may also be described as follows, but are not limited to the following:

[0226] [Note 1] A human resource matching system comprising: a value extraction means for extracting value information from the participant's work history data; a characteristic extraction means for extracting project information that indicates the characteristics of a project; a calculation means for calculating the degree of fit between the value information and the project information; a response means for generating a matching result between the participant and the project based on the degree of fit; and a dialogue means for confirming the participant's level of interest through an interactive interface using a machine learning model based on the matching result, wherein the response means determines the matching result based on weighting information between the participant's skills and the value information.

[0227] [Note 2] The value extraction means is the talent matching system described in Note 1, which generates skill vectors and value vectors from the participant's chat logs and work history.

[0228] [Note 3] The characteristic extraction means is the personnel matching system described in Note 1 or 2, which expresses the project information along three axes: technical, social, and industry.

[0229] [Appendix 4] The dialogue means is a talent matching system according to any one of Appendix 1 to 3, wherein the dialogue means checks the degree of interest of the participants in stages and updates the weighting information based on the check results.

[0230] [Appendix 5] The personnel matching system according to any one of Appendix 1 to 4, further comprising a matching result display means for displaying the matching results, wherein the matching result display means displays the degree of matching between the participant and the domain knowledge relating to the project in numerical form.

[0231] [Appendix 6] The personnel matching system according to any one of Appendix 1 to 5, comprising a slider that can adjust the weighting information.

[0232] [Appendix 7] The personnel matching system described in any one of the appendices 1 to 6, further comprising a training recommendation means for recommending training courses necessary for the participant based on the matching results.

[0233] [Appendix 8] A personnel matching system according to any one of the appendices 1 to 7, further comprising a feedback display means for displaying feedback from individuals with experience in similar projects.

[0234] [Note 9] The talent matching system described in any one of Notes 1 to 8, wherein the value information includes information regarding the participants' preferred working styles, and the matching result is determined based on the preferred working styles.

[0235] [Note 10] A computer-readable storage medium that stores a program which causes a computer to perform the following steps: a value extraction step of extracting value information from the participant's work history data; a characteristic extraction step of extracting project information that indicates the characteristics of the project; a calculation step of calculating the degree of fit between the value information and the project information; a correspondence step of generating a matching result between the participant and the project based on the degree of fit; and a dialogue step of confirming the degree of interest of the participant through an interactive interface using a machine learning model based on the matching result, wherein the correspondence step determines the matching result based on weighting information between the participant's skills and the value information.

[0236] Furthermore, some or all of the configurations described in Appendices 2 to 9, which are dependent on Appendice 1 above, may also be dependent on Appendice 10 in the same way as those described in Appendices 2 to 9. Moreover, not limited to Appendices 1 and 10, some or all of the configurations described as appendices may also be dependent on various hardware, software, various recording means for recording software, or systems, without departing from the embodiments described above.

[0237] Furthermore, each disclosure of the above-mentioned prior art documents cited herein is incorporated herein by reference. Although embodiments of the present invention have been described above, the present invention is not limited to these embodiments. It will be understood by those skilled in the art that these embodiments are merely illustrative and that various modifications are possible without departing from the scope and spirit of the present invention. That is, the present invention naturally includes the entire disclosure, including the claims, and various modifications and alterations that can be made by those skilled in the art in accordance with the technical idea.

[0238] 10 Information processing device 11 Value extraction means 12 Characteristic extraction means 13 Calculation means 14 Corresponding means 15 Dialogue means 20 Terminal 100 Communication control unit 101 Input processing unit 102 Analysis processing unit 103 Matching processing unit 104 Dialogue processing unit 105 Output processing unit 106 Storage unit 311 Processor 312 Memory 313 Input / output interface 314 Communication interface

Claims

1. A human resource matching system comprising: a value extraction means for extracting value information from a participant's work history data; a characteristic extraction means for extracting project information that indicates the characteristics of a project; a calculation means for calculating the degree of fit between the value information and the project information; a response means for generating a matching result between the participant and the project based on the degree of fit; and a dialogue means for confirming the participant's level of interest through an interactive interface using a machine learning model based on the matching result, wherein the response means determines the matching result based on weighting information between the participant's skills and the value information.

2. The talent matching system according to claim 1, wherein the value extraction means generates a skill vector and a value vector from the participant's chat log and work history.

3. The talent matching system according to claim 1 or 2, wherein the characteristic extraction means expresses the project information along three axes: technical, social, and industry.

4. The talent matching system according to any one of claims 1 to 3, wherein the dialogue means progressively checks the level of interest of the participants and updates the weighting information based on the check results.

5. The personnel matching system according to any one of claims 1 to 4, further comprising a matching result display means for displaying the matching results, wherein the matching result display means displays the degree of matching between the participant and the domain knowledge relating to the project in numerical form.

6. The personnel matching system according to any one of claims 1 to 5, further comprising a slider for adjusting the weighting information.

7. The personnel matching system according to any one of claims 1 to 6, further comprising a training recommendation means for recommending training courses necessary for the participant based on the matching results.

8. The personnel matching system according to any one of claims 1 to 7, further comprising a feedback display means for displaying feedback from individuals with experience in similar projects.

9. The talent matching system according to any one of claims 1 to 8, wherein the value information includes information regarding the participant's preferred working style, and the matching result is determined based on the preferred working style.

10. A computer-readable storage medium that stores a program which causes a computer to perform the following steps: a values ​​extraction step for extracting values ​​information from a participant's work history data; a characteristics extraction step for extracting project information that indicates the characteristics of a project; a calculation step for calculating the degree of fit between the values ​​information and the project information; a correspondence step for generating a matching result between the participant and the project based on the degree of fit; and a dialogue step for confirming the participant's level of interest through an interactive interface using a machine learning model based on the matching result, wherein the correspondence step determines the matching result based on weighting information between the participant's skills and the values ​​information.

Images