A game industry talent ability evaluation method and system

Abstract

This invention relates to the field of game assessment technology, and more particularly to a method and system for assessing the abilities of talent in the game industry. The method divides task stages based on a basic profile of the target candidate, identifies key observation points, and deploys data acquisition terminals. It collects real-time data on the timing of operational inputs and deviations in skill release, analyzes and extracts input frequency and operational deviation features, and constructs a feature-based time-lapse curve. Subsequently, it uses continuous wavelet transform to perform time-frequency analysis on the time-lapse information, generating operational wavelet features including time-frequency distribution maps and power spectral density maps. The assessment model constructed from these features is then quantitatively calculated, outputting a practical ability rating. Based on the rating results, job matching feedback is implemented. This method overcomes the limitations of traditional static statistics by deeply mining the dynamic time-frequency features of operational behavior, achieving a scientific quantitative assessment of the rhythm, stability, and accuracy of talent's micro-operations, significantly improving the objectivity and accuracy of game talent selection and job matching.

Description

Technical Field

[0001] This invention relates to the field of game evaluation technology, and in particular to a method and system for evaluating the abilities of talent in the game industry. Background Technology

[0002] With the booming development of the esports industry and the increasing demands for industrialized game production, the gaming industry's need for high-level, professional talent is growing. Especially in scenarios such as esports team selection, game testing recruitment, and professional player ability assessment, how to scientifically and objectively measure candidates' actual operational abilities has become a crucial aspect of human resource selection. Traditional talent assessment methods often rely on subjective observation, rank ratings, or simple operation counting (such as APM). These methods are insufficient to comprehensively reflect a player's core competencies in complex combat environments, such as micro-management skills, reaction stability, skill release precision, and operational rhythm.

[0003] In existing technologies, some evaluation systems attempt to perform quantitative analysis by recording players' input frequency or skill hit rate. However, their evaluation dimensions are singular, typically focusing only on the statistical average of static indicators and ignoring the dynamic changes in operational behavior over time. For example, two input behaviors with the same frequency might be different: one is a high-quality micro-operation with a stable rhythm, while the other is an invalid click in a panicked state. Traditional methods struggle to distinguish between the two. Furthermore, existing systems lack the ability to deeply mine the time-frequency characteristics hidden in operational signals (such as explosiveness, periodicity, and response delay), leading to discrepancies between evaluation results and actual competitive performance.

[0004] The above content is only used to help understand the technical solution of the present invention and does not represent an admission that the above content is prior art. Summary of the Invention

[0005] The main objective of this invention is to provide a method and system for assessing the capabilities of talent in the gaming industry. This aims to solve the technical problem that existing game talent assessment methods rely on a single static statistical indicator and lack in-depth analysis of the dynamic characteristics of operational behavior, resulting in difficulty in accurately distinguishing operational rhythm and quality, and significant deviations in assessment results.

[0006] To achieve the above objectives, the present invention provides a method for assessing the capabilities of talent in the game industry, the method comprising: Acquire basic profile data of the target assessment object, divide the task into stages according to the preset assessment checkpoint system, determine the key observation point set, and deploy the practical data collection terminal; Start the practical data acquisition terminal to collect real-time game operation data of the target evaluation object. The real-time game operation data includes operation input timing data and skill release deviation data. The real-time game operation data is analyzed, the input frequency features and operation deviation features of multiple observation points in the key observation point set are extracted, and the input frequency duration curve and deviation feature duration curve are constructed, and the output is the feature duration information; Based on continuous wavelet transform, time-frequency analysis is performed on the feature duration information to obtain operation wavelet feature information, wherein the operation wavelet feature information includes operation time-frequency distribution feature map and operation power spectral density map; A game talent operation ability assessment model is constructed. The operation wavelet feature information is used as input, and the game talent operation ability assessment model is run to perform quantitative assessment of the ability and output practical ability rating information. Based on the practical skills rating information, the game job matching feedback for the target assessment subjects is carried out.

[0007] Optionally, the step of acquiring basic profile data of the target assessment object, dividing the task into stages according to a preset assessment checkpoint system, determining the key observation point set, and deploying a practical data acquisition terminal includes: Obtain the resume dataset of the target assessment object, extract the types of game characters that the target assessment object is good at and the operation level characteristics of the target assessment object, and output the target object profile data; Based on the knowledge graph of game operation ability assessment, and combined with the types of game characters that the target assessment subjects are good at, multiple test task segments are determined; Based on the operational segment characteristics of the target evaluation object and combined with the accuracy requirements of the capability evaluation, key observation points are distributed in multiple test task segments to determine the set of key observation points. Based on the set of key observation points, a data acquisition module is configured to form the practical data acquisition terminal, wherein the data acquisition module includes a keyboard, mouse or gamepad input monitoring component and a game client performance acquisition interface.

[0008] Optionally, the step of parsing the real-time game operation data, extracting the input frequency features and operation deviation features of multiple observation points in the key observation point set, and constructing the input frequency duration curve and the deviation feature duration curve, outputting feature duration information, including: Based on the real-time game operation data, the first real-time operation data of the first observation point is extracted, and the first real-time operation data includes the first operation input frequency and the first skill release deviation. The frequency of the first operation input is analyzed, the operation input acceleration is extracted, and based on the timestamp of the first real-time operation data, an input acceleration-time curve is constructed, and the output is the first input feature duration curve; Analyze the first skill release deviation, obtain the operation deviation amplitude, and construct a deviation amplitude-time curve based on the timestamp of the first real-time operation data, outputting the first deviation feature duration curve; By traversing multiple observation points in the key observation point set, feature extraction is performed based on the real-time game operation data to obtain multiple sets of input feature duration curves and deviation feature duration curves, which are then stored as feature duration information.

[0009] Optionally, the step of performing time-frequency analysis on the feature duration information based on continuous wavelet transform to obtain the operational wavelet feature information includes: Based on the target object profile data, extract the baseline operation frequency information of this type of game character; Using the Morlet wavelet as the target wavelet mother function and the aforementioned reference operating frequency information as the scale constraint, the wavelet transform parameters are initialized. Perform a continuous wavelet transform on the first input feature duration curve to obtain the first input wavelet coefficient matrix; perform a continuous wavelet transform on the first deviation feature duration curve to obtain the first deviation wavelet coefficient matrix; The first input wavelet coefficient matrix and the first deviation wavelet coefficient matrix are combined and stored as the first wavelet analysis data set. The feature duration information is traversed to perform continuous wavelet transform, and multiple wavelet analysis data groups are obtained and stored as wavelet analysis datasets.

[0010] Optionally, the step of performing time-frequency analysis on the feature duration information based on continuous wavelet transform to obtain the operated wavelet feature information further includes: Based on the wavelet analysis data set, the wavelet energy density is calculated to obtain an energy density matrix set, which includes an input energy density matrix and a bias energy density matrix. Based on the energy density matrix set, a time-frequency spectrum set is plotted with time as the first coordinate axis and the scale constraint as the second coordinate axis; Integrate the first wavelet analysis data set using the modulus square of the wavelet coefficient matrix to obtain the power spectrum set; The time-frequency spectrum set and the power spectrum set are stored to obtain operational wavelet feature information.

[0011] Optionally, the construction of the game talent operation ability assessment model further includes: Collect historical evaluation records from game industry evaluation experts to obtain a basic evaluation sample set; Based on industry big data and combined with target game character characteristic data, an expanded evaluation sample set is generated; By merging the basic assessment sample set and the expanded assessment sample set, a training sample set is obtained, wherein the training sample set includes sample wavelet feature information, sample operation rhythm component information, sample operation complexity order information, and sample practical ability level. Based on the training sample set, construct and train the game talent operation ability assessment model.

[0012] Optionally, the step of performing game job matching feedback for the target assessment object based on the practical ability rating information includes: The practical ability rating information includes specific ability scores corresponding to multiple key observation points, and the multiple key observation points are marked with corresponding ability dimension labels. Based on the hiring standards of game companies, a talent response matrix is ​​defined, which includes multiple sets of ability scoring ranges and related job recruitment or training suggestions. Based on the talent response matrix, the practical ability rating information is traversed and matched, corresponding suggested solutions are matched, and the evaluation results of the target assessment object are output.

[0013] Furthermore, to achieve the above objectives, the present invention also provides a talent competency assessment system for the game industry, the system comprising: The profile preparation module is used to acquire basic profile data of the target assessment object, divide the task into stages according to the preset assessment checkpoint system, determine the key observation point set, and deploy the practical data acquisition terminal. The practical data acquisition module is used to start the practical data acquisition terminal and collect the real-time game operation data of the target evaluation object. The real-time game operation data includes operation input timing data and skill release deviation data. The feature curve module is used to parse the real-time game operation data, extract the input frequency features and operation deviation features of multiple observation points in the key observation point set, construct the input frequency duration curve and the deviation feature duration curve, and output the feature duration information. The wavelet time-frequency module is used to perform time-frequency analysis on the feature duration information based on continuous wavelet transform to obtain operation wavelet feature information, wherein the operation wavelet feature information includes operation time-frequency distribution feature map and operation power spectral density map. The model rating module is used to construct a game talent operation ability assessment model. Taking the operation wavelet feature information as input, the game talent operation ability assessment model is run to perform a quantitative assessment of the ability and output practical ability rating information. The matching feedback module is used to perform game job matching feedback for the target assessment object based on the practical ability rating information.

[0014] In addition, to achieve the above objectives, the present invention also provides a talent competency assessment device for the game industry, the device comprising: a memory, a processor, and a talent competency assessment program for the game industry stored on the memory and executable on the processor, the talent competency assessment program for the game industry configured to implement the steps of the talent competency assessment method for the game industry as described above.

[0015] In addition, to achieve the above objectives, the present invention also provides a computer-readable storage medium storing a game industry talent competency assessment program, which, when executed by a processor, implements the steps of the game industry talent competency assessment method as described above.

[0016] This invention provides a method for assessing talent in the gaming industry. By introducing continuous wavelet transform for time-frequency analysis, this method overcomes the limitations of existing technologies that rely solely on static statistical indicators. It accurately captures the dynamic rhythmic characteristics of operational behaviors (such as explosive micro-operations and periodic operation patterns), effectively distinguishing between high-quality micro-operations with stable rhythms and ineffective clicks in a panicked state, significantly improving the accuracy and discriminative power of the assessment. Furthermore, it innovatively constructs a dual-dimensional analysis system of input frequency duration curves and deviation characteristic duration curves. This system not only records operation frequency but also quantitatively analyzes the temporal changes in operational acceleration and skill release deviations, comprehensively reflecting the player's reaction. The assessment focuses on core competency dimensions such as speed, operational stability, and skill precision. It transforms game operation behaviors into visualized time-frequency distribution feature maps and operation power spectral density maps, making abstract operational capabilities concrete and measurable. This provides objective and interpretable data support for talent evaluation, significantly reducing biases caused by subjective judgments in traditional assessments. Based on basic profile data, the assessment's task-stage division and key observation point settings ensure a high degree of alignment between assessment content and target job requirements. Through intelligent correlation between practical ability rating information and game company hiring standards, it achieves precise talent recommendation from rank rating to job fit, improving recruitment efficiency and talent retention rates. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating an embodiment of the game industry talent competency assessment method of the present invention; Figure 2 This is a structural block diagram of an embodiment of the talent competence assessment system for the game industry of the present invention.

[0018] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0020] Reference Figure 1 , Figure 1 This is a flowchart illustrating an embodiment of the game industry talent competency assessment method of the present invention, which presents an embodiment of the game industry talent competency assessment method of the present invention.

[0021] In one embodiment, the method for assessing the capabilities of talent in the gaming industry includes: Step S100: Obtain basic profile data of the target assessment object, divide the task into stages according to the preset assessment checkpoint system, determine the key observation point set, and deploy the practical data collection terminal.

[0022] The target assessment subjects can be individuals participating in talent competency assessments in the gaming industry, such as esports player candidates or game tester applicants. They can serve as the primary source of assessment data, and their actions constitute the basis for evaluation. Basic profile data can be a dataset describing the basic attributes and background information of the target assessment subjects. It can be used to drive task phase division and key observation point setting, aligning assessment content with job requirements. In this embodiment, basic profile data can be obtained through questionnaires, authorization of historical game accounts, or third-party platform interfaces. For example, basic profile data may include, but is not limited to, one or more of the following: player historical rank data, preferred heroes / characters, and game duration distribution. The preset assessment level system can be a standardized set of assessment scenarios designed for specific game types or positions. It can provide a structured and reproducible operational environment, ensuring consistency in assessment conditions. Furthermore, the preset assessment level system may include, but is not limited to, MOBA laning suppression levels, FPS fixed-point reaction levels, and RTS resource scheduling levels.

[0023] The task phase can be a time or logical interval divided according to game progress or tactical objectives within the assessment level. It can be used to define the effective time window for key observation points and focus on the evaluation of specific ability dimensions. In an exemplary embodiment, the task phase can be segmented from the preset assessment level system after matching the job ability model with basic profile data. The key observation point set can be a set of several operation events or time points with evaluation significance preset within the task phase. It can be used to guide the practical data acquisition terminal to focus on data recording during high information density periods. In a specific embodiment, the key observation point set can mark high-value interaction nodes in the level script according to game mechanics and job ability requirements. For example, the key observation point set can include, but is not limited to, one or more of the following: skill release judgment frame, vision switching trigger point, team battle start time, etc. The practical data acquisition terminal can be a hardware and software system deployed in the assessment environment to capture the game operation behavior data of the target assessment object. It can be used to achieve lossless acquisition of operation input timing data and skill release deviation data. Furthermore, the practical data acquisition terminal can capture the operation flow in real time through game client API, input device driver layer, or screen recording + motion recognition technology. In one specific embodiment, the hands-on data acquisition terminal can receive trigger signals from key observation point sets, activate high-precision sampling within a specified task phase, and output real-time game operation data to subsequent steps.

[0024] Obtaining basic profile data of the target assessment subjects can be achieved by extracting identity and game background information from user-submitted materials or authorized interfaces, thus providing a personalized basis for subsequent task phase division. Dividing the task phases according to a pre-set assessment level system can be done by dividing the complete assessment level into several logical phases based on tactical objectives or time slices, thereby focusing the assessment on specific game scenarios related to job skills. Identifying key observation point sets and deploying practical data collection terminals can be done by marking high-value operational nodes within the task phase and configuring collection terminals to monitor these nodes, thereby improving the targeting and information density of data collection.

[0025] Step S200: Start the practical data acquisition terminal to collect real-time game operation data of the target evaluation object. The real-time game operation data includes operation input timing data and skill release deviation data.

[0026] Real-time game operation data can be the raw operation behavior records generated by the target evaluation object during the evaluation process. It can be used as the raw input for feature extraction and includes timestamps and spatial coordinate information. For example, real-time game operation data can include, but is not limited to, one or more of the following: mouse movement trajectory sequences, keyboard key event streams, and touch screen swipe vector sequences. Operation input timing data can be sequence data recording the time and frequency of operation commands. It can be used to calculate input frequency features, reflecting operation rhythm and response speed. Skill release deviation data can be spatial error data between the actual skill release position and the ideal target position. It can be used to calculate operation deviation features, measuring skill accuracy and control stability. Starting the practical data acquisition terminal can activate the data capture module and begin recording the target evaluation object's operation behavior, thus achieving the technical effect of entering the data acquisition state and preparing to acquire real-time game operation data. Acquiring the target evaluation object's real-time game operation data can be achieved by obtaining timestamped operation events through underlying input listening or game log parsing, thereby obtaining the raw operation input timing data and skill release deviation data.

[0027] Step S300: Analyze the real-time game operation data, extract the input frequency features and operation deviation features of multiple observation points in the key observation point set, and construct the input frequency duration curve and the deviation feature duration curve, outputting the feature duration information.

[0028] The input frequency feature can be a quantitative indicator of the number of operation commands per unit time and its rate of change, which can be used to characterize the intensity and rhythm fluctuation of operations, and is the basis for constructing the input frequency duration curve. The operation deviation feature can be a statistical measure of the skill release position error and its time derivative (such as the deviation rate of change), which can be used to reflect the stability of operation accuracy, and is the basis for constructing the deviation feature duration curve. The input frequency duration curve can be a continuous function curve plotted with time as the horizontal axis and input frequency as the vertical axis, which can be used to visualize the time-varying characteristics of operation rhythm and reveal patterns such as explosive micro-operations or operation stagnation. In an exemplary embodiment, the input frequency duration curve and the deviation feature duration curve can together constitute feature duration information for continuous wavelet transform processing. The deviation feature duration curve can be a continuous function curve plotted with time as the horizontal axis and operation deviation feature value as the vertical axis, which can be used to characterize the changing trend of skill release accuracy over time and identify control instability in a state of panic. Furthermore, the deviation feature duration curve and the input frequency duration curve can form a two-dimensional analysis system, complementarily reflecting operation quality.

[0029] Feature duration information can be a dual-channel time-series data structure composed of input frequency duration curves and deviation feature duration curves. This structure can be used as input for continuous wavelet transforms, carrying dynamic evolution information of operational behavior. Parsing real-time game operation data involves cleaning, aligning, and structuring the raw operation events, thus providing a data foundation with a unified format and noise suppression for feature extraction. Extracting input frequency features and operational deviation features from multiple observation points in a key observation point set involves statistically analyzing the operation frequency and its derivatives within the time window of the key observation points, and calculating the mean and variance of skill release errors, thereby generating a quantified feature sequence that can be used to construct duration curves. Constructing the input frequency duration curve and deviation feature duration curve involves interpolating and smoothing the extracted features in chronological order to form a continuous function curve, thus establishing a dual-dimensional time-series representation system and preserving dynamic evolution information of operations. The output is feature duration information, which can be achieved by encapsulating the two duration curves into a unified data structure, providing standardized input for continuous wavelet transforms.

[0030] Step S400: Based on continuous wavelet transform, perform time-frequency analysis on the feature duration information to obtain operation wavelet feature information, wherein the operation wavelet feature information includes operation time-frequency distribution feature map and operation power spectral density map.

[0031] Continuous wavelet transform can be a mathematical tool for time-frequency analysis. By scaling and translating mother wavelets, it performs multi-scale decomposition of signals, allowing for the simultaneous analysis of local features of operational behavior in both the time and frequency domains, and the extraction of dynamic rhythm patterns. In a specific embodiment, continuous wavelet transform can employ complex-valued or real-valued wavelet basis functions such as Morlet and Mexican Hat to convolve the feature duration information. The operational wavelet feature information can be a time-frequency domain feature representation obtained after continuous wavelet transform, which can be used as input to a game talent operation ability assessment model, carrying high-order operation pattern information. Furthermore, the operational wavelet feature information can include, but is not limited to, one or more of the following: operational time-frequency distribution feature map, operational power spectral density map, and wavelet coefficient energy distribution matrix. The operational time-frequency distribution feature map can be a two-dimensional heatmap with time as the horizontal axis, frequency as the vertical axis, and wavelet coefficient amplitude as color intensity, which can be used to visually display the instantaneous frequency components and duration of the operation rhythm, such as periodic clicks or sudden combos.

[0032] The operation power spectral density map can be seen as the energy distribution of the operation signal across various frequency components, and can be used to quantify the energy proportion of different operation rhythm modes (such as stable high-frequency micro-operation vs. random low-frequency clicks). Based on continuous wavelet transform, time-frequency analysis of the characteristic duration information can be performed by applying continuous wavelet transform to each duration curve and calculating wavelet coefficients at each scale. In an exemplary embodiment, this operation can be decomposed into time and frequency components using complex Morlet wavelets, preserving phase information to identify periodicity; or edge detection analysis can be performed using real Mexican Hat wavelets to highlight the operation burst points, thereby achieving the technical effect of overcoming the limitations of static indicators and capturing the instantaneous frequency characteristics of the operation rhythm. Obtaining the operation wavelet feature information can be achieved by extracting statistical quantities such as amplitude, energy, or entropy from the continuous wavelet transform results, thereby generating a high-dimensional, interpretable time-frequency feature representation.

[0033] Step S500: Construct a game talent operation ability assessment model. Using operation wavelet feature information as input, run the game talent operation ability assessment model to perform quantitative assessment of ability and output practical ability rating information.

[0034] The game talent operational ability assessment model can be a multi-dimensional ability quantification model built based on machine learning, which can be used to map operational wavelet feature information into structured practical ability rating information. In a specific embodiment, the game talent operational ability assessment model can be trained under supervision using operational wavelet feature information labeled with expert scores, and can adopt random forest, XGBoost, or neural network architecture. For example, the game talent operational ability assessment model can include, but is not limited to, one or more of the following: reaction speed sub-model, operational stability sub-model, and skill accuracy sub-model. The practical ability rating information can be a quantitative score result of various core operational abilities of the target assessment object, which can be used as the basis for job matching feedback, replacing traditional rank labels to achieve accurate recommendations. Constructing the game talent operational ability assessment model can be done by training a regression or classification model using historical assessment data, with operational wavelet feature information as input and ability score as output. Furthermore, this operation can be achieved by using a gradient boosting tree model to explain the contribution of each ability dimension using feature importance; or by using a multi-task neural network to simultaneously predict the three sub-dimensions of reaction speed, stability, and accuracy, thereby achieving the technical effect of mapping from abstract operational features to structured ability ratings. Using operational wavelet feature information as input, a game talent operational ability assessment model is run to quantitatively evaluate capabilities. This can be achieved by inputting the operational wavelet feature information of the test subject into a trained model and performing forward inference, thereby outputting objective and quantitative practical ability rating information. The output of practical ability rating information can be achieved by formatting the model inference results into JSON or a structured report, thus providing capability data that can be parsed by downstream systems.

[0035] Step S600: Based on the practical skills rating information, perform game job matching feedback for the target assessment subjects.

[0036] The game-related job matching feedback can be an adaptation suggestion generated based on the comparison between practical ability rating information and corporate hiring standards. This can be used to output a person-job matching report or a recommended job list, improving recruitment decision-making efficiency. Based on practical ability rating information, the game-related job matching feedback for the target assessment object can be performed by matching the practical ability rating information with the company's preset job ability thresholds. In an exemplary embodiment, this operation can be achieved by calculating Euclidean distance or cosine similarity to output a list of the closest positions; or by using a rule engine to determine whether the hard conditions of a specific position (such as the high stability and low bias required for an FPS sniper) are met, thereby achieving a closed-loop technical effect from ability data to recruitment decisions.

[0037] Taking the selection of support players for MOBA professional teams as an example, the game industry talent ability assessment method in this embodiment can be as follows: a target assessment candidate submits their historical rank and commonly used hero data (basic profile data). Based on this, the system selects the "bottom lane protection" level from the preset assessment level system and divides it into three task stages: "laning phase," "small-scale team fight phase," and "baron contest phase." Within the "laning phase," the "time of enemy dash skill release" is set as the key observation point set. The practical data acquisition terminal records the frequency of mouse clicks and the deviation of healing skill landing points during this period. The system constructs an input frequency duration curve to show that the candidate has high-frequency stable operations at the moment of enemy dash, and the deviation feature duration curve shows that the skill landing point error is less than the threshold. After continuous wavelet transform, the operation time-frequency distribution feature map shows a continuous energy peak in the 2-4Hz frequency band, indicating that the candidate has periodic precise micro-operation ability. Based on this, the game talent operation ability assessment model outputs high "reaction speed" and "skill accuracy" scores. Finally, the system recommends that the candidate is suitable for the "protective support" position, rather than making a generalized judgment based on their diamond rank.

[0038] For example, in the scenario of recruiting an FPS game test engineer, the talent assessment method for the game industry in this embodiment can be as follows: the applicant completes a "fixed-point reaction test" level, and the basic profile data shows that they are good at sniper game equipment; the system sets "sudden target turning" as the key observation point in the "moving target shooting" task stage; the practical data acquisition terminal captures the firing delay and bullet impact point deviation; the input frequency duration curve shows a single accurate click rather than continuous firing, and the deviation characteristic duration curve converges rapidly within 0.3 seconds after the turn; the continuous wavelet transform has no significant energy in the high frequency band (>8Hz), ruling out the possibility of panicked clicks; the operation power spectral density map shows that the energy is concentrated in the low frequency steady-state component; the evaluation model outputs high "operational stability" and "spatial awareness" scores; the system matches the "game equipment feel test position" because it requires accurate feedback on single-shot hit experience, rather than just looking at the master rank.

[0039] In one embodiment, basic profile data of the target assessment object is acquired, task stages are divided according to a preset assessment checkpoint system, a set of key observation points is determined, and a practical data collection terminal is deployed, including: Obtain the resume dataset of the target assessment object, extract the game character types that the target assessment object is good at and the operation level characteristics of the target assessment object, and output the target object profile data.

[0040] The resume dataset can be a collection of historical behavior and achievement records accumulated by the target evaluation object in a game platform or e-sports event, which can be used as the original basis for extracting game character type and operation rank features. In this embodiment, the resume dataset can be obtained through game account authorization, event database interface, or third-party data platform API. For example, the resume dataset may include, but is not limited to, ranked match history, usage time of commonly used heroes, and professional event appearance records. The game character type can be the character category that the target evaluation object frequently uses or performs well in a specific game, which can be used to drive the selection of test task segments to ensure that the evaluation content is consistent with the player's actual ability range. In an exemplary embodiment, the game character type can be obtained by clustering indicators such as character usage frequency, win rate, or KDA from the resume dataset. The operation rank feature can be a quantitative or classification identifier reflecting the competitive level of the target evaluation object, which can be used to guide the key observation point distribution strategy and adapt to the evaluation accuracy requirements of players of different levels. Furthermore, the operation rank feature can be extracted based on the game's official rank system or internal ELO rating system. Target object profile data can be a structured individual ability description composed of game character type and skill level characteristics. It can be used to replace general basic profile data, providing more refined personalized input for assessment. In one specific embodiment, target object profile data can be used as query conditions for a game skill assessment knowledge graph to generate test task segments.

[0041] Obtaining the target assessment subject's resume dataset can be achieved by pulling historical data from the target assessment subject on a specified gaming platform through an authorized interface, thus providing raw materials for constructing the target subject's profile. Extracting the target assessment subject's preferred game character types and skill level characteristics can be done by performing statistical analysis and pattern recognition on the resume dataset, outputting character preference tags and skill levels, thus transforming raw resume data into structured ability characteristics. Outputting the target subject's profile data can be achieved by encapsulating the extracted character type and skill level characteristics into standardized JSON or vector formats, thus generating personalized input that can be queried using a knowledge graph.

[0042] Based on a knowledge graph for assessing game operation skills, and combined with the types of game characters that the target assessment subjects are good at, multiple test task segments are determined.

[0043] The game operation ability assessment knowledge graph can be a semantic association network constructed with game mechanics, job responsibilities, and operation ability dimensions as nodes. It can be used to map abstract job ability requirements into executable test task segments. In this embodiment, the game operation ability assessment knowledge graph can be constructed through expert annotation, game design document parsing, and historical assessment data mining. For example, the game operation ability assessment knowledge graph may include, but is not limited to, a multiplayer online tactical arena auxiliary ability subgraph, a first-person shooter game aiming control subgraph, and a real-time strategy game multi-line operation subgraph. Test task segments can be ability-oriented assessment sub-levels retrieved from the knowledge graph based on game character type. These segments can be used to refine the task stage division, focusing the assessment on character-specific ability scenarios.

[0044] Based on a knowledge graph for game operation ability assessment, and combined with the game character types that the target assessment subject is proficient in, multiple test task segments are determined. This can be achieved by retrieving ability nodes associated with the character type from the knowledge graph and mapping them to preset level segments. Furthermore, this operation can be implemented by using graph neural networks for semantic similarity matching to recommend the most relevant task segments, or by using a rule engine to traverse the edge relationships of the knowledge graph to directly return predefined task templates. This achieves the technical effect of semantic alignment between the assessment content and the candidate's expertise and job requirements.

[0045] Based on the operational segment characteristics of the target evaluation object and combined with the accuracy requirements of capability evaluation, key observation points are distributed across multiple test task segments to determine the key observation point set.

[0046] The accuracy requirements for ability assessment can be defined by the data acquisition density and signal-to-noise ratio requirements set for different operational levels. This can constrain the distribution strategy of key observation points, ensuring that high-level players are fully observed in high-discrimination intervals. The distribution of key observation points can be a dynamic spatiotemporal layout of key observation points within test task segments according to accuracy requirements. This can optimize the allocation of observation resources and avoid oversampling in low-information periods or missing high-value operations. Based on the operational level characteristics of the target assessment object and combined with the accuracy requirements of ability assessment, key observation points are distributed across multiple test task segments to determine the key observation point set. This can involve densely setting high-difficulty operation observation points in task segments corresponding to high levels, while focusing on basic reaction points in low-level areas. Furthermore, this operation can be achieved by dynamically adjusting the observation point time window width and spatially sensitive area radius according to the level, or by using information entropy to assess the variance of historical similar players' operations and adding observation points in high-variance intervals. This can improve the technical effect of capturing key operational behaviors and enhancing assessment discrimination.

[0047] Based on the key observation point set, a data acquisition module is configured to form a practical data acquisition terminal. The data acquisition module includes a keyboard, mouse or gamepad input listening component and a game client performance acquisition interface.

[0048] The data acquisition module can be a collection of hardware and software functional units of a practical data acquisition terminal, used to synchronously capture input device signals and game client states. For example, the data acquisition module may include, but is not limited to, keyboard, mouse, or gamepad input monitoring components, game client performance acquisition interfaces, and time synchronization calibration units. The keyboard, mouse, or gamepad input monitoring components can be driver-level or application-level acquisition units that directly monitor low-level input events from peripheral devices, used to acquire latency-free, high-precision operation input timing data. In one specific embodiment, the keyboard, mouse, or gamepad input monitoring components can be implemented through operating system input APIs, device driver hooks, or dedicated SDKs. The game client performance acquisition interface can be an interface that extracts internal states such as skill release position and hit determination from the game engine or client logs, used to obtain the ideal target position and actual landing point coordinates required for skill release deviation data. In this embodiment, the game client performance acquisition interface can be implemented by calling the game's provided debugging interface, memory reading, or frame synchronization log parsing. Based on the set of key observation points, a data acquisition module is configured to form a practical data acquisition terminal. This can be achieved by activating the input listening component and binding it to the game client performance acquisition interface, and starting high-frequency sampling within the time window of the key observation points. This can achieve the technical effect of building a high-fidelity practical data acquisition capability for specific observation needs.

[0049] Taking the evaluation of high-ranking MOBA junglers as an example, the game industry talent ability evaluation method in this embodiment can be as follows: The system obtains the resume dataset of a King-ranked player, extracts their preferred "rhythm-based jungler" role type and "high GPM (gold per minute)" operation rank characteristics, and generates target object profile data; based on the game operation ability evaluation knowledge graph, it retrieves test task segments such as "invading and counter-jungling" and "dragon control timing"; due to their high-ranking characteristics, the system densely sets key observation points such as "smite release judgment frame" and "movement path in blind spot" in the "enemy red zone invasion" sub-segment; the data acquisition module simultaneously enables mouse click monitoring and game client skill hit log interface to capture operation details within a 0.1-second time window; the high-quality signals finally collected support subsequent wavelet analysis to accurately identify their "explosive jungle clearing rhythm" and "skill prediction accuracy", avoiding the problem of insufficient coverage of their high-level abilities in traditional fixed levels.

[0050] In one embodiment, real-time game operation data is parsed, input frequency features and operation deviation features of multiple observation points in a key observation point set are extracted, and input frequency duration curves and deviation feature duration curves are constructed. The output is feature duration information, including: Based on real-time game operation data, the first real-time operation data of the first observation point is extracted. The first real-time operation data includes the frequency of the first operation input and the deviation of the first skill release. The frequency of the first operation input is analyzed, the operation input acceleration is extracted, and based on the timestamp of the first real-time operation data, an input acceleration-time curve is constructed, and the output is the first input feature duration curve; Analyze the first skill release deviation, obtain the operation deviation amplitude, and construct a deviation amplitude-time curve based on the timestamp of the first real-time operation data. The output is the first deviation feature duration curve. By traversing multiple observation points in the key observation point set, feature extraction is performed based on real-time game operation data to obtain multiple sets of input feature duration curves and deviation feature duration curves, which are then stored as feature duration information.

[0051] The first observation point can be the first specific operation event or time window processed in the key observation point set, which can be used as the starting processing unit of the feature extraction process to verify and standardize subsequent multi-point processing logic. The first real-time operation data can be a subset of the original game operations captured within the time window of the first observation point, which can be used as a local data source for input frequency and deviation feature extraction. Further, the first real-time operation data can include the first operation input frequency and the first skill release deviation, etc. The first operation input frequency can be the operation instruction count sequence per unit time within the first observation point, which can be used to calculate the operation input acceleration and reflect the rhythm change trend. The operation input acceleration can be the first derivative of the operation input frequency with respect to time, which characterizes the degree of acceleration or deceleration of the operation rhythm. In this embodiment, the operation input acceleration can be calculated by numerical differentiation (such as central difference) of the input frequency sequence, which can be used to reveal dynamic behavior patterns such as combo initiation, sudden stop, or frantic clicking. For example, the operation input acceleration can include, but is not limited to, one or more of the following: positive acceleration segment, negative deceleration segment, and zero acceleration steady-state segment. A timestamp can be a precise time marker of when an operation event occurs, and can be used to provide a timeline reference for building a timeline curve, ensuring time alignment.

[0052] The input acceleration-time curve can be a continuous function curve plotted with time as the horizontal axis and operational input acceleration as the vertical axis. It can be used to visualize the dynamic rate of change of the operational rhythm and identify explosive or inhibited operational behaviors. In an exemplary embodiment, the input acceleration-time curve can be a specific implementation of the first input feature duration curve, participating in the formation of feature duration information. The first input feature duration curve can be the input-side feature time sequence of the first observation point represented by the input acceleration-time curve, which can be used to structurally express the dynamic evolution characteristics of the operational rhythm within the observation point. The first skill release deviation can be the spatial error vector between the actual landing point of the skill and the ideal target position within the first observation point. It can be used to calculate the operational deviation amplitude and measure the accuracy of a single or sequential skill release. The operational deviation amplitude can be the Euclidean norm or Manhattan distance of the skill release deviation vector, characterizing the magnitude of the deviation. In a specific embodiment, the operational deviation amplitude can be obtained by calculating the magnitude of the deviation coordinate components, which can be used to quantify the operational control accuracy, eliminate directional interference, and focus the error intensity. Furthermore, the operational deviation amplitude can include, but is not limited to, one or more of radial deviation amplitude, lateral offset amplitude, and longitudinal offset amplitude.

[0053] The deviation amplitude-time curve can be a continuous function curve plotted with time on the horizontal axis and the operational deviation amplitude on the vertical axis. It can be used to demonstrate the fluctuation of skill release accuracy over time and identify intervals of decreased stability. For example, the deviation amplitude-time curve can be a specific implementation of the first deviation feature duration curve, contributing to the feature duration information. The first deviation feature duration curve can be the deviation-side feature time sequence of the first observation point, represented by the deviation amplitude-time curve, used to structurally express the dynamic stability of operational accuracy within that observation point. Multiple sets of input feature duration curves are a set of input acceleration-time curves generated after traversing all observation points, which can be used to comprehensively cover the dynamic characteristics of operational rhythm in each key task stage. Multiple sets of deviation feature duration curves can be a set of deviation amplitude-time curves generated after traversing all observation points, which can be used to comprehensively cover the operational accuracy stability characteristics in each key task stage.

[0054] The feature duration information can be a structured dual-channel time-series dataset composed of multiple sets of input feature duration curves and multiple sets of deviation feature duration curves. This dataset can be used as input for continuous wavelet transform, carrying high-order dynamic operation features. In one specific embodiment, compared to the duration curves constructed based on frequency and original deviation in the previous scheme, the feature duration information is upgraded to feature curves with stronger physical meaning based on acceleration and deviation amplitude, improving the semantic interpretability of time-frequency analysis. Extracting the first real-time operation data of the first observation point can be achieved by slicing the real-time game operation data according to the time window of the first observation point. Furthermore, extracting the first real-time operation data of the first observation point can be achieved by isolating the operation signals of specific high-value time periods, facilitating refined feature extraction. Analyzing the first operation input frequency and extracting the operation input acceleration can be achieved by performing numerical differentiation on the first operation input frequency sequence. Furthermore, by analyzing the frequency of the first operational input and extracting the operational input acceleration, the local acceleration can be calculated using the three-point center difference formula, or the Savitzky-Golay filter can be used to smooth the signal while simultaneously calculating the derivative to suppress noise amplification. This allows the static frequency index to be upgraded to a dynamic rhythm change rate feature.

[0055] Based on the timestamps of the first real-time operation data, an input acceleration-time curve is constructed. This can be achieved by interpolating the calculated operation input acceleration according to the corresponding timestamps to form a continuous curve. Furthermore, constructing the input acceleration-time curve based on the timestamps of the first real-time operation data can be achieved by establishing a dynamic representation of the operation rhythm with clear time alignment. The output is the first input feature duration curve, which can be achieved by encapsulating the input acceleration-time curve into a standardized data structure. Further, the output as the first input feature duration curve can be achieved by generating input-side feature units that can be uniformly processed in subsequent processes. The first skill release deviation is analyzed to obtain the operation deviation amplitude, which can be achieved by calculating the magnitude or L2 norm of the deviation vector. Further, analyzing the first skill release deviation to obtain the operation deviation amplitude can be achieved by converting the multi-dimensional spatial error into a scalar intensity index, facilitating time-series modeling.

[0056] Based on the timestamps of the first real-time operation data, a deviation amplitude-time curve is constructed, which can be achieved by aligning the operation deviation amplitude according to the timestamps and then interpolating and smoothing it. Furthermore, based on the timestamps of the first real-time operation data, constructing the deviation amplitude-time curve can create a visualized temporal trajectory of skill accuracy stability. The output is the first deviation feature duration curve, which can be formatted as a structured feature object. Furthermore, the output is the first deviation feature duration curve, which can generate deviation-side feature units that can be uniformly processed in subsequent processes. Traversing multiple observation points in the key observation point set, feature extraction is performed based on real-time game operation data, which can be achieved by repeating the above acceleration and deviation amplitude extraction process for each observation point. Furthermore, traversing multiple observation points in the key observation point set, feature extraction based on real-time game operation data, can achieve full-coverage feature modeling of the key operation intervals throughout the entire evaluation cycle. Obtaining multiple sets of input feature duration curves and deviation feature duration curves can be achieved by aggregating the two types of curves generated from all observation points. Furthermore, obtaining multiple sets of input feature duration curves and deviation feature duration curves can be achieved by constructing a complete two-dimensional dynamic feature set. Storing as feature duration information can be achieved by packaging and storing multiple sets of curves according to their observation point ID index. Furthermore, storing as feature duration information can provide structurally clear and semantically explicit input data for continuous wavelet transforms.

[0057] Taking the assessment of FPS professional players' recoil stability during shooting as an example, the talent assessment method for the gaming industry in this embodiment can be as follows: In the "moving target sweeping" observation point, the system extracts the mouse movement frequency sequence of 10 consecutive projectiles fired by the player, calculates their input acceleration, and finds that the acceleration is positive in the first 3 frames (rapid downward movement), and approaches zero in the last 7 frames (stable maintenance). The input acceleration-time curve shows a typical "rapid drop-stable control" pattern. At the same time, the skill release deviation (distance between the bullet impact point and the bullseye) is converted into deviation amplitude. The deviation amplitude-time curve shows that the deviation is large in the first 2 shots (>30 pixels), and then quickly converges to <10 pixels and remains stable. This combination of curves indicates that the player has excellent control over the recoil initiation response and continuous control ability. The system stores the two curves of this observation point in the feature duration information for subsequent wavelet analysis to identify its high-frequency steady-state control components, which is different from the high acceleration jitter and continuous deviation oscillation pattern commonly seen in beginners.

[0058] In one embodiment, based on continuous wavelet transform, time-frequency analysis is performed on the feature duration information to obtain the operated wavelet feature information, including: Based on the target object profile data, extract the baseline operation frequency information of this type of game character; The wavelet transform parameters are initialized using the Morlet wavelet as the target wavelet mother function and the reference operating frequency information as the scale constraint. Perform a continuous wavelet transform on the first input feature duration curve to obtain the first input wavelet coefficient matrix; perform a continuous wavelet transform on the first deviation feature duration curve to obtain the first deviation wavelet coefficient matrix; The first input wavelet coefficient matrix and the first deviation wavelet coefficient matrix are combined and stored as the first wavelet analysis data set; Perform continuous wavelet transforms on the feature time information to obtain multiple wavelet analysis data sets, which are then stored as wavelet analysis datasets.

[0059] In this context, the game character can refer to the role category played or excelled at by the target evaluation subject in a specific game, and can be used as a classification basis for extracting baseline operation frequency information. In a specific embodiment, the game character can be a specific role type parsed from the target object's profile data, used for frequency band prior setting. The baseline operation frequency information can be the dominant rhythm frequency range of a specific game character in typical operation scenarios, and can be used to provide prior constraints on the scale parameters for continuous wavelet transform, focusing on the effective frequency band. For example, the baseline operation frequency information can be derived from the operation data of historical high-performing players, such as the 1-2Hz control rhythm commonly used by MOBA support roles, and the 5-8Hz concentration of instantaneous operations for FPS sniper roles. Furthermore, the baseline operation frequency information can include, but is not limited to, one or more of low-frequency control rhythm, mid-frequency combo rhythm, and high-frequency reaction rhythm. The Morlet wavelet can be a complex-valued continuous wavelet mother function, in the form of a complex exponentially modulated Gaussian window, which can be used as a time-frequency analysis tool, achieving a good balance between time and frequency resolution, and is suitable for capturing periodic and transient operation patterns. In one exemplary embodiment, the Morlet wavelet can be used as the wavelet mother function, whose scaling parameters are constrained by the reference operating frequency information to achieve adaptive band focusing.

[0060] The wavelet mother function can be the basic waveform function used for scaling and translation in continuous wavelet transform, and can be used to determine the localization characteristics and frequency band response shape of time-frequency analysis. Scale constraints can be upper and lower bounds imposed on the wavelet transform scale parameters, and can be used to avoid wasting computational resources in irrelevant frequency bands (such as extremely low-frequency drift or extremely high-frequency noise) and improve the feature signal-to-noise ratio. In this embodiment, the scale constraint can be derived from the reference operating frequency information through Fourier correspondence. Wavelet transform parameters can be configuration items in continuous wavelet transform, including mother function type, scale range, sampling density, etc., and can be used to control the granularity and coverage of time-frequency decomposition. In a specific embodiment, the wavelet transform parameters can be initialized jointly by Morlet wavelet and reference operating frequency information to achieve adaptive role configuration. The first input wavelet coefficient matrix can be a two-dimensional array of complex or real coefficients obtained after performing a continuous wavelet transform on the first input feature duration curve, with rows corresponding to time and columns corresponding to scale (frequency), and can be used to characterize the dynamic energy distribution of operating acceleration in the time-frequency domain, revealing the burstiness and periodicity of rhythm. Furthermore, the first input wavelet coefficient matrix may include, but is not limited to, amplitude coefficient submatrix, phase coefficient submatrix, energy density submatrix, etc.

[0061] The first deviation wavelet coefficient matrix can be a time-frequency coefficient matrix obtained by performing a continuous wavelet transform on the first deviation feature duration curve. It can be used to characterize the structural features of skill release accuracy fluctuations in the time-frequency domain and identify stability decay during high-pressure periods. For example, the first deviation wavelet coefficient matrix can include, but is not limited to, low-frequency trend deviation spectrum, transient jitter deviation spectrum, and periodic error spectrum. The first wavelet analysis data set can be a dual-channel time-frequency feature unit composed of the first input wavelet coefficient matrix and the first deviation wavelet coefficient matrix, used to preserve the coupling relationship between operational rhythm and accuracy in the time-frequency domain, supporting multi-dimensional joint assessment of capabilities. Multiple wavelet analysis data sets can be a collection of wavelet analysis data sets generated by traversing all observation points, which can be used to cover the time-frequency features of each key task stage in the entire evaluation process. The wavelet analysis dataset can be a structured high-dimensional feature library containing all wavelet analysis data sets, which can be used as direct input to the game talent operational capability assessment model, supporting fine-grained capability quantification. In one exemplary embodiment, the wavelet analysis dataset, compared to the generalized operational wavelet feature information in the preceding scheme, is explicitly generated by role-adaptive wavelet transform, exhibiting stronger job alignment and physical interpretability.

[0062] Based on the target object's profile data, the baseline operation frequency information of this type of game character is extracted. This can be done by querying a pre-built character-frequency mapping table or by clustering historical data of high-performing players of the same type to obtain the dominant frequency band. Furthermore, this operation can be implemented in the above manner, thus providing domain priors for wavelet transform and achieving frequency band focusing. Using the Morlet wavelet as the target wavelet mother function and the baseline operation frequency information as the scale constraint, the wavelet transform parameters are initialized. This can be done by converting the baseline operation frequency f0 into the Morlet wavelet center frequency, setting the scale range s∈[s_min, s_max] to correspond to f∈[0.5f0, 2f0]. ​​Further, this operation can be achieved by using logarithmic scaling to densify scale points within the dominant frequency band, or by using linear scaling but dynamically adjusting the window width to match the temporal locality of different frequencies. This allows for adaptive time-frequency resolution configuration for characters, improving the sensitivity of key pattern detection.

[0063] A continuous wavelet transform is performed on the first input feature duration curve to obtain the first input wavelet coefficient matrix. This can be achieved by convolving the input acceleration-time curve with Morlet wavelets of different scales, outputting a complex coefficient matrix. Furthermore, this operation, implemented in the above manner, can extract the time-varying frequency components of the operation rhythm, such as the periodicity or sudden acceleration of combos. A continuous wavelet transform is performed on the first deviation feature duration curve to obtain the first deviation wavelet coefficient matrix. This can be achieved by performing the same wavelet transform process on the deviation amplitude-time curve. Furthermore, this operation, implemented in the above manner, can reveal the time-frequency structure of skill errors, such as whether broadband jitter occurs during team battles. The first input wavelet coefficient matrix and the first deviation wavelet coefficient matrix are combined and stored as the first wavelet analysis data set. This can be achieved by concatenating the two matrices along the channel dimension or storing them with an associated index. Furthermore, this operation, implemented in the above manner, can construct a dual-modal time-frequency feature unit, supporting comprehensive judgment of operation quality. Continuous wavelet transforms are performed on the feature duration information to obtain multiple wavelet analysis data sets. This can be achieved by repeating the above transformation and combination process for each set of input / bias feature duration curves. Furthermore, this operation can be implemented in the above manner, thereby achieving full coverage of time-frequency features throughout the entire evaluation cycle. The data is stored as a wavelet analysis dataset, which can be achieved by organizing all wavelet analysis data sets into HDF5 or Parquet format datasets according to observation point IDs. Furthermore, this operation can be implemented in the above manner, thereby providing the evaluation model with structurally uniform, high-dimensional, and dense input features.

[0064] Taking the evaluation of burst combos by mid-lane mages in MOBA games as an example, the talent assessment method for the gaming industry in this embodiment can be as follows: The system identifies the target's proficiency in "high-burst mage" roles from the target's profile data and extracts the baseline operation frequency information as 3-6Hz (corresponding to the QWER combo rhythm); the Morlet wavelet scale parameters are initialized to focus on this frequency band; the input acceleration-time curve of the "team fight begins" observation point is transformed, and the first input wavelet coefficient matrix shows a strong energy stripe at 4.2Hz for 2 seconds, indicating a stable combo rhythm; at the same time, the wavelet transformation of the deviation amplitude-time curve shows that the error energy is concentrated in the low frequency (<1Hz), indicating that the skill landing point is generally offset but there is no high-frequency jitter, reflecting calm casting; the two are merged into the first wavelet analysis data set; in contrast, panicked players show wideband scattered energy in the same frequency band and the deviation signal has a sudden peak at 3-5Hz, indicating uncontrolled hand tremors; this difference is completely captured by the wavelet analysis dataset, supporting the evaluation model to accurately distinguish the operation quality.

[0065] In one embodiment, based on continuous wavelet transform, time-frequency analysis is performed on the feature duration information to obtain the operated wavelet feature information, and the method further includes: Based on the wavelet analysis data set, the wavelet energy density is calculated to obtain the energy density matrix set, which includes the input energy density matrix and the bias energy density matrix. Based on the energy density matrix, a time-frequency spectrum set is plotted with time as the first coordinate axis and scale constraint as the second coordinate axis. Integrating the first wavelet analysis data set using the modulus square of the wavelet coefficient matrix yields the power spectrum set. Store time-frequency spectrum sets and power spectrum sets to obtain operational wavelet feature information.

[0066] The wavelet energy density, which can be the square of the wavelet coefficient modulus, characterizes the local energy intensity of a signal at a specific time and scale (frequency). It can be used to transform abstract wavelet coefficients into physically meaningful energy distributions, supporting visualization and quantitative analysis. In this embodiment, the wavelet energy density can be obtained by calculating the square of the modulus of the complex or real wavelet coefficient matrix element by element. The energy density matrix set can be a dual-channel energy representation set composed of the input energy density matrix and the deviation energy density matrix, which can be used to characterize the energy distribution of operational rhythm dynamics and skill precision fluctuations in the time and frequency domains, respectively. For example, the energy density matrix set can include, but is not limited to, the input energy density matrix and the deviation energy density matrix. The input energy density matrix can be a two-dimensional array of energy densities calculated from the first input wavelet coefficient matrix, which can be used to reflect the activity level of operational acceleration changes under various time-frequency combinations, revealing rhythmic patterns such as combos and sudden stops.

[0067] The deviation energy density matrix can be a two-dimensional array of energy density calculated from the first deviation wavelet coefficient matrix. It can be used to reflect the energy concentration area of ​​skill release error fluctuations and identify periods of control instability under high pressure. Time can be the time coordinate of the operation, serving as the first coordinate axis of the time-frequency map and providing a time positioning reference. Scale constraints can be the range of scale parameters limited by the reference operation frequency information in the continuous wavelet transform. They can be used as the second coordinate axis of the time-frequency map, corresponding to the reciprocal of the frequency, reflecting the frequency band focusing characteristics. In an exemplary embodiment, scale constraints participate in map construction as the vertical axis, strengthening the correlation with role priors. The time-frequency map set can be a collection of heatmaps plotted with time as the horizontal axis, scale constraints as the vertical axis, and energy density as color intensity. It can be used to visually display the energy distribution structure of the operation on the time-frequency plane, such as periodic patches or sudden stripes. Furthermore, the time-frequency map set can include, but is not limited to, input time-frequency maps, deviation time-frequency maps, and multi-observation point time-frequency map sequences.

[0068] The modulus square of the wavelet coefficient matrix can be the square of the absolute value of the complex values ​​of the wavelet coefficients, i.e., the mathematical expression of energy density. It can be used as the integrand of the power spectrum integral, providing a basis for frequency domain energy calculation. Integration can be a numerical summation or integration operation of the modulus square of the wavelet coefficient matrix along the scale (frequency) dimension. This can be used to compress the frequency domain dimension, generating a time-to-total energy relationship curve, equivalent to a local power spectral density. The power spectrum atlas can be a set of energy-time curves obtained by integrating at each time point along the scale dimension. It can be used to characterize the total response intensity of operations at each moment, and to identify explosive operations at key nodes (such as the end of skill cooldown). In a specific embodiment, the power spectrum atlas may include, but is not limited to, the input power spectrum, the deviation power spectrum, and the segmented power spectrum comparison map. Operational wavelet feature information can be a set of visualized and measurable high-order operational features composed of the time-frequency spectrum atlas and the power spectrum atlas. It can be used as the final input of a game talent operational ability assessment model, carrying a concrete representation of abstract abilities such as rhythm and stability. In this embodiment, the wavelet feature information is composed of a dual spectrum derived from energy density, which places greater emphasis on visualization and physical interpretability compared to the previous scheme.

[0069] Based on the wavelet analysis data set, the wavelet energy density is calculated to obtain an energy density matrix set. This can be achieved by calculating the modulus squares of the input and bias wavelet coefficient matrices in the wavelet analysis data set separately. Furthermore, this operation can be implemented by taking the modulus squares of each element of the complex or real wavelet coefficient matrix, thus transforming the complex coefficients into energy distributions, laying the foundation for visualization and quantitative analysis. Based on the energy density matrix set, a time-frequency spectrum set is plotted with time as the first coordinate axis and scale constraints as the second coordinate axis. This can be achieved by rendering the energy density matrix as a color heatmap using a time-scale grid. Further, this operation can be achieved by using logarithmic color levels to enhance the visibility of low-energy details, or by overlaying contour lines to highlight significant energy concentration areas, thereby generating a time-frequency distribution image that intuitively reflects the operational rhythm structure. Integrating the modulus squares of the wavelet coefficient matrix over the first wavelet analysis data set yields a power spectrum set. This can be achieved by summing the energy density at each time point along the scale dimension to generate a time-total energy sequence. Furthermore, this operation can be achieved by using trapezoidal numerical integration to process non-uniform scale sampling, or by weighted integration within the dominant frequency band to highlight the contribution of the relevant frequency bands, thereby extracting the overall intensity of the operational response at each moment and supporting the recognition of key node behaviors. Storing time-frequency and power spectrum sets to obtain operational wavelet feature information can be achieved by packaging the two types of spectra into image tensors or structured feature vectors according to observation point IDs, thus forming a final feature input with strong interpretability and high discriminative power.

[0070] Taking the multi-tasking stress test of RTS professional players as an example, the game industry talent ability assessment method in this embodiment can be as follows: In the observation point of "simultaneous three-line encounter", the system calculates the energy density after performing wavelet transform on the input acceleration duration curve of the player. The time-frequency spectrum shows three synchronous periodic energy patches in the 2-3Hz frequency band, corresponding to the stable micro-operation rhythm of the three lines. The deviation energy density spectrum shows no significant high-frequency energy in the same time period, indicating that the operation accuracy has not decreased due to multi-line pressure. The power spectrum shows a sharp peak at 0.5 seconds after the start of the encounter, reflecting the player's rapid response ability when the crisis first appears. In contrast, the time-frequency spectrum of ordinary players shows broadband diffuse noise, the power spectrum has no clear peak, and the deviation spectrum shows a surge of energy in the high-frequency band, indicating that they are flustered. This dual spectrum is completely stored in the operation wavelet feature information, enabling the evaluation model to clearly identify the high-order ability of "calm multi-line control".

[0071] In one embodiment, constructing a game talent operational ability assessment model further includes: Collect historical evaluation records from game industry evaluation experts to obtain a basic evaluation sample set.

[0072] The historical evaluation records of game industry assessment experts can be a collection of structured or semi-structured scores and comments on player performance made by senior esports coaches, professional team analysts, or game testing directors. These records can serve as authoritative labeling sources for constructing the basic assessment sample set, injecting prior domain knowledge. Furthermore, the historical evaluation records of game industry assessment experts can be exported from internal corporate talent assessment systems, tournament review databases, or expert annotation platforms. In an exemplary embodiment, the historical evaluation records of game industry assessment experts are mapped to the original game operation data, supporting subsequent feature extraction and label alignment. The basic assessment sample set can be an initial training sample set formed by associating historical evaluation records with corresponding original game operation data. This set can be used to provide a mapping relationship between real human judgment and operational behavior, ensuring that the model's learning conforms to industry-recognized capability standards.

[0073] Collecting historical evaluation records from game industry assessment experts can be achieved by exporting timestamped expert rating records from corporate HR systems or esports club databases. Furthermore, this collection can be accomplished by interfacing with structured database interfaces or batch exporting annotation platform logs, thereby obtaining high-quality, domain-aligned capability tags. Obtaining a basic assessment sample set can be achieved by aligning expert evaluation records with their corresponding raw game operation data (processed into wavelet features using the aforementioned process) using IDs. Further, obtaining a basic assessment sample set can be achieved by establishing a unified identifier indexing mechanism, thereby constructing initial supervised learning data pairs.

[0074] Based on industry big data and combined with target game character characteristic data, an expanded evaluation sample set is generated.

[0075] Industry big data can be a large-scale, anonymized operational behavior database covering multiple games, skill levels, and character roles. This data can support data augmentation and synthetic sample generation, improving model generalization capabilities. In one specific embodiment, industry big data can be aggregated from partner game developer APIs, publicly available tournament logs, or user-authorized data. Target game character characteristic data can describe the requirements for operational rhythm, precision, complexity, etc., of a specific game character in its mechanism design. This data can guide the reasonable constraints of expanded samples, ensuring that the synthetic data conforms to the character's semantic capabilities. For example, target game character characteristic data can originate from game balance documents, professional player training manuals, or knowledge graph node attributes. The expanded evaluation sample set can be a virtual evaluation sample set synthesized based on industry big data and character characteristic data through interpolation, perturbation, or generative methods. This can alleviate the problem of insufficient real-labeled data and expand the coverage density of samples in the character-skill-style space.

[0076] Based on industry big data and target game character characteristic data, an expanded evaluation sample set can be generated. This can be achieved by filtering similar character operation flows from industry big data and performing parameterized perturbations or GAN generation based on character characteristic data constraints. Furthermore, based on industry big data and target game character characteristic data, the expanded evaluation sample set can be generated by using variational autoencoders to interpolate intermediate style samples under character characteristic constraints, or by applying controllable noise to the operation flows of high-performing players to simulate performance under different stress levels. This expands the diversity of training data and covers rare but reasonable operation patterns.

[0077] By integrating the basic assessment sample set and the extended assessment sample set, a training sample set is obtained. The training sample set includes sample wavelet feature information, sample operation rhythm component information, sample operation complexity order information, and sample practical ability level.

[0078] The training sample set can be a complete supervised learning dataset formed by fusing basic and expanded samples, which can be used to provide highly diverse, representative, and multi-dimensional labeled input-output pairs for model training. In this embodiment, the training sample set explicitly includes four key fields, emphasizing explicit modeling of rhythm and complexity more than previous schemes. The wavelet feature information of the samples can be the numerical feature vector extracted from the time-frequency spectrum set and power spectrum set corresponding to each training sample, which can be used as the main input of the model to carry the high-order time-frequency structure of the operation dynamics. The rhythm component information of the sample operation can be the rhythm type quantification index extracted after pattern recognition of the time-frequency spectrum, which can be used to explicitly characterize whether the operation has rhythm patterns such as periodic combos, sudden responses, or steady-state maintenance. In an exemplary embodiment, the rhythm component information of the sample operation can be obtained from the energy density matrix through peak detection, periodic analysis, or clustering algorithms. For example, the rhythm component information of the sample operation can include, but is not limited to, one or more of the following: periodic intensity coefficient, sudden energy proportion, rhythm stability entropy value, etc.

[0079] The order of sample operation complexity can be a comprehensive measure reflecting the number of skill collaborations, micro-operation density, and decision-making level in a single operation sequence. It can be used to distinguish between simple repetitive clicks and high-order composite micro-operations (such as A-step + skill prediction + vision control). Furthermore, the order of sample operation complexity can be calculated based on skill combination diversity, effective number of operations per unit time, and context dependency depth. In a specific embodiment, the order of sample operation complexity may include, but is not limited to, one or more of the following: skill collaboration order, micro-operation density index, and context dependency depth. The sample practical ability level can be a structured ability rating label generated by expert evaluation or rule mapping, which can be used as a supervisory signal to guide the model in learning the mapping relationship from features to abilities. For example, the sample practical ability level may include, but is not limited to, one or more of the following: reaction speed level, operation stability level, and skill accuracy level. The training sample set is obtained by merging the basic assessment sample set and the extended assessment sample set. This can be achieved by proportionally merging the two types of samples and unifying the feature and label formats. Furthermore, by integrating the basic assessment sample set and the expanded assessment sample set, the training sample set can be aligned by setting sampling weights or adversarial verification mechanisms, thereby forming a complete training set that balances realism and generalization.

[0080] Based on the training sample set, a game talent operation ability assessment model is constructed and trained.

[0081] The game talent operation ability assessment model can be a multi-input, multi-task machine learning model built based on a training sample set. It can be used to achieve an objective quantitative mapping from operation frequency features to practical ability rating information. In one specific embodiment, the game talent operation ability assessment model adopts a gradient boosting tree, multilayer perceptron, or graph neural network architecture, using sample wavelet feature information as the main input and fusing rhythm and complexity auxiliary features for end-to-end training. In this embodiment, the game talent operation ability assessment model explicitly relies on multi-dimensional supervision signals (including rhythm components and complexity order) for training, exhibiting stronger semantic understanding and discrimination capabilities compared to previous schemes. Building and training the game talent operation ability assessment model based on the training sample set can involve concatenating sample wavelet feature information, rhythm component information, and complexity order information into an input vector, and training a regression or classification model using the sample practical ability level as the label. Furthermore, based on the training sample set, the game talent operation ability assessment model can be constructed and trained by using a multi-task learning framework to predict the sub-dimensions of reaction speed, stability, and accuracy, or by introducing an attention mechanism to dynamically weight the wavelet feature contributions of different observation points. This allows the model to make an overall ability judgment by comprehensively considering time-frequency structure, rhythm pattern, and operation complexity.

[0082] Taking the advanced micro-management assessment of MOBA junglers as an example, the talent assessment method for the gaming industry in this embodiment can be as follows: The system obtains the score (high reaction speed, high complexity) of a professional jungler in the "counter-gank" level from expert records; its sample wavelet feature information shows strong periodic energy in the 3-5Hz frequency band; the sample operation rhythm component information is marked as "sudden-periodic hybrid type"; the sample operation complexity order information is 3rd order (simultaneously handling positioning, Smite timing, and skill prediction); this sample is used to train the model. During the test, an applicant performs the same level. Although the APM is similar, its rhythm component is "random wideband", and the complexity order is only 1st order (only repeating basic attacks). The deviation energy surges during team fights; the model integrates the three types of features to output a low stability and low complexity score, correctly identifying that it does not possess the advanced micro-management ability required for a professional jungler, avoiding the misjudgment of traditional APM screening.

[0083] In one embodiment, based on practical skills rating information, game job matching feedback is performed on the target assessment subject, including: The practical ability rating information includes specific ability scores corresponding to multiple key observation points, and multiple key observation points are marked with corresponding ability dimension labels; Specifically, the specialized ability score can be a quantitative score of a specific ability dimension evaluated for a single key observation point. It can be used to provide fine-grained, scenario-based ability performance data to support multi-dimensional job matching. In an exemplary embodiment, the specialized ability score can be output by a game talent operational ability assessment model based on the wavelet features of the corresponding observation point. Furthermore, the specialized ability score can include, but is not limited to, one or more of the following: reaction speed score, micro-operation stability score, and skill accuracy score. The ability dimension label can be a semantic identifier that marks the core competitive ability category corresponding to the key observation point. It can be used to assign interpretable business semantics to the specialized ability score, achieving ability-job semantic alignment. For example, the ability dimension label can be preset during the level design stage and bound to game mechanics and job requirements.

[0084] Based on the hiring standards of game companies, a talent response matrix is ​​defined, which includes multiple sets of ability scoring ranges and related job recruitment or training suggestions. The hiring standards for game companies can be standardized documents or structured rule sets out by the game company for different positions, which can be used as the basis for constructing a talent response matrix to ensure that the recommendation results meet actual recruitment needs. In one specific embodiment, the hiring standards for game companies can be jointly formulated by the HR department and the business team, and include a job competency model. The talent response matrix can be a rule table that establishes a mapping relationship between competency score ranges and job recruitment or training recommendations, which can be used to automate the conversion from quantified competency to recruitment decisions. Furthermore, the talent response matrix can be formally constructed based on the hiring standards of game companies through a rule engine or expert system. For example, the talent response matrix can adopt direct recruitment rule groups, targeted training rule groups, and elimination warning rule groups, etc. Multiple competency score ranges can be a set of threshold ranges set for different competency dimensions, which can be used to define whether candidates meet the competency thresholds or development potential of specific positions. In an exemplary embodiment, multiple competency score ranges can include high potential ranges (e.g., 85-100 points), qualified ranges (e.g., 70-84 points), and unqualified ranges (<70 points), etc. Job placement or training recommendations can be talent management strategies generated based on competency matching results. These recommendations can guide HR in making hiring, training, or dismissal decisions, improving the efficiency of person-job fit. Furthermore, job placement or training recommendations can include direct recruitment to front-line teams, targeted training in youth training camps, or recommendations for testing roles rather than professional roles.

[0085] Based on the talent response matrix, the practical ability rating information is traversed and matched, corresponding suggested solutions are matched, and the evaluation results of the target assessment object are output.

[0086] The evaluation results can be a structured talent assessment report for recruiters or candidates, presenting job matching conclusions and capability gap analysis to support subsequent decision-making. In this embodiment, the output is driven by a talent response matrix, with a clear job orientation, unlike the previous general rating that only outputs capability levels. The practical capability rating information includes specialized capability scores corresponding to multiple key observation points, each labeled with a corresponding capability dimension tag. This can be achieved by decomposing the model output by observation point ID and associating it with preset capability dimension tags, thus giving the rating information contextual semantics and supporting dimension-level matching. Based on the hiring standards of game companies, a talent response matrix is ​​defined, which can transform the capability requirements in the hiring standards into formal rules, such as "FPS precision shooting position: micro-operation stability ≥ 80 points and reaction speed ≥ 75 points → direct hiring". Furthermore, this operation can be achieved by using a decision tree structure to encode multi-condition combination rules, or by using fuzzy logic to handle boundary cases (such as 78 points being close to the 80-point threshold), thereby establishing an executable mapping logic between capabilities and positions. Based on the talent response matrix, the practical ability rating information is traversed and matched. This can be done for each job rule, checking whether the candidate's scores for each specific ability fall within a specified range, thus enabling parallel matching for multiple positions and identifying the optimal fit. Corresponding suggested solutions are then matched, and the evaluation results for the target assessment object are output. This can be done by integrating the successfully matched suggested solutions into a structured report, including recommended positions, strengths, and areas for improvement, thereby providing actionable and interpretable recruitment decision support information.

[0087] Taking the evaluation of MOBA support player candidates as an example, the talent assessment method in this embodiment can be as follows: After a candidate completes the assessment, their practical ability rating information shows that they score 88 in the "Vision Control" observation point (ability dimension label: Tactical Awareness); 92 in the "Teamfight Protection" observation point (label: Micro-operation Stability); but only 65 in the "Lane Suppression" observation point (label: Offensive Tempo). The talent response matrix defines the requirements for the "MOBA Support Player" position as: Tactical Awareness ≥ 85, Micro-operation Stability ≥ 90, and no hard requirement for Offensive Tempo → the recommended solution is "direct recruitment"; while the "Roaming Jungler" position requires Offensive Tempo ≥ 80 → mismatch. The system automatically outputs a recommendation for them to serve as a hard support, and prompts "Offensive Tempo is a weakness, it is not recommended to switch to the roaming player position", avoiding the mismatch caused by the traditional method of only looking at the overall score of 80.

[0088] In addition, refer to Figure 2 To achieve the above objectives, the present invention also provides a talent competency assessment system for the game industry, the system comprising: The profile preparation module 10 is used to acquire basic profile data of the target assessment object, divide the task into stages according to the preset assessment checkpoint system, determine the key observation point set, and deploy the practical data acquisition terminal. The practical data acquisition module 20 is used to start the practical data acquisition terminal and collect the real-time game operation data of the target evaluation object. The real-time game operation data includes operation input timing data and skill release deviation data. The feature construction module 30 is used to parse the real-time game operation data, extract the input frequency features and operation deviation features of multiple observation points in the key observation point set, construct the input frequency duration curve and the deviation feature duration curve, and output the feature duration information. Wavelet time-frequency module 40 is used to perform time-frequency analysis on the feature duration information based on continuous wavelet transform to obtain operation wavelet feature information, wherein the operation wavelet feature information includes operation time-frequency distribution feature map and operation power spectral density map. The model rating module 50 is used to construct a game talent operation ability assessment model. It takes the operation wavelet feature information as input, runs the game talent operation ability assessment model to perform quantitative assessment of ability, and outputs practical ability rating information. The matching feedback module 60 is used to perform game job matching feedback for the target assessment object based on the practical ability rating information.

[0089] Other embodiments or specific implementations of the talent competence assessment system for the game industry described in this invention can refer to the above-mentioned method embodiments, and will not be repeated here.

[0090] In addition, to achieve the above objectives, the present invention also provides a talent competency assessment device for the game industry, the device comprising: a memory, a processor, and a talent competency assessment program for the game industry stored on the memory and executable on the processor, the talent competency assessment program for the game industry configured to implement the steps of the talent competency assessment method for the game industry as described above.

[0091] In addition, to achieve the above objectives, the present invention also provides a computer-readable storage medium storing a game industry talent competency assessment program, which, when executed by a processor, implements the steps of the game industry talent competency assessment method as described above.

[0092] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A game industry talent ability assessment method, characterized in that, The method includes: Acquire basic profile data of the target assessment object, divide the task into stages according to the preset assessment checkpoint system, determine the key observation point set, and deploy the practical data collection terminal; Start the practical data acquisition terminal to collect real-time game operation data of the target evaluation object. The real-time game operation data includes operation input timing data and skill release deviation data. The real-time game operation data is analyzed, the input frequency features and operation deviation features of multiple observation points in the key observation point set are extracted, and the input frequency duration curve and deviation feature duration curve are constructed, and the output is the feature duration information; Based on continuous wavelet transform, time-frequency analysis is performed on the feature duration information to obtain operation wavelet feature information, wherein the operation wavelet feature information includes operation time-frequency distribution feature map and operation power spectral density map; A game talent operation ability assessment model is constructed. The operation wavelet feature information is used as input, and the game talent operation ability assessment model is run to perform quantitative assessment of the ability and output practical ability rating information. Based on the practical skills rating information, the game job matching feedback for the target assessment subjects is carried out.

2. The method for assessing talent capabilities in the game industry as described in claim 1, characterized in that, The process of acquiring basic profile data of the target assessment object, dividing the task into stages according to the preset assessment checkpoint system, determining the key observation point set, and deploying practical data collection terminals includes: Obtain the resume dataset of the target assessment object, extract the types of game characters that the target assessment object is good at and the operation level characteristics of the target assessment object, and output the target object profile data; Based on the knowledge graph of game operation ability assessment, and combined with the types of game characters that the target assessment subjects are good at, multiple test task segments are determined; Based on the operational segment characteristics of the target evaluation object and combined with the accuracy requirements of the capability evaluation, key observation points are distributed in multiple test task segments to determine the set of key observation points. Based on the set of key observation points, a data acquisition module is configured to form the practical data acquisition terminal, wherein the data acquisition module includes a keyboard, mouse or gamepad input monitoring component and a game client performance acquisition interface.

3. The method for assessing talent capabilities in the game industry as described in claim 2, characterized in that, The process involves parsing the real-time game operation data, extracting input frequency features and operation deviation features from multiple observation points in the key observation point set, constructing input frequency duration curves and deviation feature duration curves, and outputting feature duration information, including: Based on the real-time game operation data, the first real-time operation data of the first observation point is extracted, and the first real-time operation data includes the first operation input frequency and the first skill release deviation. The frequency of the first operation input is analyzed, the operation input acceleration is extracted, and based on the timestamp of the first real-time operation data, an input acceleration-time curve is constructed, and the output is the first input feature duration curve; Analyze the first skill release deviation, obtain the operation deviation amplitude, and construct a deviation amplitude-time curve based on the timestamp of the first real-time operation data, outputting the first deviation feature duration curve; By traversing multiple observation points in the key observation point set, feature extraction is performed based on the real-time game operation data to obtain multiple sets of input feature duration curves and deviation feature duration curves, which are then stored as feature duration information.

4. The method for assessing talent capabilities in the game industry as described in claim 3, characterized in that, The step of performing time-frequency analysis on the feature duration information based on continuous wavelet transform to obtain the operational wavelet feature information includes: Based on the target object profile data, extract the baseline operation frequency information of this type of game character; Using the Morlet wavelet as the target wavelet mother function and the reference operating frequency information as the scale constraint, the wavelet transform parameters are initialized. Perform a continuous wavelet transform on the first input feature duration curve to obtain the first input wavelet coefficient matrix; perform a continuous wavelet transform on the first deviation feature duration curve to obtain the first deviation wavelet coefficient matrix; The first input wavelet coefficient matrix and the first deviation wavelet coefficient matrix are combined and stored as the first wavelet analysis data set. The feature duration information is traversed to perform continuous wavelet transform, and multiple wavelet analysis data groups are obtained and stored as wavelet analysis datasets.

5. The method for assessing talent capabilities in the game industry as described in claim 4, characterized in that, The step of performing time-frequency analysis on the feature duration information based on continuous wavelet transform to obtain the operated wavelet feature information further includes: Based on the wavelet analysis data set, the wavelet energy density is calculated to obtain an energy density matrix set, which includes an input energy density matrix and a bias energy density matrix. Based on the energy density matrix set, a time-frequency spectrum set is plotted with time as the first coordinate axis and the scale constraint as the second coordinate axis; Integrate the first wavelet analysis data set using the modulus square of the wavelet coefficient matrix to obtain the power spectrum set; The time-frequency spectrum set and the power spectrum set are stored to obtain operational wavelet feature information.

6. The method for assessing talent capabilities in the game industry as described in claim 5, characterized in that, The construction of the game talent operation ability assessment model also includes: Collect historical evaluation records from game industry evaluation experts to obtain a basic evaluation sample set; Based on industry big data and combined with target game character characteristic data, an expanded evaluation sample set is generated; By merging the basic assessment sample set and the expanded assessment sample set, a training sample set is obtained, wherein the training sample set includes sample wavelet feature information, sample operation rhythm component information, sample operation complexity order information, and sample practical ability level. Based on the training sample set, construct and train the game talent operation ability assessment model.

7. The method for assessing talent capabilities in the game industry as described in claim 6, characterized in that, The process of matching game-related positions to the target assessment subjects based on the practical skills rating information includes: The practical ability rating information includes specific ability scores corresponding to multiple key observation points, and the multiple key observation points are marked with corresponding ability dimension labels. Based on the hiring standards of game companies, a talent response matrix is ​​defined, which includes multiple sets of ability scoring ranges and related job recruitment or training suggestions. Based on the talent response matrix, the practical ability rating information is traversed and matched, corresponding suggested solutions are matched, and the evaluation results of the target assessment object are output.

8. A talent competency assessment system for the game industry, characterized in that, The system includes: The profile preparation module is used to acquire basic profile data of the target assessment object, divide the task into stages according to the preset assessment checkpoint system, determine the key observation point set, and deploy the practical data collection terminal. The practical data acquisition module is used to start the practical data acquisition terminal and collect the real-time game operation data of the target evaluation object. The real-time game operation data includes operation input timing data and skill release deviation data. The feature curve module is used to parse the real-time game operation data, extract the input frequency features and operation deviation features of multiple observation points in the key observation point set, construct the input frequency duration curve and the deviation feature duration curve, and output the feature duration information. The wavelet time-frequency module is used to perform time-frequency analysis on the feature duration information based on continuous wavelet transform to obtain operation wavelet feature information, wherein the operation wavelet feature information includes operation time-frequency distribution feature map and operation power spectral density map. The model rating module is used to construct a game talent operation ability assessment model. Taking the operation wavelet feature information as input, the game talent operation ability assessment model is run to perform a quantitative assessment of the ability and output practical ability rating information. The matching feedback module is used to perform game job matching feedback for the target assessment object based on the practical ability rating information.

9. A talent assessment device for the game industry, characterized in that, The device includes: a memory, a processor, and a game industry talent competency assessment program stored on the memory and executable on the processor, the game industry talent competency assessment program being configured to implement the steps of the game industry talent competency assessment method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a game industry talent competency assessment program, which, when executed by a processor, implements the steps of the game industry talent competency assessment method as described in any one of claims 1 to 7.

Images