Special population auxiliary employment service method, device and equipment and storage medium

Abstract

The present disclosure provides a special population auxiliary employment service method, device, equipment and storage medium, the method comprises: obtaining the employment problem of a user, and identifying the intention corresponding to the employment problem; based on the intention, automatically obtaining the multi-dimensional portrait required to reply to the employment problem, wherein the multi-dimensional portrait comprises at least one of personal characteristic portrait reflecting the specific needs of the user, post portrait reflecting the post demand, enterprise environment portrait reflecting the enterprise environment and policy portrait; based on the intention and the multi-dimensional portrait, screening the skill module suitable for replying to the employment problem from the pre-constructed skill pool; based on the screened skill module, generating the reply to the employment problem and pushing it to the user. The method of the present disclosure can improve the level of auxiliary employment service for special population.

Description

Technical Field

[0001] This disclosure relates to the field of artificial intelligence technology, and in particular to a method, apparatus, device, and storage medium for assistive employment services for special populations. Background Technology

[0002] Currently, AI agents are gradually being introduced into assisted employment services for special populations. Existing systems first obtain user input through dialogue, then perform intent recognition based on the context. Subsequently, based on the recognized intent, they match and invoke the corresponding skill module from a pre-set skill pool to execute the corresponding assisted employment service. However, these systems rely solely on general intent recognition results during skill scheduling. This single-dimensional scheduling mechanism means that the invoked skill modules cannot well adapt to the specific needs of special populations, resulting in low accuracy of assisted employment services. Therefore, improving the level of assisted employment services for special populations has become a pressing technical problem for those skilled in the art. Summary of the Invention

[0003] In view of this, this disclosure proposes a method, apparatus, equipment and storage medium for assisted employment services for special populations, which can improve the level of assisted employment services for special populations.

[0004] According to a first aspect of this disclosure, a method for providing assisted employment services to special populations is provided, comprising: The system obtains information about a user's employment status and identifies the intent corresponding to that employment status, wherein the user is a special group of people with certain physical disabilities. Based on the stated intent, a multi-dimensional profile required to answer the employment question is automatically obtained, wherein the multi-dimensional profile includes at least one of the following: a personal characteristic profile reflecting the user's specific needs, a job profile reflecting the job requirements, a corporate environment profile reflecting the corporate environment, and a policy profile. Based on the intent and the multi-dimensional profile, skill modules suitable for answering the employment question are selected from a pre-built skill pool, wherein the skill modules are computing modules that perform specific business tasks; Based on the selected skill modules, a response to the employment question is generated and pushed to the user.

[0005] In one possible implementation, when selecting skill modules suitable for answering the employment question from a pre-built skill pool based on the intent and the multi-dimensional profile, the process includes: Based on the intent, multiple first candidate skill modules that fit the intent are selected from the skill pool; Calculate the matching degree between each first candidate skill module and the multidimensional profile, and based on the matching degree between each first candidate skill module and the multidimensional profile, select a set number of second candidate skill modules with the highest matching degree from each first candidate skill module; The final skill module that is suitable for answering the employment question is determined from the second candidate skill modules.

[0006] In one possible implementation, calculating the matching degree between the first candidate skill module and the multidimensional profile includes: Obtain the module multi-dimensional profile corresponding to the first candidate skill module and the historical evolution weight of the module multi-dimensional profile under the multi-dimensional profile; Calculate the initial matching degree between the module's multidimensional profile and the multidimensional profile; The initial matching degree is revised using the historical evolution weight to obtain the final matching degree between the module multidimensional profile and the multidimensional profile; The final matching degree between the module's multidimensional profile and the multidimensional profile is taken as the matching degree between the first candidate skill module and the multidimensional profile.

[0007] In one possible implementation, the historical evolution weights are dynamically adjusted based on user feedback data.

[0008] In one possible implementation, adjusting the historical evolution weights based on user feedback data includes: Determine the sentiment tendency of the feedback data, wherein the sentiment tendency of the feedback data includes at least one of negative feedback, neutral feedback, and positive feedback; The latest historical evolution weights are adjusted based on the sentiment trends of the feedback data.

[0009] In one possible implementation, determining the final skill module suitable for answering the employment question from the second candidate skill modules includes: Based on the personal feature profile in the multi-dimensional profile, the user's physical functional constraints are determined; Based on the user's physical functional constraints, the final skill module suitable for answering the employment question is determined from the second candidate skill modules.

[0010] In one possible implementation, when generating an answer to the employment question based on the selected skill modules and pushing it to the user, it includes: Determine the execution order of each of the selected skill modules; The skill modules are invoked sequentially to generate answers to the employment questions. The response will be pushed and displayed to the user.

[0011] According to a second aspect of this disclosure, a device for assistive employment for a special population is provided, comprising: The data acquisition module is used to acquire users' employment questions and identify the intentions corresponding to the employment questions, wherein the users are special groups with certain physical disabilities; The profile association module is used to automatically obtain a multi-dimensional profile required to answer the employment question based on the intent. The multi-dimensional profile includes at least one of the following: a personal characteristic profile reflecting the user's specific needs, a job profile reflecting the job requirements, a corporate environment profile reflecting the corporate environment, and a policy profile. A filtering module is used to filter out skill modules that are suitable for answering the employment question from a pre-built skill pool based on the intent and the multi-dimensional profile, wherein the skill module is a computing module that performs a specific business task; The execution module generates answers to the employment questions based on the selected skill modules and pushes them to the user.

[0012] According to a third aspect of this disclosure, an assistive employment device for special populations is provided, comprising: a processor; a memory for storing processor-executable instructions; wherein the processor is configured to perform the method described in the first aspect of this disclosure.

[0013] According to a fourth aspect of this disclosure, a non-volatile computer-readable storage medium is provided that stores computer program instructions thereon, wherein the computer program instructions, when executed by a processor, implement the method described in the first aspect of this disclosure.

[0014] This disclosure provides a method, apparatus, device, and storage medium for assisted employment services for special populations. The method includes: acquiring a user's employment question and identifying the intent corresponding to the employment question; automatically acquiring a multi-dimensional profile required to answer the employment question based on the intent, wherein the multi-dimensional profile includes at least one of a personal characteristic profile reflecting the user's specific needs, a job profile reflecting job requirements, a corporate environment profile reflecting the corporate environment, and a policy profile; selecting skill modules suitable for answering the employment question from a pre-built skill pool based on the intent and the multi-dimensional profile; and generating an answer to the employment question based on the selected skill modules and pushing it to the user. Because this disclosure, in addition to identifying the intent, further acquires a multi-dimensional profile including a personal characteristic profile, a job profile, a corporate environment profile, and a policy profile, and selects skill modules based on the intent and the multi-dimensional profile, the skill modules invoked can more accurately adapt to the personalized needs of special populations, thus improving the level of assisted employment services for special populations.

[0015] Other features and aspects of this disclosure will become clear from the following detailed description of exemplary embodiments with reference to the accompanying drawings. Attached Figure Description

[0016] The accompanying drawings, which are included in and form part of this specification, illustrate exemplary embodiments, features, and aspects of this disclosure together with the specification and serve to explain the principles of this disclosure.

[0017] Figure 1 A flowchart illustrating a method for providing assisted employment services to special populations according to an embodiment of the present disclosure is shown. Figure 2 A schematic block diagram of an assisted employment device for special populations according to an embodiment of the present disclosure is shown; Figure 3 A schematic block diagram of an assisted employment device for special populations according to an embodiment of the present disclosure is shown. Detailed Implementation

[0018] Various exemplary embodiments, features, and aspects of this disclosure will now be described in detail with reference to the accompanying drawings. The same reference numerals in the drawings denote elements that have the same or similar functions. Although various aspects of the embodiments are shown in the drawings, they are not necessarily drawn to scale unless specifically indicated otherwise.

[0019] The term “exemplary” as used herein means “serving as an example, embodiment, or illustration.” Any embodiment illustrated herein as “exemplary” is not necessarily to be construed as superior to or better than other embodiments.

[0020] Furthermore, to better illustrate this disclosure, numerous specific details are set forth in the following detailed description. Those skilled in the art will understand that this disclosure can be practiced without certain specific details. In some instances, methods, means, components, and circuits well known to those skilled in the art have not been described in detail in order to highlight the main points of this disclosure.

[0021] <Method Implementation> Figure 1 A flowchart illustrating a method for providing assisted employment services to special populations according to an embodiment of this disclosure is shown. Figure 1 As shown, the method includes steps S1100-S1400.

[0022] S1100 obtains the user's employment questions and identifies the intent corresponding to the employment questions.

[0023] First of all, it should be noted that the special population in this disclosure refers to people with disabilities who have certain physical functional impairments, and the users served by this disclosure are mainly such special populations.

[0024] When a user needs employment services, they can input the employment question they need to consult into the system that implements the methods disclosed herein. After the employment question is input into the system, the system can automatically obtain the user's employment question.

[0025] After obtaining the user's employment questions, a pre-trained intent recognition model can be used to identify the intent behind the questions. This intent can include job searches, salary calculations, and inquiries about subsidy policies for specific groups.

[0026] S1200 automatically obtains a multi-dimensional profile needed to answer an employment question based on the intent corresponding to the employment question. The multi-dimensional profile includes at least one of the following: a personal characteristic profile reflecting the user's specific needs, a job profile reflecting the job requirements, a corporate environment profile reflecting the corporate environment, and a policy profile reflecting the policy content.

[0027] First and foremost, it should be noted that the system has pre-set databases for personal characteristics, job requirements, corporate environment, and policies.

[0028] This personal profile database stores the personal profiles of all registered users. Each user's profile includes their disability category, disability level, physical limitations, and other physical characteristics. User profiles can be dynamically generated based on user data retrieved from the Disabled Persons' Federation or registration information submitted by the user.

[0029] This job requirement profile database stores the job requirement profiles for all hiring units listed in the system. Each job requirement profile includes job skill requirements, environmental adaptability requirements, physical exertion levels, and other job characteristics. The job skill requirements encompass both hard skills and soft skills. Hard skills include, but are not limited to, specialized operational abilities that can be directly learned through training or hands-on practice, such as computer operation, sign language communication, text entry, and product assembly. Soft skills include, but are not limited to, comprehensive professional qualities involving interpersonal interaction and self-regulation, such as teamwork, emotional stability, and time management. Environmental adaptability requirements include, but are not limited to, work posture requirements (e.g., prolonged sitting, prolonged standing, alternating sitting and standing), spatial mobility requirements (e.g., whether it requires traveling between multiple locations), sensory work environment (e.g., whether it involves strong light stimulation, high noise environments, low light conditions, strong screen dependence), auxiliary equipment compatibility (e.g., whether it supports screen reader software, voice input, hearing aid compatibility), and the feasibility of remote / home work (e.g., whether fully online work is allowed). The physical exertion level includes a quantifiable index grading the overall physical exertion intensity of the job, specifically divided into three levels: mild, moderate, and severe. Mild refers to jobs that primarily involve seated or static operations, requiring minimal standing, walking, or moving of items, resulting in minimal physical exertion; examples include data labeling, online customer service, and text proofreading. Moderate refers to jobs that require intermittent standing, short-distance walking, or handling light items, resulting in moderate physical exertion; examples include receptionists, warehouse clerks, and cashiers. Severe refers to jobs that require prolonged standing, frequent walking, repeated bending, or handling heavy objects, resulting in significant physical exertion; examples include production line operators, express delivery sorters, and cleaning staff. The job requirement profile for this advertised position can be dynamically generated based on the job description and the actual working environment. For example, for a job posting for "Data Labeler" by the "Data Labeling Department of a certain technology company", the job requirements profile may include: skill requirements are "text annotation, image selection, and label classification", environment requirements are "purely online operation, no need to move, support for screen reader software, and allowance for remote work from home", and physical exertion level is "light".

[0030] This enterprise environment profile database stores the enterprise environment profiles of all employers included on the platform. Each enterprise environment profile includes information such as the nature of the enterprise (e.g., state-owned enterprises, private enterprises, social welfare enterprises, etc.), industry category (e.g., information technology, manufacturing, service industries, etc.), accessibility facilities configuration (e.g., the availability of facilities and services such as accessible pathways, accessible elevators, tactile paving, wheelchair ramps, low-level service counters, accessible restrooms, sign language interpretation services, screen reader support, etc.), employment history of disabled employees (e.g., the number of positions previously filled with disabled employees, the distribution of disability categories, retention rate, and turnover rate, reflecting the enterprise's friendliness towards disabled employees), and social security and housing provident fund contribution standards, among other enterprise environment characteristics. Enterprise environment profiles can be dynamically generated based on multi-source data, including enterprise registration information, on-site survey data, historical employment records, and publicly available government enterprise credit information. For example, the corporate environment profile of a technology-based private enterprise may include: the enterprise nature is "private enterprise", the industry category is "information technology service industry", the accessibility facilities configuration is "wheelchair ramps, accessible elevators, and low-level service counters", the history of employing disabled people is "a total of 12 visually impaired employees and 8 physically disabled employees have been hired in the past three years, with a retention rate of 85%", and the social security and housing provident fund contribution standards are "paid according to the local minimum contribution base, with a housing provident fund ratio of 5%".

[0031] This policy profile database stores various policy profiles related to supported employment for people with disabilities. Each policy profile includes policy characteristics such as applicable conditions, a summary of policy content, policy validity period, scope of applicable groups, application procedures, and required materials. Policy profiles can be automatically generated based on government-issued policy documents using natural language processing and structured information extraction technology, or they can be manually annotated and reviewed by platform policy operators before being entered into the database. For example, a policy profile for a "Supported Employment Subsidy Policy for People with Disabilities" might include: applicable area "a certain municipal district"; applicable group "severely physically disabled persons holding a People's Republic of China Disability Certificate and having been stably employed in a supported employment institution for at least 3 months"; policy content "a monthly subsidy of 500 yuan per person"; application condition "the employer has paid social insurance for the disabled person"; application procedure "the employer submits an application form, a copy of the disability certificate, proof of social insurance payment, etc. to the district disabled persons' federation"; and policy validity period "from January 1, 2025 to December 31, 2027".

[0032] The aforementioned personal characteristic profile database, job requirement profile database, enterprise environment profile database, and policy profile database are physically decoupled from each other and maintained independently. Each database maintains the accuracy and timeliness of the data through different data sources and update mechanisms.

[0033] When the system is running, upon receiving a user's real-time employment question, the pre-built intent recognition model first identifies the intent corresponding to the employment question, then determines the multi-dimensional profile required for that intent, and finally extracts the corresponding multi-dimensional profile from the corresponding profile library. Through this "physically decoupled, runtime dynamically assembled" architecture design, the system can flexibly respond to different combinations of multi-dimensional profile data required in different business scenarios.

[0034] S1300, based on intent and multi-dimensional profiles, selects skill modules from a pre-built skill pool that are suitable for answering employment-related questions.

[0035] First, it's important to note that the system has a skill pool containing pre-defined skill modules. These modules are the smallest functional units the system can invoke to perform specific business tasks. Within this pool, the same business task may have multiple skill modules with the same functional type but different applicable conditions. For example, under the task of "salary calculation," the skill pool includes a standard salary calculation module for general employees, a composite salary calculation module for severely disabled individuals (automatically linked to self-reliance allowances and social security reductions), and a special salary calculation module for blind massage therapists (including hourly rates and special subsidies). While these skill modules output similar business results, their internal calculation logic and the corresponding user profiles differ. A difference in any user profile dimension may result in different applicable skill modules.

[0036] In one possible implementation, when selecting skill modules suitable for answering employment questions from a pre-built skill pool based on intent and multi-dimensional profiles, the following steps may be included: Step 1: Based on the user's intent related to the employment problem, select multiple first-line candidate skill modules from the skill pool that match the intent. Specifically, based on the user's intent related to the employment problem, recall all skill modules from the skill pool that can perform the task corresponding to that intent as multiple first-line candidate skill modules that match the intent.

[0037] Step 2: Calculate the matching degree between each first candidate skill module and the multidimensional profile required to answer employment questions, and based on the matching degree between each first candidate skill module and the multidimensional profile required to answer employment questions, select the set number of second candidate skill modules with the highest matching degree from each first candidate skill module.

[0038] For each first candidate skill module, the following steps may be included when calculating the matching degree between the first candidate skill module and the multidimensional profile required to answer employment questions: First, obtain the multi-dimensional profile of the first candidate skill module and the historical evolution weight of the first candidate skill module under the multi-dimensional profile required to answer employment questions.

[0039] First, it's important to clarify that each skill module in the skill pool has a pre-configured multi-dimensional profile. This profile characterizes the range of data dimensions applicable to the skill module when performing tasks. Specifically, this multi-dimensional profile is modeled using a unified dimensional system with the underlying multi-dimensional data profile layer of the system. It includes a description of the compatibility of at least one of the following dimensions: the personal profile features adapted to the skill module, the associated job requirement profile features, the dependent company profile features, and the corresponding policy profile features (such as the applicable local policy scope and social security and housing provident fund contribution standards). During the calculation of this step, the multi-dimensional profile configured for the first candidate skill module can be directly retrieved from the system.

[0040] Furthermore, the system records the historical evolution weights of each skill module in the skill pool under different multi-dimensional profile combinations. These historical evolution weights are quantified values ​​of adaptation accumulated based on user feedback after the skill module performs tasks for a specific multi-dimensional profile combination during its historical service process. They characterize the matching reliability between the skill module and the corresponding multi-dimensional profile combination. Specifically, whenever a skill module first serves a multi-dimensional profile combination, the system generates an initial historical evolution weight value for that combination. Subsequently, each time the skill module serves the same multi-dimensional profile combination again, the system dynamically adjusts the historical evolution weights for that combination based on user feedback obtained during this service process, thus continuously updating the matching reliability. During the calculation of this step, the historical evolution weights of the first candidate skill module under the multi-dimensional profile combination required to answer the current employment question can be directly obtained from the system.

[0041] In one possible implementation, the historical evolution weight of each skill module under a specific multi-dimensional profile is dynamically adjusted based on user feedback data. The specific adjustment steps can be as follows: First, obtain user feedback data for the skill module under a specific multi-dimensional profile. This feedback data refers to subsequent interactive behavior data and / or explicit evaluation data generated after the user outputs the execution result in the skill module, including but not limited to: the user's text input content, the user's page operation behavior, the user's explicit evaluation of the output result (such as clicking "helpful," "useless," "like," "report error," etc.), the user's session duration, and the session termination method.

[0042] Secondly, the sentiment tendency of the feedback data is determined. The sentiment tendency of the feedback data includes at least one of negative, neutral, and positive feedback. Specifically, the collected user feedback data can be input into a pre-trained sentiment analysis model, which then outputs the sentiment tendency of the user feedback. The logic for determining the sentiment tendency of the feedback data by the sentiment analysis model is as follows: Situations indicating a negative sentiment feedback include at least one of the following: 1) The user continues to ask follow-up questions on the same topic in different ways (e.g., rephrasing the question and asking again); or pre-defined negative sentiment words are identified in the user's input text (e.g., "I don't understand," "Wrong," "Incorrect," "Completely unusable," "Useless," "Bad review," etc.). 2) The user clicks the "Refresh" or "Recommend" control; or the user clicks the "Report Error" or "Complaint" control. 3) The user directly closes the current chat window.

[0043] The following are scenarios where the feedback is determined to be neutral: 1) The user's topic changes to a new topic unrelated to the current skill module's execution result. 2) The user does not perform any explicit evaluation action after receiving the output result. 3) The user silently accepts the current result and does not generate any subsequent interaction within a preset time window (e.g., within 1 minute), and the conversation ends naturally or times out.

[0044] Situations where the emotional tendency is determined to be positive feedback include at least one of the following: 1) Preset positive emotional words are identified in the text content entered by the user (such as confirmation statements such as "Understood," "Thank you," "Okay," "This is it," etc.). 2) The user clicks positive evaluation controls such as "Thank you," "Helpful," "Like," or "Favorite"; or the user stays on the results page output by the skill module for more than a preset time (such as 30 seconds) and performs further actions (such as clicking "Apply for a position," "Download attachments," "View more," etc.). 3) After viewing the results, the user actively uses positive interaction (such as clicking "Confirm Completion," selecting "Continue Using," etc.) as a way to end the conversation.

[0045] Next, based on the sentiment of the feedback data, the latest historical evolution weights are adjusted. Specifically, if the user provides positive feedback, the historical evolution weight between the skill module and the current multi-dimensional profile combination is increased from its original value to enhance the scheduling priority of the skill module when encountering the same or similar multi-dimensional profile combinations in the future. If the user provides negative feedback, the historical evolution weight is decreased from its original value to reduce the scheduling priority of the skill module in the same or similar scenarios in the future. If the user provides neutral feedback, the historical evolution weights are not adjusted.

[0046] In one possible implementation, the adjustment formula used by the skill module to reduce the historical evolution weight under a specific multidimensional profile when receiving negative feedback is as follows: In the formula, The negative reward threshold For updated historical evolution weights, This refers to the historical evolution weights currently stored in the system. The negative reward threshold can be set based on the severity of the user's negative behavior. For example, if it's mild dissatisfaction (such as the user rephrasing the same topic or clicking "change batch"), then... =First negative adjustment value 1. If it is a moderate obstacle (such as user input containing negative words such as "cannot understand", "incorrect", "cannot operate at all", etc.), then another =Second negative adjustment value 2. If it's a severe conflict (e.g., the user directly closes the current dialog box or clicks "Report Error"), then... = The third negative adjustment value is 3.5. Wherein, the first negative adjustment value... Second negative adjustment value The third negative adjustment value is preferably the first negative adjustment value = 1, the second negative adjustment value = 2, and the third adjustment value = 3.5.

[0047] In one possible implementation, the adjustment formula used by the skill module to increase the historical evolution weight under a specific multi-dimensional profile when receiving positive feedback is as follows: In the formula, This is the positive reward threshold. This threshold can be set based on the intensity of the user's positive behavior. For example, if the user's positive behavior is an explicit, proactive like / favorite, indicating a very good solution, then an additional reward threshold can be set. = First positive adjustment value 2. If the user's positive behavior implicitly adopts / remains, then another... = Second positive adjustment value 1. If the user's positive behavior is silently defaulted, then it can be set to 1. =The third positive adjustment value, where the first positive adjustment value Second positive adjustment value The third positive adjustment value is preferably the first positive adjustment value = 2, the second positive adjustment value = 1, and the third adjustment value = 0.5.

[0048] Second, calculate the initial matching degree between the module's multi-dimensional profile and the multi-dimensional profile required to answer employment questions. Specifically, first, map the module's multi-dimensional profile and the multi-dimensional profile required to answer employment questions to a unified vector space to obtain the feature vectors of the module's multi-dimensional profile and the multi-dimensional profile required to answer employment questions. Then, calculate the cosine similarity between the two feature vectors as the initial matching degree between the module's multi-dimensional profile and the multi-dimensional profile required to answer employment questions.

[0049] Third, the initial matching degree is revised using historical evolution weights to obtain the final matching degree between the module's multidimensional profile and the multidimensional profile required to answer employment questions. Specifically, the product between the initial matching degree and the historical evolution weights is calculated, and this product is used as the final matching degree between the module's multidimensional profile and the multidimensional profile required to answer employment questions.

[0050] Fourth, the final matching degree between the multi-dimensional profile of the module and the multi-dimensional profile required to answer employment questions is used as the matching degree between the first candidate skill module and the multi-dimensional profile.

[0051] By introducing evolution weights for each skill module, the scheduling of skill modules can have a self-evolutionary capability based on historical interaction feedback. That is, by using historical evolution weights as a product factor to weight and correct the initial matching degree, skill modules with high-frequency positive feedback can obtain higher final matching degree scores in homogeneous portrait scenarios, thereby achieving positive reinforcement and negative suppression of the scheduling strategy and effectively avoiding the repeated use of ineffective skills.

[0052] By referring to the above method, the matching degree between each first candidate skill module and the multi-dimensional profile required to answer employment questions can be calculated. Then, the set number of second candidate skill modules with the highest matching degree can be selected from each first candidate skill module.

[0053] Step 3: Determine the final skill module from the second candidate skill modules that best matches the answers to employment questions. This may specifically include the following steps: First, based on the personal feature profile in the multi-dimensional profile, the user's physical functional constraints are determined. Specifically, the user's personal profile features are mapped to a standard physical functional constraint space to obtain the user's physical functional constraints.

[0054] First, it should be noted that the standard physical functional constraint space refers to a standardized, multi-dimensional functional requirement / obstacle type description space predefined by the system, used to uniformly represent the physical functional constraint status of job seekers. This standard physical functional constraint space includes multiple constraint dimensions, each corresponding to a type of core physical function. The standard physical functional constraint space can include at least one of visual constraints, hearing constraints, physical constraints, cognitive constraints, and language constraints. In a specific embodiment, the standard physical functional constraint space = [visual constraints, hearing constraints, physical constraints, cognitive constraints].

[0055] The process of mapping a user's personal profile features to a standard physical functional constraint space to obtain the user's physical functional constraints can include the following steps: First, extract key functional attributes from the personal profile. Then, map the key functional attributes to the values ​​of the corresponding dimensions in the standard physical functional constraint space, and arrange these values ​​in order to obtain the user's physical functional constraints.

[0056] In one possible implementation, mapping key functional attributes to values ​​of corresponding dimensions in the standard physical functional constraint space is based on a preset attribute-dimension mapping rule. This attribute-dimension mapping rule records: 1) Constraint space dimensions corresponding to various key functional attributes. For example, attributes such as "visual impairment", "visually impaired", "blind person", and "unable to recognize screen graphics" correspond to the "visual constraint dimension"; attributes such as "hearing impairment", "hearing impairment", and "deaf person" correspond to the "hearing constraint dimension"; attributes such as "limb disability", "lower limb mobility impairment", "inability to stand", "requires wheelchair", and "inability to stand for long periods" correspond to the "limb constraint dimension"; and attributes such as "intellectual disability", "mental disability", and "cognitive impairment" correspond to the "cognitive constraint dimension".

[0057] 2) Value rules for various key functional attributes in the corresponding dimensions of the constraint space: When a key functional attribute corresponding to a constraint space dimension exists in the user profile, the value of that constraint space dimension is set to the first preset value (e.g., 1), indicating that there is a functional constraint; when a key functional attribute corresponding to a constraint space dimension does not exist in the user profile, the value of that constraint space dimension is set to the second preset value (e.g., 0), indicating that there is no functional constraint.

[0058] In an embodiment where the standard physical function constraint space = [visual constraint, hearing constraint, limb constraint, cognitive constraint], the preset attribute-dimension mapping rule can be as follows: If the user profile contains key functional attributes related to visual function, such as "visual impairment", "visual disability", "blind person", or "inability to recognize screen graphics", then the value of the visual constraint in the standard physical functional constraint space is set to 1 (indicating the existence of a visual constraint); otherwise, it is marked as 0.

[0059] If the user profile contains key functional attributes related to hearing function, such as "hearing disability," "hearing impairment," or "deafness," then the value of the hearing constraint in the standard physical functional constraint space is set to 1 (indicating the existence of a hearing constraint); otherwise, it is marked as 0.

[0060] If the user profile contains key functional attributes related to limb movement function, such as "limb disability", "lower limb mobility impairment", "inability to stand", "requires wheelchair travel", or "inability to stand for long periods or move frequently", then the value of the limb constraint in the standard physical functional constraint space is set to 1 (indicating the existence of limb constraint); otherwise, it is marked as 0.

[0061] If the user profile contains key functional attributes related to cognitive function, such as "intellectual disability," "mental disability," or "cognitive impairment," then the value of the cognitive constraint in the standard physical functional constraint space is set to 1 (indicating the existence of a cognitive constraint); otherwise, it is marked as 0.

[0062] After extracting key functional attributes from the personal profile, the preset attribute-dimension mapping rules can map these key functional attributes to the corresponding dimensions in the standard physical functional constraint space, thus obtaining the user's physical functional constraints. For example, suppose a job seeker is a "Level 1 visually impaired" person. The system extracts the key functional attribute "visual impairment" from the user's candidate profile. According to the above mapping rules: visual constraint dimension: visual constraint exists → marked as 1; hearing constraint dimension: no hearing-related constraints → marked as 0; limb constraint dimension: no limb-related constraints → marked as 0; cognitive constraint dimension: no cognitive-related constraints → marked as 0; then the final generated physical functional constraint is: Muser=[1,0,0,0].

[0063] Second, based on the user's physical functional constraints, the final skill module suitable for answering employment questions is determined from the second candidate skill modules.

[0064] First, it's important to clarify that each skill module in the system's functional pool is pre-configured with a functional requirement (i.e., the physical functional requirements the user needs to possess) matrix, Mskill. This functional requirement matrix and the user's physical functional constraint vector reside in the same standard physical functional constraint space, possessing the same dimensions and sorting order. The values ​​of each dimension are determined based on the objective dependence of the skill module on the user's physical functions during execution, specifically including the following judgment principles: If the execution result of the skill module is presented in a visual interface and the user cannot obtain the same information through alternative means (such as screen reader software, voice assistance, etc.), then the skill module is determined to be highly dependent on visual function, and the corresponding dimension is set to 1; otherwise, the value is 0.

[0065] If the execution of this skill module depends on the reception of voice commands or the acquisition of audio information, and the user cannot complete the interaction through alternative methods (such as subtitles, text prompts, etc.), then the skill module is determined to be strongly dependent on the hearing function, and the corresponding dimension is set to 1; otherwise, the value is 0.

[0066] If the execution of the skill module requires the user to perform physical operations (such as clicking, dragging, standing, walking, etc.), and cannot be completed by alternative methods (such as voice control, eye tracking control, assistive devices, etc.), then the skill module is determined to be highly dependent on physical functions, and the corresponding dimension is set to 1; otherwise, the value is 0.

[0067] If the execution process of the skill module involves complex information understanding, logical reasoning, or decision-making, and exceeds the user's cognitive ability, then the skill module is determined to be highly dependent on cognitive function, and the corresponding dimension is set to 1; otherwise, the value is 0.

[0068] In an embodiment where the standard physical functional constraint space is a 4-dimensional space [visual constraint, hearing constraint, limb constraint, cognitive constraint], the system has the following three candidate skill modules: Skill Module A (Speech-to-Text Conversion Assistant): This skill delivers information to the user via voice broadcast. It does not rely on the user's visual function, but requires the user to have normal hearing to receive voice information. Therefore, the functional requirement matrix Mskill-A = [0, 1, 0, 0] configured for this skill module indicates that this skill module strongly relies on hearing and does not strongly rely on visual, physical, or cognitive functions.

[0069] Skill Module B (Pure Visual Image Annotation Tool): This skill requires the user to identify and annotate graphic information on the screen, heavily relying on the user's visual function. Therefore, the function requirement matrix Mskill-B = [1, 0, 0, 0] is configured for this skill module, indicating that this skill heavily relies on visual function and not heavily relies on hearing, physical, and cognitive functions.

[0070] Skill Module C (Text Policy Interpretation Agent): This skill outputs policy interpretation results in text form. Users can use it in conjunction with auxiliary tools such as screen reader software. It does not heavily rely on any one of the following: vision, hearing, physical or cognitive functions. Therefore, the function requirement matrix Mskill-C is configured for this skill module matrix as [0, 0, 0, 0].

[0071] When determining the final skill module suitable for answering the employment question from the second candidate skill modules based on the user's physical functional constraints, the following operations are performed for each second candidate skill module: The corresponding functional requirement matrix Mskill is extracted; the Mskill matrix is ​​compared with the user's physical functional constraints Muser; and the comparison result is used to determine whether there is a conflict between the skill requirements of the second candidate skill module and the user's physical functional capabilities. If a conflict exists, the second candidate skill module is filtered out; otherwise, it is retained. After all second candidate skill modules have undergone the comparison of their Mskill matrix with the user's physical functional constraints Muser, the retained second candidate skill modules are used as the final skill modules suitable for answering the employment question.

[0072] In one possible implementation, when comparing the functional requirement matrix Mskill corresponding to the second candidate skill module with the user's physical functional constraint Muser to determine whether there is a conflict between the skill requirement of the second candidate skill module and the user's physical functionality, the following steps may be included: comparing the values ​​of each corresponding dimension in the functional requirement matrix Mskill and the user's physical functional constraint Muser one by one; if the value of one dimension is consistent, it is determined whether there is a conflict between the skill requirement of the skill module and the user's physical functionality; if the values ​​of all dimensions are inconsistent, it is determined that there is no conflict between the skill requirement of the skill module and the user's physical functionality.

[0073] In another possible implementation, when determining the final skill module suitable for answering the employment question from the second candidate skill modules based on user-based physical function constraints, the following steps may be included: First, for each second candidate skill module, the corresponding functional requirement matrix Mskill is extracted. This Mskill matrix is ​​then compared with the user's physical functional constraints Muser. The comparison result determines whether there is a conflict between the skill requirement of the second candidate skill module and the user's physical functional constraints. Based on the conflict determination, a mask value is generated for that second candidate skill module. Specifically, if there is no conflict, the mask value is 1, indicating that the skill module needs to be retained; if there is a conflict, the mask value is 0, indicating that the skill module needs to be filtered out. Following this method, the mask value for each second candidate skill module can be obtained, and mask vectors for all second candidate modules are generated according to a specified order. For example, if the masks calculated for skill modules A, B, and C are 0, 1, and 0 respectively, then the mask vector for all second candidate modules is [0, 1, 0].

[0074] Furthermore, the final matching degree of each second candidate skill module with the multi-dimensional profile required to answer the employment question is obtained, and the matching degree score vector of all second candidate modules is generated in the same specified order. For example, if the final matching degrees of skill module A, skill module B, and skill module C with the multi-dimensional profile required to answer the employment question are 0.85, 0.92, and 0.78 respectively, then the matching degree score vector of all second candidate modules is [0.85, 0.92, 0.78].

[0075] Finally, the mask vectors of all second candidate modules and the matching score vectors of all second candidate modules are plotted. Skill modules with plotting results that are not 0 are retained as the final skill modules for matching the employment question. Continuing with the previous example, the plotting result is [0, 0.92, 0], which means that among skill modules A, B, and C, only skill module B is retained as the final skill module for matching the employment question.

[0076] In this embodiment, a physical safety barrier mechanism for AI skill scheduling in the context of supported employment for people with disabilities is constructed by introducing the user's physical functional constraints to screen the final skill modules for answering employment questions. Based on the core physical functional attributes of job seekers with disabilities, such as disability type and level, this mechanism establishes an insurmountable hard constraint filter at the front end of the skill scheduling chain. This intercepts any skill modules that are physiologically incompatible with the user from the final candidate set in advance, thereby fundamentally ensuring the absolute security and accessibility of the scheduling results at the physical execution level and avoiding interaction failures or usage obstacles caused by mismatch between skills and the user's physical condition.

[0077] S1400, based on the selected skill modules, generates answers to employment questions and pushes them to the user. Specifically, it determines the execution order of the selected skill modules; calls each skill module in sequence to generate answers to employment questions; and pushes the answers to the user.

[0078] This disclosure provides a method for providing assisted employment services to special populations, including: obtaining a user's employment question and identifying the intent corresponding to the employment question; automatically obtaining a multi-dimensional profile required to answer the employment question based on the intent, wherein the multi-dimensional profile includes at least one of a personal characteristic profile reflecting the user's specific needs, a job profile reflecting job requirements, a corporate environment profile reflecting the corporate environment, and a policy profile; selecting skill modules suitable for answering the employment question from a pre-built skill pool based on the intent and the multi-dimensional profile; and generating an answer to the employment question based on the selected skill modules and pushing it to the user. Because this disclosure, in addition to identifying the intent, further obtains a multi-dimensional profile including a personal characteristic profile, a job profile, a corporate environment profile, and a policy profile, and selects skill modules based on both the intent and the multi-dimensional profile, the skill modules invoked can more accurately adapt to the personalized needs of special populations, thus improving the level of assisted employment services for special populations.

[0079] <Device Embodiment> Figure 2 A schematic block diagram of an assisted employment device for a special population according to an embodiment of the present disclosure is shown. Figure 2 As shown, the device 100 includes: The data acquisition module 110 is used to acquire the user's employment questions and identify the intent corresponding to the employment questions; The profile association module 120 is used to automatically obtain a multi-dimensional profile required to answer the employment question based on the intent. The multi-dimensional profile includes at least one of the following: a personal characteristic profile reflecting the user's specific needs, a job profile reflecting the job requirements, a corporate environment profile reflecting the corporate environment, and a policy profile. The filtering module 130 is used to filter out skill modules that are suitable for answering the employment question from a pre-built skill pool based on the intent and the multi-dimensional profile. The execution module 140 generates answers to the employment questions based on the selected skill modules and pushes them to the user.

[0080] <Equipment Example> Figure 3 A schematic block diagram of an assistive employment device for special populations according to an embodiment of the present disclosure is shown. Figure 3 As shown, the assisted employment device 200 for special populations includes a processor 210 and a memory 220 for storing executable instructions of the processor 210. The processor 210 is configured to implement any of the aforementioned assisted employment service methods for special populations when executing the executable instructions.

[0081] It should be noted here that the number of processors 210 can be one or more. Furthermore, the assistive employment device 200 for special populations in this embodiment may also include an input device 230 and an output device 240. The processors 210, memory 220, input device 230, and output device 240 can be connected via a bus or other means, without specific limitations here.

[0082] The memory 220, as a computer-readable storage medium, can be used to store software programs, computer-executable programs, and various modules, such as the program or module corresponding to the assisted employment service method for special populations in this embodiment of the present disclosure. The processor 210 executes various functional applications and data processing of the assisted employment device 200 for special populations by running the software programs or modules stored in the memory 220.

[0083] Input device 230 can be used to receive input digital numbers or signals. These signals may include key signals related to user settings and function control of the device / terminal / server. Output device 240 may include a display device such as a screen.

[0084] <Storage Medium Examples> According to a fourth aspect of this disclosure, a non-volatile computer-readable storage medium is also provided, on which computer program instructions are stored, which, when executed by processor 210, implement any of the aforementioned methods for assisted employment services for special populations.

[0085] The various embodiments of this disclosure have been described above. These descriptions are exemplary and not exhaustive, and are not limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical applications, or technical improvements to the technology in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A method for providing assisted employment services to a special population, characterized in that, include: The system obtains information about a user's employment status and identifies the intent corresponding to that employment status, wherein the user is a special group of people with certain physical disabilities. Based on the stated intent, a multi-dimensional profile required to answer the employment question is automatically obtained, wherein the multi-dimensional profile includes at least one of the following: a personal characteristic profile reflecting the user's specific needs, a job profile reflecting job requirements, a corporate environment profile reflecting the corporate environment, and a policy profile. Based on the intent and the multi-dimensional profile, skill modules suitable for answering the employment question are selected from a pre-built skill pool, wherein the skill modules are computing modules that perform specific business tasks; Based on the selected skill modules, a response to the employment question is generated and pushed to the user.

2. The method according to claim 1, characterized in that, When selecting skill modules suitable for answering the employment question from a pre-built skill pool based on the stated intent and the multi-dimensional profile, the process includes: Based on the intent, multiple first candidate skill modules that fit the intent are selected from the skill pool; Calculate the matching degree between each first candidate skill module and the multidimensional profile, and based on the matching degree between each first candidate skill module and the multidimensional profile, select a set number of second candidate skill modules with the highest matching degree from each first candidate skill module; The final skill module that is suitable for answering the employment question is determined from the second candidate skill modules.

3. The method according to claim 2, characterized in that, Calculating the matching degree between the first candidate skill module and the multidimensional profile includes: Obtain the module multi-dimensional profile corresponding to the first candidate skill module and the historical evolution weight of the module multi-dimensional profile under the multi-dimensional profile; Calculate the initial matching degree between the module's multidimensional profile and the multidimensional profile; The initial matching degree is revised using the historical evolution weight to obtain the final matching degree between the module multidimensional profile and the multidimensional profile; The final matching degree between the module's multidimensional profile and the multidimensional profile is taken as the matching degree between the first candidate skill module and the multidimensional profile.

4. The method according to claim 3, characterized in that, The historical evolution weights are dynamically adjusted based on user feedback data.

5. The method according to claim 4, characterized in that, When adjusting the historical evolution weights based on user feedback data, the following is included: Determine the sentiment tendency of the feedback data, wherein the sentiment tendency of the feedback data includes at least one of negative feedback, neutral feedback, and positive feedback; The latest historical evolution weights are adjusted based on the sentiment trends of the feedback data.

6. The method according to claim 2, characterized in that, When determining the final skill module suitable for answering the employment question from the second candidate skill modules, the process includes: Based on the personal feature profile in the multi-dimensional profile, the user's physical functional constraints are determined; Based on the user's physical functional constraints, the final skill module suitable for answering the employment question is determined from the second candidate skill modules.

7. The method according to claim 2, characterized in that, When generating answers to the employment questions based on the selected skill modules and pushing them to the user, the process includes: Determine the execution order of each of the selected skill modules; The skill modules are invoked sequentially to generate answers to the employment questions. The response will be pushed and displayed to the user.

8. A device for assistive employment services for a special population, characterized in that, include: The data acquisition module is used to acquire users' employment questions and identify the intentions corresponding to the employment questions, wherein the users are special groups with certain physical disabilities; The profile association module is used to automatically obtain a multi-dimensional profile needed to answer the employment question based on the intent. The multi-dimensional profile includes at least one of the following: a personal characteristic profile reflecting the user's specific needs, a job profile reflecting job requirements, a corporate environment profile reflecting the corporate environment, and a policy profile. A filtering module is used to filter out skill modules that are suitable for answering the employment question from a pre-built skill pool based on the intent and the multi-dimensional profile, wherein the skill module is a computing module that performs a specific business task; The execution module generates answers to the employment questions based on the selected skill modules and pushes them to the user.

9. An assistive employment service device for a special population, characterized in that, include: processor; Memory used to store processor-executable instructions; The processor is configured to implement the method of any one of claims 1 to 7 when executing the executable instructions.

10. A non-volatile computer-readable storage medium storing computer program instructions thereon, characterized in that, When the computer program instructions are executed by the processor, they implement the method described in any one of claims 1 to 7.

Images