Mutual old-age support dispatching management method and system based on resident co-governance and medium

Abstract

The application provides a mutual aid old-age care dispatching management method and system based on resident co-governance, and a medium. The method comprises the following steps: based on the information of each service subject in a service subject information library, each service subject in the service subject information library is divided into various service circles, and each service circle is a multi-layer nested circle with a service object as the center, which comprises a first service circle, a second service circle, a third service circle and a fourth service circle from inside to outside; when a service demand is received, at least based on the service demand, a recommended service circle is determined; a service request notification is pushed to each service subject in the recommended service circle, and the service request notification is generated based on the service demand; when confirmation information of any service subject in the recommended service circle is received, the service subject that sends the confirmation information is determined as a target service subject; and in response to a service confirmation operation, the service demand is determined to be completed. The scheme can improve the automation level and information processing efficiency of mutual aid old-age care management.

Description

Technical Field

[0001] This invention relates to the field of smart elderly care technology, and more specifically to a method, system, and medium for scheduling and managing mutual-aid elderly care based on resident co-governance. Background Technology

[0002] As my country's population continues to age rapidly, the demand for multi-level and diversified elderly care services among urban and rural residents is growing. The mutual-aid elderly care model, rooted in traditional kinship, geographical ties, and neighborhood support, is gradually becoming an indispensable component and key development direction of the elderly care service system due to its advantages: low-cost operation—relying on voluntary participation and volunteer service mechanisms to significantly reduce service supply costs; high trust—service providers are mostly from familiar social circles, resulting in significantly higher acceptance and satisfaction among the elderly; and high synergy with the grassroots governance system—seamlessly integrating into existing organizational structures such as village committees, community residents' committees, and senior citizens' associations.

[0003] Currently, this village / community-based mutual aid elderly care model has the following shortcomings: First, the existing model treats all entities as an undifferentiated, flat resource pool. Any service request (such as visits, companionship, shopping, or other low-complexity, light services) requires global searching and unified scheduling. This approach incurs high computational costs and lengthy response paths, contradicting the efficient self-organizing characteristics of a close-knit community. Second, after service requests are entered into the input end, the existing model relies entirely on community staff to manually read, understand, classify, and input the requests into forms. This method is not only inefficient but also suffers from subjective differences in understanding and classifying the same request among different personnel, affecting the consistency and accuracy of subsequent task assignments. Moreover, task assignment decisions depend entirely on the experience of community staff, lacking quantitative basis. These problems seriously hinder the large-scale development and promotion of mutual aid elderly care management.

[0004] In view of this, the present invention is hereby proposed. Summary of the Invention

[0005] The present invention addresses the aforementioned problems. According to one aspect of the invention, a method for scheduling and managing mutual-aid elderly care based on resident co-governance is provided, comprising: dividing each service entity in a service entity information database into various service circles based on information about each service entity in the database; each service circle being a multi-layered nested circle centered on a service recipient, consisting of a first service circle, a second service circle, a third service circle, and a fourth service circle from the inside out; the service entity information database being used to store information about each service entity within the service jurisdiction where the service recipient is located; upon receiving a service request from a service recipient, determining a recommended service circle within each service circle based at least on the service request; pushing a service request notification to each service entity within the recommended service circle, the service request notification being generated based on the service request; upon receiving confirmation information from any service entity within the recommended service circle, identifying the service entity that sent the confirmation information as the target service entity; and, in response to the service confirmation operation, determining that the service corresponding to the service request has been completed.

[0006] For example, based on the information of each service entity in the service entity information database, each service entity in the service entity information database is divided into various service circles, including: constructing a social relationship association graph based on the information of each service entity in the service entity information database, wherein the nodes in the social relationship association graph include each service entity in the service entity information database; inputting the social relationship graph into a pre-trained service circle mapping model to obtain the probability of each service entity belonging to different service circles; wherein, for any service circle, the service entities within that service circle are service entities whose probability of belonging to that service circle is greater than a first probability threshold; preferably, the service circle mapping model consists of a graph attention network and a classification output layer, and inputting the social relationship graph into the pre-trained service circle mapping model includes: processing the social relationship association graph using the graph attention network to output a high-dimensional embedding vector corresponding to each service entity in the service entity information database, wherein the high-dimensional embedding vector is used to represent the role feature of the corresponding service entity in the social relationship association graph; inputting the high-dimensional embedding vector corresponding to each service entity into the classification output layer to obtain the probability of each service entity belonging to different service circles.

[0007] For example, at least based on service needs, a recommended service circle is determined in each service circle, including: determining the service type label and urgency score of the service needs based on the service needs; determining the recommended service circle based at least on the service type label and urgency score; preferably, determining the service type label and urgency score of the service needs includes: inputting the service needs into a pre-trained language model to obtain a semantic representation of the service needs; inputting the semantic representation into an intent recognition model to determine the service type label of the service needs; preferably, the intent recognition model is also used to output the intent recognition confidence score corresponding to the service type label; the method further includes: performing named entity recognition on the semantic representation to obtain the information corresponding to each slot in the structured slot template and the entity recognition confidence score, wherein the structured slot template includes at least slots for filling in the service object name, service address, expected service time period and urgency descriptive words; in determining Before recommending the service circle, the method also includes: generating a system resolution confidence score based on intent recognition confidence, entity recognition confidence, and the completeness of key fields of service requirements; the step of determining the recommended service circle is executed when the system resolution confidence score is higher than a preset confidence threshold and all key slots in the structured slot template are not empty, and the key slots include slots for filling in the service recipient's name, service address, expected service time period, and urgency description words; preferably, determining the service type label and urgency score of the service requirement includes: detecting high-risk words in the service requirement through named entity recognition and generating risk word scores based on high-risk words; identifying urgency expressions in the service requirement through semantic analysis and generating urgency scores; generating historical health risk scores for the service recipient based on the service recipient's historical information; and calculating the weighted sum of the risk word scores, urgency scores, and historical health risk scores to obtain the urgency score.

[0008] For example, determining a recommended service circle based at least on service type labels and urgency scores includes: determining an initial service circle in a preset two-dimensional mapping table based on service type labels and urgency scores, wherein the preset two-dimensional mapping table is used to represent the correspondence between service type labels, different urgency score intervals and each service circle; determining whether the service object meets correction conditions based on the service object's historical service information and health record, the correction conditions including a first correction condition and a second correction condition, the first correction condition being that the historical service information shows that the service object's most recent preset number of similar needs were completed in service circles outside the initial service circle more than a preset number of times, the second correction condition being that the service object's health record has a high-risk chronic disease marker related to the service type, and the difference between the urgency score and any urgency score interval threshold is less than a preset difference threshold; when either correction condition is met, correcting the initial service circle according to the satisfied correction condition to determine the recommended service circle; preferably, correcting the initial service circle according to the satisfied correction conditions includes: when only the first correction condition is satisfied, using a first correction operation to correct the initial service circle, and determining the corrected initial service circle as the recommended service circle; ... determining the recommended service circle as the recommended service circle, and determining the recommended service circle as the recommended service circle, and determining the recommended service circle as the recommended service circle, and determining the recommended service circle as the recommended service circle, and determining the recommended service circle as the recommended service circle, and determining the recommended service circle as the recommended service circle, and determining the recommended service circle as the recommended service circle, and determining the recommended service circle as the recommended service circle, and determining the recommended service circle as the recommended service circle, and determining the recommended service When the second correction condition is met, the initial service circle is corrected using the second correction operation, and the corrected initial service circle is determined as the recommended service circle. When both the first and second correction conditions are met, the initial service circle is corrected using the first and second correction operations respectively to obtain the first correction result and the second correction result. The service circle that is outermost in the first and second correction results is selected as the recommended service circle. The first correction operation includes: adjusting the current initial service circle to the next service circle. The second correction operation includes: compensating the urgency score according to a preset risk compensation value, and re-determining the initial service circle in the preset two-dimensional mapping table based on the compensated urgency score. Preferably, the method further includes: when the initial service circle is the third service circle, if the load of each service entity in the third service circle is saturated, a horizontal coordination request is generated and sent to the administrator node. When the confirmation information of the administrator node is received, the load of the service entities in the third service circle of the adjacent service area of ​​the current service area is obtained. If the load of the service entities in the third service circle of the adjacent service area is not saturated, the service request notification is pushed to each service entity in the third service circle of the adjacent service area. Preferably, the correction of the initial service circle according to the satisfied correction conditions further includes: When the initial service circle is the third service circle and the revised service circle is the fourth service circle, an upgrade service circle request is generated and sent to the administrator node; upon receiving confirmation from the administrator node, the fourth service circle is determined to be the recommended service circle.

[0009] For example, pushing service request notifications to each service entity within the recommended service circle includes: determining the suitability score for each service entity within the recommended service circle; sorting the service entities within the recommended service circle according to their suitability scores from highest to lowest to obtain a push sequence; pushing service request notifications to each service entity in batches according to the push sequence with a preset push step size; stopping the push if confirmation information from a service entity in the current push batch is received within a preset waiting time; otherwise, pushing the service request notification to the next batch of service entities; preferably, the suitability score is determined by at least the sum of a spatial distance score, a historical service quality score, an order load score, and a skill matching score; wherein, the spatial distance score is determined based on the spatial distance between the service entity and the service recipient; the historical service quality score is determined based on the service entity's historical service information, which includes historical service completion rate, average response time, and service feedback. The scoring system includes: an order-taking load score, determined based on the number of service tasks the service provider has accepted but not yet completed, with a negative correlation between the order-taking load score and the number of service tasks; and a skill matching score, determined based on the degree of matching between the service provider's skill tags and the service type tags of the service requirements. Preferably, pushing service request notifications to each service provider within the recommended service circle further includes: marking service providers within the recommended service circle with an attribution probability greater than a second probability threshold as high-confidence entities, and marking other service providers as entities awaiting verification. For any high-confidence entity within the recommended service circle, the service provider's suitability score is the sum of its spatial distance score, historical service quality score, order-taking load score, and skill matching score. For any entity awaiting verification within the recommended service circle, the service provider's suitability score is the product of the sum of its spatial distance score, historical service quality score, order-taking load score, and skill matching score, and a penalty coefficient less than 1.

[0010] For example, the method further includes: when none of the service entities in the recommended service circle have accepted the order, pushing a service request notification to the next service circle in the recommended service circle; preferably, the method further includes: when the total response time of the recommended service circle reaches the service circle timeout threshold, determining that none of the service entities in the recommended service circle have accepted the order, the service circle timeout threshold being a preset fixed value or a dynamic value; preferably, the service circle timeout threshold is a dynamic value, and the service circle timeout threshold is determined by: calculating the service circle timeout threshold based on the urgency score of the service request, the time period in which the service request was initiated, and the historical average response time of the recommended service circle; preferably, calculating the service circle timeout threshold includes: based on the urgency score of the service request, the time period in which the service request was initiated, and the historical average response time of the recommended service circle .... The urgency level is scored to determine the urgency factor; a time factor is determined based on the time period in which the service request is initiated; a historical response factor is determined based on the historical average response time; a service circle timeout threshold is obtained by calculating the product of a preset base timeout, the urgency factor, the time factor, and the historical response factor; preferably, calculating the service circle timeout threshold further includes: determining whether the service circle timeout threshold is within a preset timeout interval; if the service circle timeout threshold is not within the preset timeout interval, adjusting the service circle timeout threshold to the nearest value in the preset timeout interval; preferably, the preset waiting time is the ratio of the service circle timeout threshold to the total number of service batches in the recommended service circle.

[0011] For example, the method further includes: locking the service task corresponding to the service request after the preset lock time for receiving the confirmation information for the first time; when there are multiple confirmation messages within the preset lock time, sorting the confirmation messages according to the receiving time of each confirmation message; sequentially traversing the suitability scores of the service entities corresponding to each confirmation message according to the sorting results; and determining the current service entity as the target service entity when the suitability score of the currently traversed service entity is higher than the minimum available threshold.

[0012] For example, the method further includes: upon receiving a risk escalation request from the target service subject, adjusting the urgency score based on the risk escalation request, and re-executing the step of determining a recommended service circle based at least on the service type label and the urgency score based on the adjusted urgency score; and / or, the method further includes: upon receiving correction information from the administrator, modifying the service requirements according to the correction information, and re-executing the step of determining a recommended service circle in each service circle based at least on the service requirements based on the modified service requirements.

[0013] According to another aspect of the present invention, a mutual assistance elderly care scheduling and management system based on resident co-governance is provided, which is used to implement the above-described method.

[0014] According to another aspect of the present invention, a computer-readable storage medium is provided, which stores a computer program / instructions that, when executed by a processor, implement the method described above.

[0015] Compared to existing technologies, the above solution has at least the following technical advantages: 1. When matching service needs with service providers, this solution only needs to operate within the service circle that matches the service needs, eliminating the need for a global search. This narrows the matching scope, reduces computational overhead and organizational coordination costs, and improves matching efficiency. 2. When determining the recommended service circle, this solution uses service needs as the deciding factor, not the order of service circles, and can directly select outer service circles for push, enabling flexible selection of service circles. 3. After determining the recommended service circle, this solution can automatically dispatch orders to service providers within the service circle, replacing manual subjective order dispatch and improving decision-making efficiency and accuracy. 4. This solution can improve the automation level of mutual aid elderly care management, reduce the frequency of manual intervention and human judgment errors, improve order dispatch and response efficiency, and enable the urban and rural community mutual aid elderly care system to cover a larger scale of service needs with lower operating costs, thereby improving service response efficiency, fairness, and sustainable operation capabilities.

[0016] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, and in order to make the above and other objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0017] The above and other objects, features, and advantages of the present invention will become more apparent from the more detailed description of the embodiments of the invention in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same parts or steps.

[0018] Figure 1 A schematic block diagram of a method according to an embodiment of the present invention is shown. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the present invention more apparent, exemplary embodiments according to the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are merely a part of the embodiments of the present invention, and not all of the embodiments of the present invention. It should be understood that the present invention is not limited to the exemplary embodiments described herein. Based on the embodiments of the present invention described herein, all other embodiments obtained by those skilled in the art without inventive effort should fall within the protection scope of the present invention.

[0020] As mentioned above, existing mutual-aid elderly care management methods suffer from problems such as flat organizational logic, high costs, low efficiency, reliance on manual interpretation for demand acceptance, poor processing consistency, reliance on human experience for dispatch decisions, and lack of quantitative basis. Therefore, this invention provides a method, system, and medium for mutual-aid elderly care scheduling and management based on resident co-governance. This method, through service circle division and automatic dispatch, can solve the problems of flat organizational logic and reliance on human experience for demand analysis and dispatch in existing elderly care management methods, thus contributing to the large-scale development and promotion of mutual-aid elderly care management.

[0021] According to one aspect of the present invention, a mutual-aid elderly care scheduling and management system based on resident co-governance is provided, which is used to implement the methods described below. The system may include a central platform and multiple edge nodes. The central platform and multiple edge nodes form a collaborative distributed architecture. Each edge node is deployed within a service area (a village or community). Edge nodes are used to independently implement the methods described below. The central platform is responsible for cross-node state synchronization, cross-node upgrade arbitration, and global data analysis. This architecture allows the system's processing capacity to scale linearly with the number of edge nodes, avoiding the performance bottleneck of the central node in a centralized architecture. Simultaneously, each edge node can flexibly configure its service area and scheduling parameters according to its local organizational structure and resource endowment, adapting to the differentiated governance habits of different urban and rural communities.

[0022] The system can also include intermediate nodes. For example, the central platform can be set up at the county level, and intermediate nodes can correspond to townships / streets, etc. This allows for a multi-layered architecture of county-township-community / village. The intermediate nodes can be used for approving requests such as cross-node horizontal coordination.

[0023] According to another aspect of the present invention, a method for scheduling and managing mutual-aid elderly care based on resident co-governance is provided. Figure 1 A schematic block diagram of a method according to an embodiment of the present invention is shown. Figure 1 As shown, the method may include the following steps: S110, S120, S130, S140, and S150.

[0024] In step S110, based on the information of each service entity in the service entity information database, each service entity in the database is divided into various service circles. Each service circle is a multi-layered nested circle centered on the service object, consisting of a first service circle, a second service circle, a third service circle, and a fourth service circle from the inside out. The service entity information database is used to store information about each service entity within the service jurisdiction where the service object is located. The service object mentioned in this document can be any elderly service entity in the service entity information database.

[0025] The service provider information database is built based on the service area (which can be a village or community) where the service recipient resides. In this example solution, a service provider information database can be pre-established on the central platform, recording information on each service provider in the village / community (including identity identifier, contact information, spatial coordinates (such as the specific residential building or village group location), historical service participation records, organizational affiliation, and relationship with the service recipient). Based on this information, the closeness of the relationship between each service provider and the service recipient can be calculated. For example, this information can be input into a large language model, and the large language model can be used to output the service circle division results. Alternatively, a knowledge graph combined with a graph attention network algorithm can be used to obtain the service circle classification results. Of course, service circles for different service providers can also be manually labeled and divided, which will not be elaborated upon here.

[0026] In some embodiments, the service area classification results are as follows: The first service circle (family and relatives): This includes the elderly person's family members, relatives, and authorized contacts. This circle is based on blood ties and strong trust, and is prevalent in both urban and rural areas.

[0027] The second service circle (neighborhood and volunteer mutual aid circle): In rural areas, this includes neighbors, volunteers, mutual aid service personnel, and ordinary members of the senior citizens' association within the same village / group; in urban areas, it includes neighbors in the same building / community, community volunteers, and building managers. This circle is the core carrier of mutual aid and co-governance, characterized by spatial proximity and familiarity.

[0028] The third service circle (community / village-level organizations and service backbone): In rural areas, this includes village committee staff, grid workers, senior citizens' association leaders, women's directors, village clinic doctors, and resident social workers; in urban areas, it includes neighborhood committee staff, grid workers, community service station staff, community health service center staff, and community social workers. This circle possesses certain organizational and coordination authority and preliminary professional handling capabilities.

[0029] The fourth service circle (external professional service circle) includes elderly care service centers, health centers / community health service centers, professional care institutions, and medical and elderly care resources at the district / county level or a wider range that can be linked.

[0030] In practical scenarios, based on the aforementioned four horizontal service layers, a seven-layer vertical management hierarchy can be simultaneously constructed, with the two dimensions intersecting to form a complete layered governance system. The correspondence between the seven vertical layers and the four horizontal layers is as follows: The first-tier village / community senior citizen associations and the second-tier village / community neighborhood committees / village committees jointly correspond to the grassroots organizations and coordinate resources in circles one through three, forming a nested relationship: the neighborhood committees / village committees are the legal governance entities, coordinating the operation of senior citizen associations; the senior citizen associations are the direct implementing organizations, connecting seniors and volunteers to carry out specific services. Both tiers share the same village / community-level edge node in the technical system. This node has dual-level operational permissions: the head of the neighborhood committee / village committee holds the final review and data reporting authority, while the head of the senior citizen association holds the task assignment authority within their association's scope.

[0031] The third-tier street / township senior citizen associations correspond to the cross-village overall coordination of the core resources of the three circles, and are the township-level intermediate nodes in the technical system, responsible for the horizontal allocation of resources and data aggregation within their jurisdiction.

[0032] The fourth layer, district and county-level senior citizen associations, corresponds to the access and management of professional resources in the fourth layer, and is the county-level central platform of the technical system, serving as the core deployment node of the technical system.

[0033] The fifth tier, regional level, the sixth tier, provincial level, and the seventh tier, national level, are responsible for cross-county resource allocation, provincial standard setting, and national standard setting, respectively. They achieve data aggregation and policy parameter distribution through the standardized data interface reserved in the county-level central platform, without directly intervening in the daily order dispatching operations below the county level.

[0034] The above seven-layer management structure can be adapted to local conditions by omitting some of its layers.

[0035] In step S120, upon receiving a service request from a service object, a recommended service circle is determined among the various service circles based at least on the service request.

[0036] Service requests can be submitted by the service recipient, their family members, neighborhood volunteers, community / village staff, etc., through the user terminal. Submission methods can support voice input or text input, and the content is usually a free description in natural language (e.g., "Buy some groceries for my mom and deliver them," or "Grandpa Zhang says he feels tightness in his chest and needs someone to check on him"). When the service request is voice-based, the method can also include converting the voice request into text. This facilitates the parsing of the service request.

[0037] After parsing the service request, a recommended service circle for executing the service request can be determined based on the specific details of the service request using a pre-defined correspondence database (which pre-stores the executable items in different service circles). Alternatively, the recommended service circle can be determined using semantic encoding combined with neural network models such as intent classification.

[0038] In step S130, the service request notification is pushed to each service entity within the recommended service circle. The service request notification is generated based on the service requirements.

[0039] Once the recommended service circle is determined, service request notifications can be pushed to each service entity within the circle. For example, service request notifications can be pushed to each service entity simultaneously. Alternatively, service entities can be sorted (e.g., based on their historical mutual aid ratings), and service request notifications can be pushed in batches according to the order, stopping further pushes upon receiving confirmation.

[0040] Service request notifications can utilize the structured slot information extracted below, or they can directly present the text corresponding to the service request as the service request notification. In some embodiments, the service request notification may include basic information about the service recipient, a brief description of the service, the expected time window, and the service address (after privacy anonymization processing), etc.

[0041] In step S140, when a confirmation message is received from any service entity in the recommended service circle, the service entity that sent the confirmation message is identified as the target service entity.

[0042] Upon receiving confirmation from the service provider, it indicates that a service provider has accepted the service task corresponding to the service request. At this point, the service flow of the task can be locked, the service provider's identifier and confirmation time recorded, and the order acceptance information can be fed back to the user terminal that initiated the service request. During the service provided by the target service provider, service execution status information (such as arrived, in progress, completed, etc.) can also be received to update the service progress of the service request.

[0043] In step S150, in response to the service confirmation operation, it is determined that the service corresponding to the service request has been completed.

[0044] After the service is completed, the confirmation operation can be completed in one or more of the following ways. The confirmation result is sent back to the original community / village-level edge node that accepted the service, and the node completes the final closed-loop archiving of the service record.

[0045] Seniors can confirm services through simple interactive operations on their user terminals (such as clicking the confirmation button on the screen or using voice confirmation). For scenarios using voice confirmation, after the terminal device captures the voice segment, it can call the voice recognition module to transcribe the confirmation voice into a text record, which is then archived along with the original voice file.

[0046] Family members or authorized contacts of the elderly can also complete service confirmation on their behalf through their linked user terminals. When confirming on behalf of the elderly, the identifier of the person confirming on their behalf and their relationship tag with the service recipient (such as children, spouse, authorized neighbor) are recorded.

[0047] For situations where the service recipient is unable to confirm the service completion themselves, or where family members or authorized persons are unable to confirm on their behalf in a timely manner, the person in charge of the community / village-level edge node (such as the director of the neighborhood committee or the president of the senior citizens' association) can confirm the service completion status through telephone follow-ups, home visits, or other methods, and perform manual confirmation on the terminal. During manual confirmation, photos, brief text records, etc., can be uploaded as verification vouchers. The hash value of the voucher file is stored in the log after hash calculation.

[0048] In some embodiments, the node manager can intervene manually at various stages from receiving a service request to its completion. For example, the recommended service circle can be manually replaced after it has been determined.

[0049] The aforementioned technical solution allocates different service providers to different service circles through automated service circle division. On the one hand, when matching service needs with service providers, the process only needs to be performed within the service circle that matches the service need, eliminating the need for a global search. This narrows the matching scope, reduces computational overhead and organizational coordination costs, and improves matching efficiency. On the other hand, when determining the recommended service circle, the service need is the deciding factor, not the order of the service circles. The outermost service circles can be directly selected for push, enabling flexible selection of service circles. Furthermore, after determining the recommended service circle, the solution can automatically dispatch orders to service providers within that circle, replacing manual subjective order dispatch and improving decision-making efficiency and accuracy. In summary, this solution can improve the automation level of mutual aid elderly care management, reduce the frequency of manual intervention and human judgment errors, improve order dispatch and response efficiency, and enable the urban and rural community mutual aid elderly care system to cover a larger scale of service needs with lower operating costs, thereby improving service response efficiency, fairness, and sustainable operation capabilities.

[0050] For example, based on the information of each service entity in the service entity information database, each service entity in the service entity information database is divided into various service circles, including: constructing a social relationship association graph based on the information of each service entity in the service entity information database, wherein the nodes in the social relationship association graph include each service entity in the service entity information database; inputting the social relationship graph into a pre-trained service circle mapping model to obtain the probability of each service entity belonging to different service circles; wherein, for any service circle, the service entities within that service circle are service entities whose probability of belonging to that service circle is greater than a first probability threshold.

[0051] It is understandable that a social relationship map corresponding to a service area includes all service entities, including service recipients within the area, which will not be elaborated upon here.

[0052] In this example, we consider using a knowledge graph approach to determine the service circles of each service entity. Specifically, we can construct a dynamic weighted association graph with service entities as nodes and multidimensional relationships between entities as edges.

[0053] The weights of the edges are calculated by the processor based on the following dimensions: blood kinship weights based on registration information, spatial proximity weights based on geographic information system distance measurement, mutual assistance frequency and evaluation weights based on historical service logs, and common group affiliation weights based on organizational structure (such as senior citizen associations, volunteer team member relationships).

[0054] ; In the formula The overall weight of the edges; These represent weighted scores for blood relations, spatial distance, frequency of mutual assistance and evaluation, and group affiliation, respectively. This represents the conversion factor of the above four types of weights to the overall weight.

[0055] The quantification method is as follows: If the two individuals are related in the system's registration information, assign a value of 1.0; otherwise, assign a value of 0. Calculate the actual distance based on the geographical coordinates of the two people's residences. D The distance decay function is used for calculation: d (i,j) =1 / (1+ × D ),in This is the distance attenuation coefficient. The closer the distance, the higher the sub-weight, approaching 1; the farther the distance, the closer it approaches 0. Nodes in the statistical system's log i With nodes j Cumulative number of mutual aid acts in history N After normalization, the following was obtained h (i,j) = N / ( N + β ), of which β A smoothing coefficient is used to prevent the denominator from being zero. : Count the number of mutual aid organizations M that the two people belong to (if they belong to the senior citizens' association and the volunteer team, then M=2). After obtaining the social relationship graph, mechanisms such as graph attention or other neural network models can be used to analyze it to determine the probability of each service entity belonging to different service circles. For any given service circle, all service entities whose probability of belonging to that service circle is greater than a first probability threshold can be identified as service entities within that service circle.

[0056] The aforementioned technical solution relies on a comprehensive service provider information database, constructs a social relationship graph based on service provider information, clearly identifies various related nodes and connection logics, and then uses a pre-trained service circle mapping model to accurately calculate the probability of each service provider belonging to different service circles. By setting a first probability threshold, service circle screening and division are completed. This efficiently and scientifically categorizes all service providers into various service circles, significantly improving the intelligence and accuracy of service circle division, increasing its efficiency, and avoiding errors introduced by manual division, thus providing an accurate basis for subsequent steps. Furthermore, this service circle division, based on the social relationship graph, reflects the closeness or distance between the service providers and the recipients, aligning with the highly efficient self-organizing characteristics of a familiar, close-knit community.

[0057] For example, the service circle mapping model consists of a graph attention network and a classification output layer. The social relationship graph is input into the pre-trained service circle mapping model, including: processing the social relationship graph using the graph attention network to output a high-dimensional embedding vector corresponding to each service subject in the service subject information database, wherein the high-dimensional embedding vector is used to represent the role feature of the corresponding service subject in the social relationship graph; and inputting the high-dimensional embedding vector corresponding to each service subject into the classification output layer to obtain the probability of each service subject belonging to different service circles.

[0058] Research has revealed that a single service provider may hold multiple roles (e.g., being both a neighbor and a key village doctor). In this example, to address the issue that traditional rules cannot accurately reflect the complex relationships of "one person, multiple roles" within a community, a Graph Attention Network (GAT) is considered as the specific implementation architecture for the graph neural network. GAT introduces an attention mechanism on top of the standard Graph Convolutional Network (GCN), enabling it to adaptively learn the differences in importance among different neighboring nodes. This makes it particularly suitable for handling the complex roles and uneven strength of relationships within community mutual aid networks.

[0059] In practical implementation, a graph attention network can be used to perform the following operation: For any target node (i.e., a service entity) in the graph, the feature vector pairs of the target node and its k neighboring nodes are calculated using an attention function. This adaptively outputs the importance weights of each neighboring node to the target node, rather than simply averaging all neighbor information. The calculation method of the self-attention mechanism is as follows:

[0060]

[0061]

[0062]

[0063]

[0064]

[0065] in and These are the attributes of the neighboring node and the target node, respectively. , and These are the mapping functions for the two.

[0066] Through the stacking of multi-layer neural networks and nonlinear transformations, the system generates a high-dimensional embedding vector for each node in the graph. This vector implicitly encodes the topological position, connectivity, and role characteristics of the entity in the overall community mutual aid network.

[0067] Based on this high-dimensional embedding vector, a classification output layer (such as a Softmax layer) is connected to calculate the probability distribution of the subject's affiliation to each service circle. The service circle to which it belongs is determined based on the probability value, and the probability vector is retained for multi-circle response strategy invocation. The final generated dynamic label structure example is: [Node ID]-[Circle Number]-[Service Type Code]-[Confidence Score].

[0068] In this example, the initial training of the model can use existing manually labeled geography data as a supervision signal; in areas lacking labeled data, a semi-supervised strategy can be used to guide the model to convergence with a small number of samples labeled with known rules, which will not be elaborated here.

[0069] In some embodiments, lightweight incremental updates can be used when information in the service subject information database changes. For example, when events such as registering a new member, updating a residential address, generating a new mutual aid completion record, or changing an affiliated organization occur, it is not necessary to retrain the entire graph. The algorithm framework automatically identifies the associated nodes of the changed node within its K-order neighborhood (K is usually 2 or 3), performs forward propagation operations only on nodes within that local subgraph, updates the embedding vectors and corresponding ring labels of the affected nodes, and asynchronously synchronizes the updated label data to the corresponding edge nodes.

[0070] When a change event carries new manually labeled signals, it can trigger incremental fine-tuning of the K-order local subgraph, updating only the local parameters; if it is only an attribute change (such as address update), it only performs forward propagation to refresh the embedding vector, without updating the model parameters.

[0071] In the above scheme, the service circle mapping model relies on the collaborative operation of graph attention network and classification output layer to efficiently complete the data analysis and feature mining of social relationship graph. It can accurately extract the unique role features of each service subject in the relationship graph and transform them into standardized high-dimensional embedding vectors, clearly quantifying their relationship attributes and positioning characteristics. Furthermore, it can accurately calculate the probability of each service subject belonging to different service circles through the subsequent classification output layer, greatly improving the accuracy and efficiency of service circle division of service subjects.

[0072] For example, at least based on service needs, a recommended service circle is determined in each service circle, including: determining service type tags and urgency scores for service needs based on service needs; and determining the recommended service circle based at least on service type tags and urgency scores.

[0073] In this example solution, we first consider determining the service type and urgency of the service request. The service type can be determined using a large language model or an intent recognition model. Similarly, the service request can be input into a large language model to determine its urgency score. Alternatively, we can identify and score the urgency expressions within the service request to obtain an urgency score.

[0074] After obtaining service type labels and urgency scores, recommended service zones can be determined based on these. For example, a classification model (which can be built based on a convolutional neural network) can be pre-trained, and the service type labels and urgency scores can be input into the classification model to obtain recommended service zones. Alternatively, a correspondence between different service type labels and urgency scores and various service zones can be established in advance, and the recommended service zones can be determined based on this correspondence.

[0075] The above solution quickly determines the service area based on the urgency of the demand and the type of service. This ensures that the selected service area matches the service demand, providing a more accurate recommendation range for subsequent service request notifications, reducing errors, and improving processing efficiency and consistency.

[0076] For example, determining the service type label and urgency score of a service request includes: inputting the service request into a pre-trained language model to obtain a semantic representation of the service request; and inputting the semantic representation into an intent recognition model to determine the service type label of the service request.

[0077] In some embodiments, the request text corresponding to the service request can first be input into a domain-fine-tuned pre-trained language model (such as BERT and its lightweight variants) for semantic encoding. This model can capture the contextual semantic relationships in the text through a multi-head self-attention mechanism, generating a semantic representation containing deep semantic information, including sentence vector representation and word vector representation. Then, the sentence vector representation can be input as the overall semantic feature of the text into an intent recognition model, which outputs the probability distribution of each preset service type and determines the service type label for this request based on the highest probability result. This process can be represented as follows:

[0078]

[0079] in, h Sentence vector representation, p For the probability distribution of each service category, For the finalized service type label, W This is a weight matrix used to map sentence vectors to the service category space. b For bias vectors, The probability of the k-th service category.

[0080] The above technical solution obtains accurate semantic representations by inputting service requirements into a pre-trained language model, and then uses an intent recognition model to determine service type labels and simultaneously determine urgency scores. This enables efficient and accurate classification of service requirements, significantly improves the intelligence level and efficiency of service requirement identification, reduces errors in manual judgment, and provides a more accurate basis for subsequent service recommendation and market segmentation.

[0081] For example, determining the service type label and urgency score of a service request includes: detecting high-risk words in the service request through named entity recognition and generating a risk word score based on the high-risk words; identifying urgency expressions in the service request through semantic analysis and generating an urgency score; generating a historical health risk score for the service recipient based on the service recipient's historical information; and calculating a weighted sum of the risk word score, urgency score, and historical health risk score to obtain the urgency score.

[0082] In this example, an emergency vocabulary and semantic matching rules can be pre-set. When the named entity recognition stage detects predefined high-risk words (such as "fainted," "unconscious," "unable to breathe," "fell") or identifies urgent expressions through semantic analysis, the module automatically calculates an urgency score.

[0083] The urgency level score can be expressed as:

[0084] in, E Rate the urgency level. R Rate risky keywords S Assess urgency level. H Historical health risk score, For the preset weights, and satisfying Among them, risk word scoring R This score reflects the occurrence and severity of high-risk words in the requirement text. In practical scenarios, scores corresponding to different high-risk words can be pre-set. If at least one high-risk word appears, the score with the highest score is selected as the risk word score; otherwise, the risk word score is 0. Semantic urgency score. S This score reflects the urgency expressed by the overall semantics of the text; its specific score can be automatically output by a large language model; historical health risk score. H This score, used to reflect the baseline risk level in a service recipient's past health records (i.e., historical information), can be obtained through professional scoring. Of course, H It can also be automatically calculated based on the content of the service recipient's health record. For example, it can be summed up according to preset conditions: 1. Whether the person has a chronic disease (cardiovascular disease, diabetes, etc.) +0.3 if yes; 2. Whether the person has an emergency medical record in the past 30 days +0.3 if yes; 3. Whether the person is ≥80 years old +0.2 if yes; 4. Whether the person is "living alone" or "disabled / semi-disabled" +0.2 if yes. H = min(1.0, the sum of the above terms). HThe score is automatically updated along with the health record. The service recipient's past health record can be provided by the established service provider information database, and may include fields such as the elderly person's past disease markers and recent service frequency. The calculated result is normalized and output as an urgency score between 0 and 1.

[0085] The above-mentioned scoring scheme comprehensively considers the severity level of risk words, the overall semantic tendency of the demand text, and the historical health record tags of the service recipients, and outputs a normalized urgency score. As a result, it can comprehensively and multidimensionally determine the urgency level of service demands and effectively improve the accuracy of subsequent service recommendation circles.

[0086] In some embodiments, when the urgency score exceeds a preset score threshold, the request can be automatically marked as "priority processing" or "emergency intervention" and prominently displayed on the local processing interface (such as the display screen of a village collective or community service center) to help managers (such as village collectives or community cadres) quickly pay attention to high-risk requests.

[0087] For example, the intent recognition model is also used to output the intent recognition confidence score corresponding to the service type label; the method also includes: performing named entity recognition on the semantic representation to obtain the information corresponding to each slot in the structured slot template and the entity recognition confidence score, wherein the structured slot template includes at least slots for filling in the service object name, service address, expected service time period and urgency description.

[0088] As described above, semantic representation includes sentence vector representation and word vector representation. In this example scheme, named entity recognition and intent recognition can share the output of the pre-trained language model. In this case, the three can form a joint training architecture with a shared encoding layer and dual output heads to reduce inference overhead; of course, they can also be deployed independently as two serial modules. When using the shared encoding layer approach, named entity recognition can perform word-by-word label prediction based on word-level context representation, automatically extracting the structured slot information (i.e., the structured slot template after filling in the information) required for service execution, including at least: the name or identity of the service object, the service address (which can be resolved to a specific door number or reference point), the expected service time period, and urgency descriptors. The extraction results are output in key-value pair form, for example, {object: "Grandpa Zhang", address: "Room 202, Building 3", time: "this afternoon", keyword: "chest tightness"}.

[0089] In one specific embodiment, the structured slot template includes the following fields: 1. Service object field, used to record the name, title, or identity designation of the service object; 2. Object identity field, used to record the object type, such as self, parents, neighbors, elderly living alone, etc.; 3. Service address field, used to record the original address text; 4. Standardized Location field, used to record service locations after address resolution or location standardization; 5. Reference location field, used to record auxiliary location information such as building, house number, unit number, and nearby landmarks; 6. The Expected Service Time field is used to record specific times, relative times, or vague time expressions; 7. Service Item Keyword Field, used to record keywords such as "grocery delivery", "accompanying to medical appointments", "pharmacy pickup", and "home visit"; 8. Risk term field, used to record risk-related expressions such as "chest tightness", "fell", "difficulty breathing", "fainted"; 9. Urgency description field, used to record urgency expressions such as "hurry up," "immediately," "come now"; 10. The submitter identity field is used to record the identity of the person who initiated the request, such as the submitter, family member, volunteer, or community worker; 11. Original requirement description field, used to retain a complete natural language expression; 12. Supplementary contact field: Record supplementary information such as contact information, community / village identifier, and historical service association identifier, where available.

[0090] The structured slot template can include more slots, which will not be elaborated here. When the structured slot template includes many slots, the slots used to record the service recipient's name, service address, expected service time period, and urgency description can be designated as key slots. The entity recognition confidence score is obtained by weighting the recognition probabilities of each key slot.

[0091] In some embodiments, when some slots in the structured slot information obtained from named entity recognition are empty, they can be automatically filled in based on the basic information of the service object in the database. For example, when a service address is empty, it can be automatically filled in. Alternatively, a prompt message can be returned to the terminal that sent the service request when a slot is empty, to remind the corresponding operator to supplement the information. In this case, the method further includes: when there are empty slots in the structured slot information, marking the empty slots and returning a prompt message to the terminal that sent the service request, to remind the operator to supplement the information.

[0092] When filling slots, if the text contains omissions, colloquialisms, vague expressions, or non-standard expressions, the system can combine word-level context representation, elderly care service scenario dictionary, and address rule base to perform semantic completion and preliminary standardization. For example, expressions such as "my mom," "Grandpa Zhang," "downstairs at Old Li's house," "a little later today," and "check it out quickly" can all be parsed into relatively stable structured fields within the context.

[0093] In some embodiments, the method further includes integrating the original text of the service request, service type tags, structured slot information, urgency score, and retained credentials of the service request into a record of pending requests. The retained credentials include the original audio file or original text content, speech recognition transcription results, submission timestamp, submission terminal identifier, submitter account identifier, and a snapshot of the automatic parsing results, used for subsequent verification, traceability, and accountability. This facilitates local processing and record retention. In practical scenarios, this record can first be sent to the community / village-level edge node corresponding to the service recipient's location, rather than being directly uploaded to the central platform.

[0094] Before determining the recommendation service circle, the method also includes: generating a system resolution confidence score based on intent recognition confidence score, entity recognition confidence score, and the completeness of key fields of service requirements; the step of determining the recommendation service circle is performed when the system resolution confidence score is higher than the preset confidence threshold and all key slots in the structured slot template are not empty. Key slots include slots for filling in the name of the service object, service address, expected service time period, and urgency description.

[0095] The system resolution confidence score is used to measure the reliability of the automatic resolution results of the current requirement record. In some embodiments, the system resolution confidence score can be calculated using the following formula:

[0096] in, To determine the confidence level for intent identification, Confidence level for entity identification Score the completeness of key fields. m 1. m 2. m 3 is the preset weight, and satisfies m 1+ m 2+ m 3 = 1. Wherein, It can be determined by the maximum probability result output by the intent recognition model. It can be obtained by weighting the recognition probabilities of each key slot. The determination can be made based on whether core fields such as the name of the service recipient, service address, expected service time period, and urgency description are complete. When the system's resolution confidence level falls below a preset threshold, or when key slots (such as service addresses or service recipients) are missing, fields requiring manual correction can be automatically marked and highlighted. In practical scenarios, service requirements and corresponding structured slot information can be displayed on the administrator's terminal device, highlighting fields requiring manual correction. If no correction is needed, the administrator can directly confirm, and upon receiving confirmation from the administrator, the subsequent steps of determining the recommended service circle can continue. Within edge nodes (the village collective or community where the service recipient resides), the head of the elderly community can complete the initial acceptance operation within their authorized scope, and the head of the neighborhood committee / village committee has the final confirmation right for the acceptance result. At this time, the acceptance time, acceptor identifier, and acceptance node ID can be recorded in an additive-only manner, forming the starting point record of the service processing chain. The aforementioned solution transforms colloquial descriptions into structured requirement fields using named entity recognition, replacing manual reading and data entry by community staff and improving processing efficiency. Simultaneously, by evaluating the system's parsing confidence level, it effectively eliminates invalid requirement data with incomplete information or significant parsing biases, greatly improving the accuracy and completeness of requirement information parsing. This ensures that subsequent service circle recommendation work has sufficient and valid evidence and helps guarantee the efficiency and effectiveness of subsequent resident mutual assistance.

[0097] For example, determining a recommended service circle based at least on service type labels and urgency scores includes: determining an initial service circle in a preset two-dimensional mapping table based on service type labels and urgency scores, wherein the preset two-dimensional mapping table is used to represent the correspondence between service type labels, different urgency score intervals and each service circle; determining whether the service object meets correction conditions based on the service object's historical service information and health record, the correction conditions including a first correction condition and a second correction condition, the first correction condition being that the historical service information shows that the corresponding service object's most recent preset number of similar needs were completed in service circles outside the initial service circle more than a preset number of times, the second correction condition being that the service object's health record has a high-risk chronic disease marker related to the service type, and the difference between the urgency score and the threshold value of any urgency score interval is less than a preset difference threshold; when either correction condition is met, the initial service circle is corrected according to the satisfied correction condition to determine the recommended service circle.

[0098] In some embodiments, a preset two-dimensional mapping table is shown in Table 1 below, where the urgency score is represented by a "Score". The first service circle, the second service circle, the third service circle, and the fourth service circle are abbreviated as Circle One, Circle Two, Circle Three, and Circle Four, respectively.

[0099] Table 1 Preset Two-Dimensional Mapping Table

[0100] In the table above, "Circle 3 + Warning" indicates that the starting circle is Circle 3, and a warning notification is sent to the node manager (i.e., administrator); "Circle 3 + Forced Warning" indicates that Circles 1 and 2 are skipped, and order dispatching starts directly from Circle 3, and a forced intervention notification is sent to the node manager. This mapping table is a configurable parameter, and the default configuration is uniformly managed by the central platform. Each edge node can apply to adjust the mapping relationship of specific rows and columns within its authorized scope according to the service resource status of its jurisdiction.

[0101] After determining the initial service area, it's advisable to assess whether modifications are needed based on the service recipients' historical service information and health records. The initial service area, recommended service area, and reasons for modification can be recorded in the corresponding service log for traceability.

[0102] The above technical solution can quickly determine the initial service circle in the table based on service type tags and urgency scores, and can optimize and adjust the service circle according to the actual correction conditions. Thus, it can not only ensure the efficiency and basic standardization of the initial matching of service circle recommendations, but also achieve personalized and accurate correction by combining users' past service habits and health risk status, effectively improving the scientificity, relevance and adaptability of service circle recommendations. In this way, it can find service providers that meet the needs of service recipients, improve the mutual assistance effect and the user experience of service recipients.

[0103] For example, the initial service circle is modified according to the modified conditions, including: when only the first modification condition is met, the initial service circle is modified using a first modification operation, and the modified initial service circle is determined as the recommended service circle; when only the second modification condition is met, the initial service circle is modified using a second modification operation, and the modified initial service circle is determined as the recommended service circle; when both the first and second modification conditions are met, the initial service circle is modified using the first modification operation and the second modification operation respectively to obtain the first modification result and the second modification result; the service circle that is outer between the first modification result and the second modification result is selected as the recommended service circle; wherein, the first modification operation includes: adjusting the current initial service circle to the next service circle; the second modification operation includes: compensating the urgency score according to a preset risk compensation value, and re-determining the initial service circle in a preset two-dimensional mapping table based on the compensated urgency score.

[0104] The first and second correction operations are illustrated below with specific examples. For instance, if more than half of the similar requests from a service recipient in the last N times (default N=3) ultimately required escalation to an outerer service circle to complete the service, the recommended starting service circle can be moved up one level (i.e., adjusted to the next service circle). For example, if the mapping table recommends circle two, but history shows that the elderly person's meal assistance needs were repeatedly escalated to circle three, then circle two can be moved up one level to circle three, starting directly from circle three.

[0105] If the health record of the service recipient contains a high-risk chronic disease marker related to the current service need type, and the difference between the current urgency score and the critical value of the urgency score interval is less than the preset difference threshold (e.g., the difference between the urgency score of 0.38 and the critical value of 0.4 is less than the preset difference threshold of 0.05), then a preset risk compensation value δ (default δ=0.05) can be added to the initial urgency score before the table is looked up again to avoid the starting service circle being too low due to the score being slightly lower than the threshold.

[0106] In the above technical solution, when at least one correction condition is met, the initial service circle is corrected through the corresponding correction operation. This can improve the adaptability of the final recommended service circle, ensure the mutual assistance effect, and prevent the service demand from being unable to be fulfilled due to insufficient capabilities of the service entities within the service circle.

[0107] For example, pushing service request notifications to each service entity within the recommended service circle includes: determining the suitability score for each service entity within the recommended service circle; sorting the service entities within the recommended service circle according to their suitability scores from highest to lowest to obtain a push sequence; pushing service request notifications to each service entity in batches according to the push sequence with a preset push step size; stopping the push if confirmation information is received from a service entity in the current push batch within a preset waiting time; otherwise, pushing service request notifications to the next batch of service entities; wherein, the suitability score is determined by at least the sum of spatial distance score, historical service quality score, order load score, and skill matching score.

[0108] In this embodiment, the spatial distance score is determined based on the spatial distance between the service provider and the service recipient; the historical service quality score is determined based on the service provider's historical service information, which includes historical service completion rate, average response time, and service feedback score; the order load score is determined based on the number of service tasks that the service provider has currently accepted but not completed, and the order load score is negatively correlated with the number of service tasks; and the skill matching score is determined based on the degree of matching between the service provider's skill tags and the service type tags of the service requirements.

[0109] In this example, a multi-objective weighted approach is considered to evaluate each service provider. Specifically, the evaluation is based on spatial distance, historical service quality, current order load, and skill matching, serving as the basis for deciding the order of subsequent push notifications.

[0110] Spatial distance scores are determined based on the spatial distance between the service provider and the service recipient. Specifically, the road network distance or straight-line distance between two points can be calculated based on the real-time location coordinates of the service provider (obtained through terminal device positioning) and the service address coordinates parsed from the service request. The closer the distance, the higher the score in this dimension (different distance ranges and their corresponding scores can be preset, and the score can be determined through this correspondence).

[0111] For the historical service quality dimension, we can retrieve the service completion records left in the system by the service provider to calculate its service completion rate, average response time, and service feedback ratings from service recipients or their families after the service was completed. Historical Service Quality Rating Q history The evaluation is composed of three sub-indicators: service completion rate, average response time, and evaluation feedback score. The evaluation is weighted and synthesized after being independently normalized, and a time decay mechanism is introduced to give higher reference weight to recent service records.

[0112] Among them, the service completion rate score Q completion It can be determined in the following ways: The percentage of successful service completions after accepting an order is calculated as follows: completion_rate = number of successful completions / total number of orders accepted; this value is already in the range [0,1], so no additional normalization is needed. =completion_rate. Assign a neutral baseline value to newly registered entities with no prior history. = 0.6 indicates that moderate trust is given, but not higher than that given to entities with a good track record.

[0113] Average response time score Q response This can be determined by: statistically analyzing the average time from receiving a push notification to confirming order acceptance for that entity throughout its history. (Unit: minutes), normalized using an inverse proportional decay function: ;in This is a reference baseline duration (configurable parameter, default is 10 minutes). The faster the response, the better. The closer a value is to 1, the slower the response; the closer a value is to 0. For example, =5 minutes ≈0.67, =10 minutes =0.50, =30 minutes =0.25. A neutral benchmark value is assigned to newly registered entities with no historical records. = 0.5.

[0114] Evaluation feedback scoreQ rating The following method is used to determine the score: If evaluation data exists, calculate the evaluation scores received by the entity after all historical service completions (assuming a score of 1 to 5), and normalize them to the [0,1] interval: ;in This is the arithmetic mean of historical evaluation scores. If the entity has no evaluation records (e.g., no evaluations received during new registration or previous services), a neutral baseline value is assigned. = 0.6.

[0115] The above three sub-indicators are weighted and summed according to preset weights to obtain the comprehensive historical service quality score:

[0116] in The default weighting is primarily based on completion rate, followed by response time, with evaluation score serving as a supplementary reference. Each weight coefficient is a configurable parameter, which administrators can adjust according to the service management objectives of their jurisdiction.

[0117] In some embodiments, considering that the service capabilities of a service provider may change over time, a time decay weight is introduced into the calculation of each sub-indicator, making recent service records contribute more to the score than older records. For a service record t days ago, its weighting coefficient in the statistics is... ,in Attenuation coefficient (configurable parameter, default) =0.01, corresponding to a weight decay of approximately 50% for records from about 70 days ago.

[0118] The statistical mean of each sub-indicator is recalculated after incorporating time decay weights, and then substituted into the above normalization and synthesis formulas to obtain the final mean. Q history For newly registered entities with no historical records, the system assigns a neutral benchmark score. For the current order load dimension, the number of service tasks that the service entity has accepted but not yet completed can be queried, and the remaining order capacity can be calculated based on its preset maximum order capacity. The lower the load and the more sufficient the remaining capacity, the higher the score for this dimension. When the current load of an entity has reached or exceeded its capacity threshold, the score for this dimension is zero. At this time, it can be removed from the candidate set or downgraded for promotion. In a specific embodiment, the order load score can be the ratio of the remaining order capacity to the maximum order capacity.

[0119] For the skills matching dimension, service type tags can be matched with the available skills tags registered by the service provider. Tag overlap scoring or cosine similarity based on embedding vectors can be used, and the result is taken as the skills matching score.

[0120] After the scores of each dimension are normalized, they can be weighted and summed according to the preset weight coefficients to obtain the comprehensive suitability score of the service provider for the current needs.

[0121]

[0122] in + + + =1, Scoring is done for the distance dimension. Scoring is done based on the load dimension. Scoring is assigned to the skills matching dimension. The weighting coefficients are configurable parameters. For example, the weighting coefficient for the spatial distance dimension can be dynamically adjusted based on urban-rural regional differences; in rural areas, distance sensitivity can be appropriately reduced to accommodate the dispersed residential characteristics. Furthermore, community / village-level edge node leaders can adjust these weightings based on their actual operational experience within their jurisdiction. For instance, in communities emphasizing a neighborhood mutual assistance atmosphere, the weighting of the historical service quality dimension can be appropriately increased; in scenarios emphasizing rapid response, the weighting of the spatial distance dimension can be appropriately increased.

[0123] After obtaining the suitability score, the data can be sorted from highest to lowest, and service request notifications can be pushed to each service entity in batches according to a preset push step size (e.g., 3-5). After pushing a batch, a preset waiting time is waited. If no confirmation information is received within the preset waiting time (e.g., all entities in the current batch do not respond or explicitly refuse), the push will proceed to the next batch. Otherwise, the target service entity is determined based on the confirmation information, and the push is stopped.

[0124] The above technical solution uses a multi-objective weighted scoring system that integrates spatial distance, historical service quality, current order load, and skill matching to automatically generate a priority sequence for candidate entities. This replaces manual subjective order assignment and makes the decision-making process quantifiable, explainable, and traceable.

[0125] For example, pushing service request notifications to each service entity within the recommended service circle further includes: marking service entities within the recommended service circle with an attribution probability greater than a second probability threshold as high-confidence entities, and marking other service entities as entities to be verified; wherein, for any high-confidence entity in the recommended service circle, the suitability score of that service entity is the sum of its spatial distance score, historical service quality score, order load score, and skill matching score; for any entity to be verified in the recommended service circle, the suitability score of that service entity is the product of the sum of its spatial distance score, historical service quality score, order load score, and skill matching score, and a penalty coefficient, wherein the penalty coefficient is less than 1. The second probability threshold is greater than the first probability threshold.

[0126] In this example, the scheme considers further segmenting the service entities within the service circle based on their affiliation probability. Entities with affiliation probabilities reaching or exceeding the second probability threshold (e.g., 0.7) are marked as "high-confidence candidates" and directly included in the candidate set. Entities with affiliation probabilities below the second probability threshold but above the first probability threshold (e.g., 0.4) indicate uncertainty regarding their affiliation (e.g., the entity possesses dual roles as a neighborhood mutual aid provider and an organizational backbone). In this case, they can be marked as "candidates awaiting verification" and also included in the candidate set. However, a penalty coefficient λ (0 < λ < 1) will be added to their subsequent fit score calculation. This coefficient affects the weighted result of the comprehensive score, shifting their ranking position in the push sequence accordingly. This design aims to: avoid completely excluding such boundary-state entities, preventing over-screening and resulting in a lack of available personnel during resource constraints; and implement the principle of "prioritizing high-confidence service circle entities" by lowering their push priority. This helps to further improve the scientific nature of the push process.

[0127] For example, the method further includes: when none of the service entities in the recommended service circle accept the order, pushing a service request notification to the next service circle in the recommended service circle. For instance, if no one accepts the order in circle two, circle three is selected as the recommended service circle, and step S130 is executed again.

[0128] In this example, if none of the service providers respond or explicitly refuse within the preset waiting time for their batches, it can be assumed that none of the service providers in the current recommended service circle have accepted the order. In some solutions, a service circle timeout threshold can be preset. In this case, the method further includes: determining that none of the service providers in the recommended service circle have accepted the order when the total response time of the recommended service circle reaches the service circle timeout threshold. The service circle timeout threshold can be a preset fixed value or a dynamic value. In this solution, the preset waiting time is the ratio of the service circle timeout threshold to the total number of batches of service providers in the recommended service circle.

[0129] In this example solution, the system can automatically push updates to the next service circle if none of the service providers in the recommended service circle have accepted the order. This improves the automation of upgrade response and avoids situations where no one accepts the order.

[0130] In some implementations, while determining the service entity push sequence for the recommended service circle, the spatial distance score, historical service quality score, and skill matching score of the service entities in the next service circle can be pre-calculated. When upgrading to this service circle, only the order load score needs to be calculated; the other three scores can directly reuse the pre-calculated results, which helps to further improve computational efficiency.

[0131] In some implementation schemes, the method further includes: when the initial service circle is the third service circle, if the load of each service entity in the third service circle is saturated, a horizontal coordination request is generated and sent to the administrator node; upon receiving confirmation from the administrator node, the load of service entities in the third service circles of adjacent service circles of the current service area is obtained; if the load of service entities in the third service circles of adjacent service circles is not saturated, the service request notification is pushed to each service entity in the third service circle of the adjacent service area. In this scheme, if the recommended service circle is circle three, and the candidate entities in circle three are generally saturated, the availability of entities in the same circle among neighboring edge nodes can be queried to perform horizontal resource coordination, reducing dependence on external professional layers. In this case, the administrator node can confirm the horizontal coordination request and can also pre-define service areas that can be horizontally coordinated.

[0132] For example, the service circle timeout threshold is a dynamic value, which is determined by calculating the service circle timeout threshold based on the urgency score of the service request, the time period in which the service request was initiated, and the historical average response time of the recommended service circle.

[0133] In some embodiments, calculating the service sphere timeout threshold includes: determining an urgency factor based on an urgency score; determining a time factor based on the time period in which the service request was initiated; determining a historical response factor based on the historical average response time; and calculating the product of a preset base duration and the urgency factor, time factor, and historical response factor to obtain the service sphere timeout threshold.

[0134] In the above embodiments, a dynamic timeout threshold calculation mechanism is considered to determine the service timeout threshold of the current recommended service circle based on the urgency score of the service request, the time period in which the service request was initiated, and the historical average response time of the recommended service circle.

[0135] In some embodiments, the urgency factor can be derived from an urgency score, mapped to an adjustment coefficient via a piecewise linear function. A higher urgency score results in a shorter timeout threshold. For example, a request marked "urgent intervention" may have its waiting time reduced to 50% of the normal time. Example: =15 minutes, if the urgency score of a certain need is 0.91, then the urgency factor is... =0.4, the compressed duration of this factor is 15×0.4=6 minutes.

[0136] The time factor can be adjusted based on the time the request is submitted (daytime, nighttime, mealtime, etc.). During nighttime hours, because the service provider is more likely to be resting, the waiting threshold can be appropriately extended to avoid ineffective escalations. For example, if the same request is submitted at 22:00 at night, the time factor... =1.5, submitted at 10:00 AM =1.0.

[0137] Historical response factors can be obtained by retrieving the historical average response time of the current recommendation service area over the past thirty days at its corresponding edge node (i.e., the village or community where it is located). If historical data shows a fast response in this cycle, a shorter threshold can be maintained; if historical data shows a slow response in this cycle, the threshold can be appropriately shortened to accelerate the upgrade pace. For example: =15 minutes, the historical average response time of the second service circle of a certain village over the past 30 days. =25 minutes (slow response), then historical response factor = 15 / (15+0.5×10) = 15 / 20 = 0.75, which means that the village has historically had a slow response time, and the system has shortened the waiting time to speed up the upgrade process.

[0138] The formula for calculating the service circle timeout threshold is:

[0139] in Based on the waiting time, , These are the adjustment coefficients for each factor. Each coefficient is a configurable parameter, and the node manager can adjust them based on operational experience within their jurisdiction.

[0140] The dynamic timeout threshold method adopted by the above scheme can dynamically determine the service circle timeout threshold based on factors such as urgency and service time difference. This can fit the differences in urgency priority of service needs in different scenarios, adapt to the service operation status at different times, and reasonably define the timeout standard based on the actual response of the service circle in the past, effectively improving the scientific and rational nature of timeliness management.

[0141] For example, calculating the service circle timeout threshold further includes: determining whether the service circle timeout threshold is within a preset timeout interval; if the service circle timeout threshold is not within the preset timeout interval, adjusting the service circle timeout threshold to the nearest value in the preset timeout interval; In this example, to prevent the service circle timeout threshold from reaching extreme values, we consider ensuring that the service circle timeout threshold meets the following criteria: ≤ ≤ ,in and The minimum and maximum waiting times preset for the system (e.g.) =3 minutes =30 minutes). This is the service timeout threshold for the currently recommended service area.

[0142] For example, the method further includes: upon receiving a risk escalation request from the target service subject, adjusting the urgency score based on the risk escalation request, and re-executing the step of determining the recommended service circle based at least on the service type label and the urgency score based on the adjusted urgency score.

[0143] In this example, when the target service entity triggers a proactive risk escalation request during service execution, the urgency score can be recalculated based on the reported risk description. If the recommended service circle corresponding to the recalculation result is higher than the next service circle of the current service circle, the escalation jumps to the recommended service circle instead of being passed down circle by circle. This avoids the response delay caused by fixed linear escalation.

[0144] For example, the method further includes: after receiving the correction information from the administrator, modifying the service requirements according to the correction information, and re-executing the step of determining the recommended service circle in each service circle based at least on the service requirements based on the modified service requirements.

[0145] In this example scenario, if the administrator modifies the service requirements (e.g., changes the service type label) before determining the target service subject, the recommended service circle can be redefined. If the redefined recommended service circle is higher than the current recommended service circle, the service circle is skipped directly.

[0146] For example, the method further includes: locking the service task corresponding to the service request after the preset lock time for receiving the confirmation information for the first time; when there are multiple confirmation messages within the preset lock time, sorting the confirmation messages according to the receiving time of each confirmation message; sequentially traversing the suitability scores of the service entities corresponding to each confirmation message according to the sorting results; and determining the current service entity as the target service entity when the suitability score of the currently traversed service entity is higher than the minimum available threshold.

[0147] Research has revealed that, due to factors such as the real-time nature of signal transmission and the execution time of the locking operation, concurrent responses may occur within a very short time window. In this example solution, a preset locking time is considered after receiving the first confirmation message before locking the service task. This preset locking time can be determined based on actual needs, for example, 1 second. After locking the task, if multiple confirmation messages exist, the responders can be sorted chronologically according to their precise timestamps to initially determine the earliest responder.

[0148] Then, eligibility verification can be performed according to the sorting order and the suitability scores of each service provider. Specifically, the suitability score of the earliest responder can be retrieved first. If the suitability score is higher than the minimum usable threshold, the earliest responder is identified as the target service provider. Otherwise, it indicates that although the responder responded quickly, their overall suitability is insufficient (e.g., low skill matching or poor historical service quality). The system then moves on to the next earliest responder in the timestamp sorting and repeats the verification process until a responder that meets the threshold requirements is found. The minimum usable threshold can be automatically calculated based on the statistical distribution of the suitability scores of each service provider in the recommended service circle, taking the historical mean minus one standard deviation as the threshold, and setting global upper and lower bound constraints. If historical data accumulation is insufficient, a global default threshold is used; once sufficient data is available, it automatically switches to a locally statistical adaptive threshold.

[0149] Unselected responders will have their responses automatically recorded as "unsuccessful" and their reserved order-accepting capacity in the load calculation will be released, restoring their order-accepting status so that they can participate in matching other subsequent needs.

[0150] The above scheme can ensure the uniqueness and rationality of the dispatch result through an arbitration mechanism that combines timestamp priority and qualification threshold verification when multiple people respond concurrently.

[0151] In some implementation schemes, the service request requires at least two service providers. In this case, users can specify the required number of personnel when submitting the request, or the node manager can determine and mark the required number of personnel based on the request content. When the two markings are inconsistent, the node manager's marking shall prevail, and the system shall simultaneously push a notification to the request initiator to adjust the number of personnel. After the service team is formed, it is verified whether the skill tag set of the team members covers all the sub-tasks involved in the request. If the coverage is incomplete, the node manager shall be notified of the skill gap.

[0152] In some implementation schemes, the initial service circle is modified according to the modified conditions, which further includes: when the initial service circle is the third service circle and the modified service circle is the fourth service circle, generating an upgrade service circle request and sending it to the administrator node; upon receiving confirmation from the administrator node, determining the fourth service circle as the recommended service circle. When the recommended service circle is circle four, the demand data packet can first be sent to the township-level intermediate node to request administrator confirmation. If confirmation is received from the administrator, the recommended service circle can be upgraded to circle four. Of course, the administrator node can also send a horizontal coordination command. Upon receiving the horizontal coordination command, the service request notification can be pushed to each service entity within the third service circle of the adjacent service jurisdiction.

[0153] When township-level coordination fails, the data packet is reported to the county-level central platform for cross-node arbitration. This may result in a multi-response scenario involving multiple service providers from different edge nodes responding simultaneously. The central platform intervenes to arbitrate the cross-node response; the decision-making logic includes: Distance priority principle: Calculate the road network distance between each responding entity and the demand address, and give priority to those that are closer.

[0154] Node affiliation balance principle: If the distance is close, priority is given to the response subject from the node with fewer historical cross-node service outputs.

[0155] Distributed synchronization of arbitration results: After arbitration is completed, the central platform synchronizes the arbitration results (selected service entity identifier, associated node, and summary of arbitration basis) to the edge node that initiated the request, the edge node to which the selected entity belongs, and the edge nodes to which all unsuccessful entities belong. Simultaneously, complete data from this cross-node coordination (request origin node, response node, distance data, and arbitration basis) is recorded in the central platform log for subsequent regional elderly care service resource scheduling and analysis.

[0156] For example, the method also includes: recording the entire process of service request processing in real time. Specifically, recording operations can be automatically triggered or manually triggered by relevant personnel at the following key nodes throughout the entire lifecycle of a service request from acceptance to completion, forming a complete and traceable service processing chain.

[0157] The data nodes recorded by the system include at least: (1) Requirement submission: Record the original requirement text or voice file, submitter identifier, submission timestamp, and structured requirement fields (service type label, urgency score, slot filling result).

[0158] (2) Node acceptance: Record the acceptance node ID, acceptor identifier, acceptance timestamp, and structured requirements content that has been confirmed or corrected after manual verification.

[0159] (3) Circle determination and push: Record the starting response circle, the circle belonging probability value usage method (high confidence or candidate to be verified), the comprehensive fit score and ranking result of each candidate subject in the push sequence.

[0160] (4) Response results: Record the identifier of the entity accepting the order and the confirmation timestamp, or the identifier of the entity refusing the order, the reason for refusal, and the timestamp. If multiple response arbitration is triggered, record the arbitration basis (timestamp order, qualification threshold verification results) and the final confirmed executor.

[0161] (5) Upgrade flow: If an upgrade is triggered, record the upgrade trigger conditions (timeout, capability mismatch, active upgrade request), upgrade path (upgrade by circle or skip-level upgrade), pre-calculated and real-time refreshed execution records, and the target service circle or target node information after the upgrade.

[0162] (6) Cross-node coordination: If cross-node horizontal coordination or the use of professional resources in Circle 4 is involved, record the arbitration basis of the central platform (distance priority, load balancing), the node ID of the selected entity and the arbitration timestamp.

[0163] (7) Service execution status: Service personnel can report changes in service execution status (departed, arrived, service in progress) through the terminal. The system records the time of status change and the corresponding geographical coordinates (after privacy desensitization processing).

[0164] (8) Service completion: Record the service completion time, service summary (to be filled in by the service personnel), and the index of the voucher materials uploaded at the time of completion.

[0165] All records are written using an add-only log method, with each log entry including an operator identifier, timestamp, and hash checksum to ensure immutability. Log data is distributed and stored locally on each edge node involved in the processing, while copies are uploaded to the central platform via an asynchronous synchronization mechanism. After aggregating the entire process logs, the central platform can generate a complete traceability view of any service request from initiation to completion.

[0166] The aforementioned solution employs distributed logging throughout the entire service lifecycle, recording data in a continuously updated manner for key stages such as request submission, node acceptance, service zone determination, task distribution, response results, upgrade workflow, and service completion. This forms a complete and traceable service processing chain from request initiation to service confirmation. Based on this, the central platform performs real-time analysis of the global log data, automatically identifying service quality risk signals such as frequent response timeouts, abnormal entity rejection rates, and high service zone upgrade ratios. This transforms passive recording into proactive early warning, providing data support for continuous service quality improvement and timely intervention, thus overcoming the shortcomings of existing technologies that only passively record data and lack proactive analysis capabilities. Consequently, a fully traceable responsibility chain can be formed, with the ability to proactively identify abnormal patterns.

[0167] The effects of the embodiments of the present invention will be further illustrated below through two specific examples in a real-world scenario.

[0168] In one specific embodiment, the scenario is described as follows: Ms. Huang, a family member of an elderly person in a community, submitted a voice-to-text request via her mobile phone: "Could you please check on my mother? She said she felt a little dizzy today. She lives in apartment 101, building 3, Sunshine Community. It would be best if you could go there before noon." In this embodiment, after the service request information is submitted through the user terminal, the system first determines which community / village-level edge node to send the request for processing. The system maintains a binding relationship table between service recipients and edge nodes on the central platform. When each elderly person is first included in the system management, a unique binding relationship is established with the corresponding community / village-level edge node based on the administrative division code and grid division of their permanent address.

[0169] In this example, the service recipient, "Ms. Huang's mother," has already registered in the system and is bound to the "Sunshine Community First Grid Edge Node." The system queries the binding relationship based on the service recipient's identity and automatically routes the request to this node. This node belongs to the second level (Community Neighborhood Committee) in terms of management hierarchy. The node has two levels of operational permissions—the head of the first-level elderly community can complete the initial acceptance within their authorized scope, while the head of the second-level neighborhood committee holds the final confirmation authority.

[0170] After receiving the service request text, the edge node performs a natural language parsing process. A domain-fine-tuned pre-trained language model semantically encodes the text "Could you please check on my mom? She said she's feeling a bit dizzy today," and performs the following tasks in parallel: Intent Classification: The model outputs a probability distribution of service type labels. The probability of "health monitoring" is 0.72 (the highest value). Based on this, the system classifies the demand as "health monitoring".

[0171] Entity extraction: The model extracts structured slot information: {Service object: "My mother" (associated with registered elderly ID 3101), address: "Building 3, Room 101, Sunshine Community", symptom keyword: "dizziness", expected time: "before noon"}.

[0172] The urgency scoring module identified the symptom keyword "dizziness" as a risk term requiring attention. Combined with the context that the elderly person had no history of serious chronic diseases in their health record, the system output an urgency score of 0.65 (out of 1.0), higher than the baseline score of 0.3 for general light services but below the emergency intervention threshold of 0.8. Based on this, the system marked the request as "priority attention" and highlighted it in yellow on the node manager's interface.

[0173] After the node manager verifies and confirms the system's automatic parsing results, the system enters the initial response service circle determination phase. Following this determination, the initial service circle is determined to be Circle 3. Further adjustments to the service circle are made based on the service recipient's historical service information and health records.

[0174] Service recipient's historical service information: The elderly person had a record of calling for services once in the past 30 days due to dizziness. At that time, the staff of the community health service center in Circle 3 came to handle the situation. The historical pattern suggests that similar needs may require the intervention of an entity with health assessment capabilities.

[0175] Based on the above three dimensions of information, the system outputs the initial response circle as circle three, with the judgment criteria being the basic mapping result (health monitoring category + score 0.65), and displays the judgment criteria on the node leader's interface for confirmation. The node leader reviewed the judgment and deemed it reasonable, confirming execution. Upon verification, the elderly person had only one similar request record in the past 30 days, and it was not upgraded to the outer circle for completion, thus not meeting the triggering conditions for the first correction condition; the current emergency score of 0.65 is not in the critical range, thus not meeting the triggering conditions for the second correction condition. Therefore, dynamic correction was not triggered, and the basic mapping result circle three was maintained.

[0176] The system selects entities from the service entity information database whose circle tag is Circle 3 and whose circle affiliation probability value is not lower than the first probability threshold to form the candidate set. This set includes 2 neighborhood committee staff, 3 grid workers, and 2 medical staff from the community health service center. These entities belong to different levels in terms of management hierarchy—neighborhood committee staff and grid workers belong to the second level, while medical staff from the community health service center belong to the fourth level (professional resources under the overall management of district and county elderly community organizations).

[0177] For each subject in the candidate set, the system performs a multi-objective weighted scoring. Take, for example, a nurse from a community health service center: Spatial distance score: The nurse is currently located at a community health service center, approximately 600 meters from the service address, with a score of 0.85.

[0178] Historical service quality rating: 100% completion rate, average response time of 8 minutes, three recent five-star ratings, score of 0.92.

[0179] Order load score: There are currently no ongoing tasks, and the remaining order capacity is sufficient, with a score of 1.0.

[0180] Skills Match Score: The skill tags include "health monitoring" and "basic care," which are highly matched with the need type, with a score of 0.95.

[0181] The nurse's probability of belonging to a specific social sphere is 0.89, which is higher than the confidence threshold of 0.7, classifying her as a "high-confidence candidate" with a penalty coefficient λ=1.0. After weighted calculation across all dimensions, her overall suitability score ranks first in the candidate set. The system sends a service request notification to her terminal according to the push notification step size.

[0182] The nurse confirmed the order within 3 minutes. The system recorded the order time and the nurse's identifier, and sent a notification to Ms. Huang's terminal stating, "Nurse Li from the community health service center has accepted the order and is expected to arrive within 15 minutes." During the service, the nurse reported "Arrived" upon arrival, and the system recorded the arrival time and location coordinates. Upon examination, the elderly woman's dizziness was found to be caused by fluctuations in blood pressure. The nurse measured her blood pressure, instructed her on medication, and observed her until her symptoms subsided. After the service was completed, the nurse filled out a service summary on her terminal and submitted it.

[0183] Ms. Huang received a service completion notification via her mobile device, clicked the confirmation button to confirm the service, and gave a five-star rating. The system marked this service record as "completed," updated the nurse's order-taking load status, and incorporated the evaluation data into its historical service quality data source. Records at each key node of the entire process are distributed and stored on edge nodes in an add-only manner and synchronized to the central platform (fourth-tier district / county senior citizen communities). Second-tier neighborhood committee leaders can view the complete processing chain of this service through node terminals, and fourth-tier district / county senior citizen communities can summarize and statistically analyze service data within their jurisdiction on the central platform.

[0184] In another specific embodiment, the scenario is an emergency rescue scenario, which is described as follows: In a rural area, the wife of villager Mr. Wang submitted a voice request to the system through the one-click call function: "My husband suddenly has severe chest pain and can't breathe. Please come quickly and take a look!" The request was initiated at 21:30.

[0185] After receiving a voice request, the edge node converts the audio to text, and the NLP parsing module performs the parsing in parallel. Intent classification outputs an "emergency assistance" label (probability 0.94), and named entity recognition extracts symptom keywords such as "chest pain" and "shortness of breath." During the urgency scoring process, high-risk word combinations are detected, and combined with the time factor (nighttime), an urgency score of 0.91 is output, exceeding the emergency intervention threshold of 0.8.

[0186] The system automatically marks the request as "urgent intervention" based on the score, displays a red warning on the node manager's interface, and forcibly sets the initial response service circle to circle three (organizational backbone level), skipping the regular waiting process of the first and second circles. At the same time, the system automatically sends an emergency notification to the node manager, reminding them to pay attention immediately.

[0187] The system screened three candidate entities within the village: two village committee staff members (second-tier management), one village clinic doctor (managed by the fourth-tier district / county health system professionally, and coordinated by the second-tier village level organizationally), and two grid workers (second-tier). After multi-objective scoring, the village clinic doctor received the highest overall score, and the system pushed a service request to him. However, this doctor was currently making house calls in a neighboring village, with an estimated return time exceeding 30 minutes, and reported "unable to accept the order" on his terminal. The system then pushed the request to the village committee staff member with the second-highest score, but this staff member had already gone to bed and did not respond. The remaining candidate entities subsequently reported being unable to accept the order or failing to respond within the time limit.

[0188] The system detected that all candidate entities in the village's third circle were currently saturated or unresponsive, triggering the intra-circle horizontal coordination mechanism.

[0189] The edge node sends a service data packet containing demand details, the resource saturation status of its village's third circle, and the history of attempted order dispatch to the third-layer (township elderly community) intermediate node, requesting horizontal coordination within the circle. The third-layer node queries nearby edge nodes (other administrative village nodes within the same township and a road network distance of 5 kilometers) for available entities with the circle label of circle three.

[0190] The query results show that there is a village doctor assistant (with a circle ownership probability of 0.78 and skill tags including "first aid" and "basic care") at the "Lijia Village Edge Node" in the neighboring village. Currently, there are no ongoing tasks, and the real-time location is approximately 3.8 kilometers from Grandpa Wang's house. The third-level node completed arbitration based on the distance priority principle, confirming this entity as the executor of cross-node coordination, and synchronized the arbitration result to the edge node of the village that initiated the request and the edge node of Lijia Village to which the selected entity belongs. In this arbitration, the distance differences among the candidate entities were significant, and the distance priority principle was sufficient to determine the executor; the node ownership balance principle was not triggered. Simultaneously, the complete data of this cross-node coordination was reported to the fourth-level (district / county elderly community) center platform for record-keeping.

[0191] After receiving the order, the village doctor's assistant from Lijiacun arrived at Grandpa Wang's home within 12 minutes. Initial examination suggested a possible precursor to an acute myocardial infarction, a condition beyond the assistant's capabilities. At this point, the system detected that the urgency score (0.91) for this service fell within the trigger range for intervention by professional resources in Circle Four, and the service type was tagged as "Emergency Rescue." Based on the mapping table corresponding to "Circle Three + Mandatory Warning," the system simultaneously issued a warning to the node manager during the order dispatch phase.

[0192] About 3 minutes after the village doctor's assistant arrived at the scene and reported "arrived", the node leader determined through the terminal that the situation required the intervention of the Circle 4 professional resources, and performed a manual confirmation operation on the node terminal to trigger the upgrade process.

[0193] The system determines that the current situation meets the conditions for directly escalating to Circle 4 (emergency score ≥ 0.8 and service type is emergency assistance), and activates the pre-calculated candidate set for Circle 4. The upgrade data packet is relayed through the township-level intermediate node and reported to the county-level central platform, which then forwards it to the township health center node (Circle 4 professional resources). Upon receiving the order, the on-duty doctor at the township health center immediately dispatches an ambulance, which arrives at the scene 25 minutes later and transfers Mr. Wang to the health center for further treatment. The village doctor's assistant remains on standby at the scene to assist the emergency personnel with the handover until the ambulance arrives.

[0194] This service involves cross-flow across three levels and three service circles: acceptance by the village's peripheral nodes (second level), horizontal coordination among neighboring villages (third level arbitration), and professional intervention by the township health center (fourth level resource allocation). Log entries were added only at the following key nodes, without modification: The request was submitted at 21:30, with an emergency score of 0.91, and the mandatory initial service circle was Circle 3. There are records of failed dispatches within Circle 3 of this village (reasons and timestamps for non-response / rejection by various entities). Third-level horizontal coordination was triggered at 21:38, with arbitration based on a distance of 3.8 kilometers, and the selected entity being the village doctor's assistant of Lijia Village. The village doctor's assistant accepted the order at 21:40, arrived at 21:52, and triggered an active escalation request at 21:55. The escalation request was relayed through the third level and forwarded to the township health center in Circle 4. The township health center in Circle 4 accepted the order at 21:58, and the ambulance arrived at 22:15.

[0195] After the above records are aggregated at the fourth-level central platform, a complete service processing chain is formed across levels and service circles. The system's post-event analysis module can retrospectively analyze the event, assess the resource gaps in the village circle, the response time of the third-level horizontal coordination, and the rationality of the fourth-level professional resource allocation, providing data support for optimizing regional resource allocation.

[0196] According to another aspect of the present invention, a computer-readable storage medium is also provided. The storage medium stores a computer program / instructions that, when executed by a processor, implement the method described above. The storage medium may, for example, include a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a portable compact disc read-only memory (CD-ROM), a USB memory, or any combination of the above storage media. The computer-readable storage medium may be any combination of one or more computer-readable storage media.

[0197] Those skilled in the art will readily understand the implementation structure, working principle, and beneficial effects of the system and computer-readable storage medium by reading the above methods. For the sake of brevity, further details will not be elaborated here.

[0198] Although exemplary embodiments have been described herein with reference to the accompanying drawings, it should be understood that the above exemplary embodiments are merely illustrative and are not intended to limit the scope of the invention. Various changes and modifications can be made therein by those skilled in the art without departing from the scope and spirit of the invention. All such changes and modifications are intended to be included within the scope of the invention as claimed in the appended claims.

[0199] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0200] In the several embodiments provided by this invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed.

[0201] Numerous specific details are set forth in the specification provided herein. However, it will be understood that embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures, and techniques have not been shown in detail so as not to obscure the understanding of this specification.

[0202] Furthermore, those skilled in the art will understand that although some embodiments described herein include certain features included in other embodiments but not others, combinations of features from different embodiments are meant to be within the scope of the invention and form different embodiments.

[0203] The various component embodiments of the present invention can be implemented in hardware, or as software modules running on one or more processors, or a combination thereof. Those skilled in the art will understand that microprocessors or digital signal processors (DSPs) can be used in practice to implement some or all of the functions of some modules in the system according to embodiments of the present invention. The present invention can also be implemented as an apparatus program (e.g., a computer program and computer program product) for performing part or all of the methods described herein. Such programs implementing the present invention can be stored on a computer-readable medium or can be in the form of one or more signals. Such signals can be downloaded from an Internet website, provided on a carrier signal, or provided in any other form. It should be noted that the above embodiments are illustrative of the invention and not restrictive, and those skilled in the art can devise alternative embodiments without departing from the scope of the appended claims.

[0204] The above description is merely a specific embodiment of the present invention or an explanation of that embodiment. The scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. The scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for scheduling and managing mutual-aid elderly care based on resident co-governance, characterized in that, include: Based on the information of each service entity in the service entity information database, each service entity in the service entity information database is divided into various service circles. Each service circle is a multi-layered nested circle centered on the service object, and from the inside out, they are the first service circle, the second service circle, the third service circle, and the fourth service circle. The service entity information database is used to store the information of each service entity within the service jurisdiction where the service object is located. Upon receiving a service request from the service object, a recommended service circle is determined in each of the service circles based at least on the service request; The service request notification is pushed to each service entity within the recommended service circle, and the service request notification is generated based on the service requirement. Upon receiving confirmation information from any service entity within the recommended service circle, the service entity that sent the confirmation information will be identified as the target service entity. In response to the service confirmation operation, it is determined that the service corresponding to the service request has been completed.

2. The method according to claim 1, characterized in that, The step of dividing each service entity in the service entity information database into various service circles based on the information of each service entity includes: Based on the information of each service entity in the service entity information database, a social relationship association graph is constructed, wherein the nodes in the social relationship association graph include each service entity in the service entity information database; The social relationship graph is input into a pre-trained service circle mapping model to obtain the probability of each service subject belonging to different service circles. Wherein, for any of the service circles, the service entities within the service circle are service entities whose probability of belonging to the service circle is greater than a first probability threshold; Preferably, the service circle mapping model consists of a graph attention network and a classification output layer, and the step of inputting the social relationship graph into the pre-trained service circle mapping model includes: The graph attention network is used to process the social relationship graph to output a high-dimensional embedding vector for each service subject in the service subject information database, wherein the high-dimensional embedding vector is used to represent the role characteristics of the corresponding service subject in the social relationship graph; The high-dimensional embedding vector corresponding to each service subject is input into the classification output layer to obtain the probability of each service subject belonging to different service circles in the service circle.

3. The method according to claim 1, characterized in that, The step of determining the recommended service circle within each of the service circles, based at least on the service demand, includes: Based on the service requirements, determine the service type label and urgency score for the service requirements; The recommended service area is determined based at least on the service type tag and the urgency score; Preferably, determining the service type label and urgency score for the service request includes: The service requirements are input into a pre-trained language model to obtain a semantic representation of the service requirements; The semantic representation is input into the intent recognition model to determine the service type label of the service request; Preferably, the intent recognition model is further used to output the intent recognition confidence level corresponding to the service type label; The method further includes: Named entity recognition is performed on the semantic representation to obtain the information corresponding to each slot in the structured slot template and the entity recognition confidence. The structured slot template includes at least slots for filling in the name of the service object, service address, expected service time period and urgency description. Before determining the recommended service area, the method further includes: Based on the intent recognition confidence, entity recognition confidence, and the completeness of the key fields of the service requirement, a system parsing confidence is generated; The step of determining the recommended service circle is performed when the system resolution confidence is higher than the preset confidence threshold and all key slots in the structured slot template are not empty. The key slots include slots for filling in the name of the service object, service address, expected service time period and urgency description. Preferably, determining the service type label and urgency score for the service request includes: High-risk words in the service requirements are detected by named entity recognition, and risk word scores are generated based on the high-risk words. Semantic analysis is used to identify urgency statements in the service requirements and to generate urgency scores. Based on the historical information of the service recipients, a historical health risk score is generated for the service recipients. The urgency score is obtained by calculating a weighted sum of the risk word score, the urgency score, and the historical health risk score.

4. The method according to claim 3, characterized in that, The determination of the recommended service area, based at least on the service type tag and the urgency score, includes: Based on the service type label and the urgency score, an initial service circle is determined in a preset two-dimensional mapping table, wherein the preset two-dimensional mapping table is used to represent the correspondence between the service type label, different urgency score ranges and each service circle; Based on the historical service information and health records of the service recipient, it is determined whether the service recipient meets the correction conditions. The correction conditions include a first correction condition and a second correction condition. The first correction condition is that the historical service information shows that the service recipient has completed services in service circles outside the initial service circle more than a preset number of times in the most recent preset number of similar requests. The second correction condition is that the service recipient's health record has a high-risk chronic disease marker related to the service type, and the difference between the urgency score and any urgency score interval threshold is less than a preset difference threshold. When any correction condition is met, the initial service circle is corrected according to the satisfied correction condition to determine the recommended service circle; Preferably, the step of correcting the initial service circle according to the satisfied correction conditions includes: When only the first correction condition is met, the initial service circle is corrected by the first correction operation, and the corrected initial service circle is determined as the recommended service circle. When only the second correction condition is met, the initial service circle is corrected by the second correction operation, and the corrected initial service circle is determined as the recommended service circle. When both the first correction condition and the second correction condition are met, the initial service circle is corrected by the first correction operation and the second correction operation respectively to obtain the first correction result and the second correction result; the service circle that is outermost between the first correction result and the second correction result is selected as the recommended service circle. The first correction operation includes: adjusting the current initial service circle to the next service circle; The second correction operation includes: compensating the urgency score according to a preset risk compensation value, and re-determining the initial service area in the preset two-dimensional mapping table based on the compensated urgency score; Preferably, the method further includes: when the initial service circle is the third service circle, if the load of each service entity in the third service circle is saturated, generating a horizontal coordination request and sending it to the administrator node; upon receiving the confirmation information from the administrator node, obtaining the load of service entities in the third service circles of the adjacent service areas of the current service area; if the load of service entities in the third service circles of the adjacent service areas is not saturated, pushing the service request notification to each service entity in the third service circle of the adjacent service areas; Preferably, the step of correcting the initial service circle according to the satisfied correction conditions further includes: When the initial service circle is the third service circle and the corrected service circle is the fourth service circle, an upgrade service circle request is generated and sent to the administrator node; upon receiving confirmation information from the administrator node, the fourth service circle is determined to be the recommended service circle.

5. The method according to claim 1, characterized in that, The step of pushing service request notifications to each service entity within the recommended service circle includes: For each service entity within the recommended service circle, determine the suitability score for that service entity; The service entities within the recommended service circle are sorted from largest to smallest according to the suitability score to obtain the push sequence; According to the push sequence, the service request notification is pushed to each service entity in batches with a preset push step size; If a confirmation message is received from a service subject in the current push batch within the preset waiting time, the push will stop; otherwise, the service request notification will be pushed to the next batch of service subjects. Preferably, the suitability score is determined by at least the sum of a spatial distance score, a historical service quality score, an order load score, and a skill matching score; wherein, the spatial distance score is determined based on the spatial distance between the service provider and the service recipient; the historical service quality score is determined based on the service provider's historical service information, including historical service completion rate, average response time, and service feedback score; the order load score is determined based on the number of service tasks that the service provider has currently accepted but not completed, and the order load score is negatively correlated with the number of service tasks; the skill matching score is determined based on the degree of matching between the service provider's skill tags and the service type tags of the service demand; Preferably, the step of pushing the service request notification to each service entity within the recommended service circle further includes: Service entities in the recommended service circle whose affiliation probability is greater than the second probability threshold are marked as high-confidence entities, and other service entities are marked as entities to be verified. Specifically, for any high-confidence entity in the recommended service circle, the suitability score of the service entity is the sum of the spatial distance score, historical service quality score, order load score, and skill matching score; for any entity to be verified in the recommended service circle, the suitability score of the service entity is the product of the sum of the spatial distance score, historical service quality score, order load score, and skill matching score and the penalty coefficient, wherein the penalty coefficient is less than 1.

6. The method according to claim 5, characterized in that, The method further includes: When none of the service providers in the recommended service circle have accepted the order, the service request notification is pushed to the next service circle in the recommended service circle. Preferably, the method further includes: When the total response time of the recommended service circle reaches the service circle timeout threshold, it is determined that none of the service entities in the recommended service circle have accepted the order. The service circle timeout threshold is a preset fixed value or a dynamic value. Preferably, the service circle timeout threshold is a dynamic value, and the service circle timeout threshold is determined in the following way: Based on the urgency score of the service request, the time period in which the service request was initiated, and the historical average response time of the recommended service circle, the service circle timeout threshold is calculated. Preferably, calculating the service circle timeout threshold includes: Based on the aforementioned urgency score, an urgency factor is determined; Determine the time factor based on the time period in which the service request was initiated; Based on the historical average response time, determine the historical response factor; Calculate the product of the preset base duration and the urgency factor, time factor, and historical response factor to obtain the service circle timeout threshold; Preferably, calculating the service circle timeout threshold further includes: Determine whether the service timeout threshold is within a preset timeout range; When the service circle timeout threshold is not within the preset timeout interval, the service circle timeout threshold is adjusted to the nearest value in the preset timeout interval to the service circle timeout threshold. Preferably, the preset waiting time is the ratio of the service circle timeout threshold to the total number of service subjects in the recommended service circle.

7. The method according to claim 1, characterized in that, The method further includes: After the preset lock time for the first receipt of the confirmation information, the service task corresponding to the service request is locked; When multiple confirmation messages exist within the preset lock time period, the confirmation messages are sorted according to their reception time and in chronological order. According to the sorting results, the suitability scores of the service entities corresponding to each confirmation information are traversed sequentially. When the suitability score of the currently traversed service entity is higher than the minimum usable threshold, the current service entity is determined as the target service entity.

8. The method according to claim 3, characterized in that, The method further includes: upon receiving a risk escalation request from the target service subject, adjusting the urgency score based on the risk escalation request, and re-executing the step of determining the recommended service circle based at least on the service type label and the urgency score based on the adjusted urgency score; And / or, The method further includes: after receiving correction information from the administrator, modifying the service requirements according to the correction information, and re-executing the step of determining the recommended service circle in each of the service circles based at least on the service requirements based on the modified service requirements.

9. A mutual-aid elderly care scheduling and management system based on resident co-governance, characterized in that, The system is used to implement the method as described in any one of claims 1-8.

10. A computer-readable storage medium, characterized in that, The system stores a computer program / instructions that, when executed by a processor, implement the method as described in any one of claims 1-8.

Images