Data processing method and system for resource scheduling

Abstract

The present application relates to the technical field of data processing, in particular to a resource scheduling-oriented data processing method and system. The resource scheduling-oriented data processing method comprises obtaining initial multi-source feature data of a target object to construct a dynamic feature portrait, and constructing a resource capability label of a candidate service resource; when the target object interacts with a current service resource, real-time collection of interaction data sequences and dynamic updating of the portrait; real-time calculation and monitoring of the dynamic fit degree of the interaction link based on the updated portrait and the current resource capability label; when the dynamic fit degree meets a preset scheduling trigger condition, the latest portrait is extracted; an intelligent scheduling decision algorithm is called, combined with the capability label and real-time working state parameters of the remaining service resources, to screen out a target replacement resource; finally, context data is synchronized to the replacement resource to take over the interaction link. The present application realizes dynamic real-time monitoring of the interaction process and intelligent and accurate matching of resources.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and specifically to a data processing method and system for resource scheduling. Background Technology

[0002] In customer service resource allocation scenarios, especially in highly interactive business areas such as sales and consulting, how to efficiently and accurately match target objects (or users, target customers, etc.) to suitable service resources (such as sales personnel, customer service specialists, etc.) has always been a key issue in improving service conversion efficiency and customer satisfaction.

[0003] Traditional resource scheduling methods often employ static rule-driven allocation mechanisms, such as regional rotation, random assignment, or simple attribute matching (e.g., industry, language), to allocate customer leads to fixed service personnel in a one-time manner. These methods lack in-depth modeling and dynamic adaptation capabilities to individual customer differences and service resource capabilities.

[0004] In recent years, with the development of artificial intelligence and big data technologies, some solutions have attempted to introduce user profiles and resource tagging systems to optimize matching logic.

[0005] However, the aforementioned existing technologies still have the following significant limitations.

[0006] Customer profiling generally relies on initial static information or limited historical behavior, failing to effectively integrate multi-source heterogeneous data such as one-way public domain trajectories and two-way private domain interactive texts. In particular, it lacks structured modeling of implicit psychological dimensions such as decision-making autonomy, risk preference, and information processing style, resulting in insufficient profiling representation capabilities and difficulty in supporting refined matching.

[0007] Existing scheduling mechanisms mostly focus on the initial allocation stage and lack the ability to monitor the quality of established interaction links in real time and proactively intervene. For example, when customers interact with current service resources inefficiently or experience negative emotional fluctuations, existing technologies cannot dynamically assess the fit based on objective data and trigger rescheduling, resulting in missed opportunities for recovery and increased risk of customer churn.

[0008] Even if some solutions introduce the concept of "dynamic", their matching logic is still mostly based on overall similarity calculation. They do not combine contextual factors such as the topic stage of the interaction and the trend of customer intent decay to conduct a weighted evaluation of the fit, nor do they establish an interpretable scheduling trigger threshold mechanism (such as multimodal triggering conditions such as behavioral mutation and instruction interruption).

[0009] When it is necessary to switch service resources, existing systems generally lack a scientific mechanism for selecting replacement resources: they do not comprehensively consider the real-time working status of candidate resources (such as load and emotional energy) and historical collaboration success rate, nor do they achieve structured synchronization and seamless handover of context data, which can easily lead to fragmented customer experience, collaboration conflicts and disputes over performance attribution. Summary of the Invention

[0010] In view of the shortcomings of the prior art described above, the purpose of this invention is to provide a data processing method and system for resource scheduling, which at least solves one of the technical problems in the prior art.

[0011] The first aspect of this invention provides a data processing method for resource scheduling, comprising: Acquire initial multi-source feature data of the target object, and construct a dynamic feature profile of the target object through a preset feature extraction model; Obtain historical execution data and attribute characteristics of each service resource in the candidate resource set, and construct resource capability tags for each service resource through preset tag generation rules; In response to assigning the target object to a service resource in the candidate resource set as the current service resource, during the process of establishing an interaction link between the current service resource and the target object, the interaction data sequence is collected in real time. The dynamic feature profile of the target object is dynamically updated according to the interaction data sequence, and the dynamic fit of the interaction link is calculated and monitored in real time based on the updated dynamic feature profile and the resource capability tag of the current service resource; when the dynamic fit meets the preset scheduling trigger condition, the latest dynamic feature profile of the target object is extracted. The preset intelligent scheduling decision algorithm is invoked to calculate the matching degree based on the latest dynamic feature profile, the resource capability tags of the remaining service resources in the candidate resource set excluding the current service resource, and the real-time working status parameters of each remaining service resource, and to select the target replacement resource from the remaining service resources. The context data of the interaction link is synchronized to the target replacement resource to control the target replacement resource to take over the interaction link with the target object.

[0012] As an optional implementation, the step of acquiring initial multi-source feature data of the target object and constructing a dynamic feature profile of the target object through a preset feature extraction model includes: Acquire the one-way behavior trajectory data of the target object generated in the public domain data source, as well as the initial two-way communication text stream generated in the internal business terminal; The feature extraction model is used to jointly extract features from the unidirectional behavioral trajectory data and the initial bidirectional communication text stream to generate initial object feature vectors of the target object in the dimensions of decision autonomy, information processing style, risk preference, and social orientation, so as to construct the dynamic feature profile.

[0013] As an optional implementation, the step of obtaining historical execution data and attribute characteristics of each service resource in the candidate resource set, and constructing resource capability tags for each service resource through preset tag generation rules, includes: Deeply mine the historical execution data of each service resource in the candidate resource set to extract explicit professional skill parameters and implicit trait parameters; The explicit professional skill parameters and the implicit trait parameters are structurally integrated to construct resource capability tags for each service resource, including communication mode dimension, professional field dimension, trait orientation dimension, and success case mapping dimension.

[0014] As an optional implementation, dynamically updating the dynamic feature profile of the target object based on the interaction data sequence includes: Natural language processing and sentiment analysis are performed on the interaction data sequence to extract the instantaneous feedback features of the target object and the intent decay features of the interaction intent within the current interaction cycle. A preset time-series decay function is introduced, and the weights of each dimension in the original dynamic feature profile of the target object are reduced or increased using the real-time feedback feature and the intent decay feature to generate the updated dynamic feature profile.

[0015] As an optional implementation, the step of calculating and monitoring the dynamic fit of the interaction link in real time based on the updated dynamic feature profile and the resource capability tags of the current service resources includes: The updated dynamic feature profile is converted into a current object feature vector, and the resource capability tags of the current service resource are converted into a current resource feature vector. Based on the vector space distance between the current object feature vector and the current resource feature vector in each corresponding dimension, calculate the sub-fit score for each dimension; Based on the topic stage of the interaction data sequence, dynamic weights are assigned to each sub-fit score, and a weighted summation algorithm is used to generate the dynamic fit that represents the state of the interaction link.

[0016] As an optional implementation, when the dynamic fit meets the preset scheduling triggering conditions, one or more of the following triggering logics are activated: Active warning trigger: When the quantized value of the dynamic fit is continuously lower than the first preset threshold and the duration exceeds the second preset threshold within a time window; Behavioral mutation trigger: Through semantic-level analysis of the interaction data sequence, it is detected that the fluctuation variance representing the negative emotions of the target object exceeds a preset abnormal mutation threshold; Command interruption trigger: Receives a collaborative support request command initiated by the current service resource.

[0017] As an optional implementation, the real-time calculation and monitoring of the dynamic compatibility of the interaction link, and the determination of whether the preset scheduling triggering conditions are met, are based on a preset matching strategy matrix. The matching strategy matrix includes at least the first to fourth quadrant regions constructed based on the target object dimension features and the service resource dimension features, and the matching strategy matrix is ​​divided into the first to fourth quadrant regions with the target object decision autonomy as the horizontal axis and the service resource dominance as the vertical axis. For the mapping point position of the updated dynamic feature profile and the resource capability label of the current service resource in the matching strategy matrix, execute adaptive control logic: If the mapping point is located in the first quadrant region, it is determined that the scheduling triggering condition is not met, and the interaction link is maintained; If the mapping point is located in the second quadrant region, it is determined to be in a risk warning state, and service resources with buffer feature labels are marked as candidate queues in the remaining service resources; If the mapping point is located in the third quadrant region, it is determined that the scheduling triggering condition is met, and the intelligent scheduling decision algorithm is immediately invoked to perform the screening of the target replacement resource; If the mapping point is located in the fourth quadrant region, an interaction strategy adjustment instruction is issued to the service node corresponding to the current service resource.

[0018] As an optional implementation, the step of calculating the matching degree based on the latest dynamic feature profile, the resource capability tags of the remaining service resources, and the real-time working status parameters of each of the remaining service resources, and selecting target replacement resources from the remaining service resources, includes: Obtain the real-time working status parameters of each of the remaining service resources. The real-time working status parameters include at least the current task load, the preset online identifier, and the emotional energy assessment value extracted based on recent interaction corpus. Based on the current task load and the online identifier, a candidate replacement subset is generated by filtering from the remaining service resources; Calculate the capability matching score between the resource capability tags of each service resource in the candidate replacement subset and the latest dynamic feature profile, and dynamically correct the capability matching score using the emotion energy assessment value; Historical collaboration data of each service resource in the candidate replacement subset is retrieved to calculate the collaboration success rate score. The service resource with the highest comprehensive score is selected as the target replacement resource by weighted optimization based on the corrected capability matching score and the collaboration success rate score.

[0019] As an optional implementation, synchronizing the context data of the interaction link to the target replacement resource to control the target replacement resource to take over the interaction link with the target object includes: Extract the full historical interaction data sequence between the current service resource and the target object, extract key insight nodes, and extract pending task status data from the preset historical interaction database; The extracted data is packaged according to a preset structured template to generate a digital handover summary; The digital handover summary is pushed to the server-side communication node corresponding to the target replacement resource for rendering and display, and the subsequent data communication route of the target object is redirected to the target replacement resource.

[0020] As an optional implementation, after controlling the interaction link between the target replacement resource and the target object, the method further includes: Real-time tracking of subsequent interaction nodes of the target replacement resource; in response to monitoring that the target replacement resource has completed a preset conversion task, obtaining the total incentive amount data corresponding to the conversion task. Extract the initial link maintenance duration feature of the current service resource, and the takeover promotion node feature of the target replacement resource; The initial link maintenance duration feature and the takeover advancement node feature are input into a preset contribution splitting algorithm model to automatically generate and output collaborative contribution allocation matrix data for the current service resource and the target replacement resource.

[0021] A second aspect of the present invention provides a data processing system for resource scheduling, comprising: The profile building module is used to acquire initial multi-source feature data of the target object and construct a dynamic feature profile of the target object through a preset feature extraction model. The capability tag generation module is used to obtain the historical execution data and attribute characteristics of each service resource in the candidate resource set, and construct resource capability tags for each service resource through preset tag generation rules; An interactive data acquisition module is used to, in response to assigning the target object to a service resource in the candidate resource set as the current service resource, collect interactive data sequences in real time during the process of establishing an interactive link between the current service resource and the target object; The dynamic matching monitoring module is used to dynamically update the dynamic feature profile of the target object according to the interaction data sequence, and calculate and monitor the dynamic matching degree of the interaction link in real time based on the updated dynamic feature profile and the resource capability tag of the current service resource; when the dynamic matching degree meets the preset scheduling triggering condition, the latest dynamic feature profile of the target object is extracted. The intelligent scheduling decision engine module is used to call a preset intelligent scheduling decision algorithm to calculate the matching degree based on the latest dynamic feature profile, the resource capability tags of the remaining service resources in the candidate resource set excluding the current service resource, and the real-time working status parameters of each remaining service resource, and to select the target replacement resource from the remaining service resources. The link handover and takeover module is used to synchronize the context data of the interaction link to the target replacement resource, so as to control the target replacement resource to take over the interaction link with the target object.

[0022] A third aspect of the present invention provides an electronic device, comprising: at least one processor; and at least one memory communicatively connected to the processor, wherein: the memory stores program instructions executable by the processor, and the processor, by invoking the program instructions, can perform the steps of the method described in the first aspect of the present invention.

[0023] A fourth aspect of the present invention provides a readable storage medium storing a computer program that is executed by a processor as described in the first aspect of the present invention.

[0024] This invention has at least the following technical effects: 1. This invention constructs a dynamic feature profile of a target object by fusing multi-source feature data and deeply integrates multi-dimensional capability tags of service resources, completely abandoning the traditional rigid fixed allocation mode (such as simple polling or regional system). Relying on real-time acquisition of interaction data sequences, this solution overcomes the shortcomings of traditional systems that rely solely on static initial information, and can keenly capture the implicit trait changes and decision-making style transfers of the target object during the interaction process, achieving more in-depth temporal updates of feature profiles and continuous accurate matching; 2. This invention introduces a real-time calculation and monitoring model for dynamic fit. At the very early stage when the matching quality between the target object and the current service resources deteriorates, the system can sensitively detect the decay of fit and accurately trigger the preset scheduling conditions, thereby transforming passive post-event remediation into proactive pre-event intervention. This effectively prevents the spread of negative interactive experiences and significantly reduces the risk of eventual churn of the target object. 3. This invention, through a pre-defined intelligent scheduling decision-making algorithm, can automatically select the optimal target replacement resource from a large set of candidate resources by comprehensively considering capability tags and real-time working status parameters. This mechanism enables the vast majority of resource switching requests to be automatically triggered by the system and provides quantitative matching criteria, greatly reducing the workload of manual review by management personnel and the rate of subjective misjudgment. 4. This invention ensures the complete transmission of core data such as interaction history and insight nodes through a lossless synchronization mechanism of digital context data, guaranteeing a smooth and seamless experience for target users during resource switching. Simultaneously, combined with an optional underlying collaborative contribution allocation algorithm, it fundamentally resolves the interest conflicts during service resource handover, greatly reducing internal collaboration friction and comprehensively improving user satisfaction and operational efficiency of the overall service network. Attached Figure Description

[0025] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 This is a flowchart illustrating a data processing method for resource scheduling according to a specific embodiment of the present invention.

[0027] Figure 2 This is a schematic diagram of matrix evolution in a specific embodiment of the present invention.

[0028] Figure 3 This is a block diagram of a data processing system oriented towards resource scheduling in a specific embodiment of the present invention. Detailed Implementation

[0029] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application. Furthermore, it should be understood that the specific embodiments described herein are only for illustration and explanation of this application and are not intended to limit this application.

[0030] It should be noted that the order of description of the following embodiments is not intended to limit the preferred order of the embodiments of this application. Furthermore, the descriptions of each embodiment in the following embodiments have their own emphasis; for parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0031] like Figure 1 As shown, the first aspect of the present invention provides a data processing method for resource scheduling, including the following:

[0032] Step S100: Obtain initial multi-source feature data of the target object, and construct a dynamic feature profile of the target object through a preset feature extraction model; wherein, the initial multi-source feature data includes at least the basic information of the target object, the source channel and the type of consulting product, and the feature extraction model performs weighted fusion of the above data based on a rule engine to construct a dynamic feature profile of the target object.

[0033] Step S200: Obtain the historical execution data and attribute characteristics of each service resource in the candidate resource set, and construct resource capability tags for each service resource through preset tag generation rules; wherein, the historical execution data includes at least the past interaction success rate, average response time and customer satisfaction score, and the attribute characteristics include at least the professional knowledge level, the leading role, the patience level and the service awareness. The tag generation rules map the above data into structured tags to characterize the capability level of the service resources in different dimensions.

[0034] Step S300: In response to assigning the target object to a service resource in the candidate resource set as the current service resource, during the process of establishing an interaction link between the current service resource and the target object, an interaction data sequence is collected in real time; wherein, the interaction data sequence includes at least dialogue text, response interval, question type change, customer emotion feedback and micro-survey score, collected at a preset collection frequency, for example once every 30 seconds, and formed into an ordered sequence by aligning with timestamps.

[0035] Step S400: Dynamically update the dynamic feature profile of the target object according to the interaction data sequence, and calculate and monitor the dynamic fit of the interaction link in real time based on the updated dynamic feature profile and the resource capability tag of the current service resource; when the dynamic fit meets the preset scheduling trigger condition, extract the latest dynamic feature profile of the target object.

[0036] For example, dynamic fit is calculated using a matching matrix model, which analyzes the distance and quadrant relationship between customer profile coordinates and sales tag coordinates on a two-dimensional plane, outputs a fit score from 0 to 100, and continuously tracks its changing trend.

[0037] Preferably, the dynamic fit is calculated using a four-quadrant matching matrix, where the first quadrant is the ideal matching area, the second quadrant is the risk area, the third quadrant is the mismatch area, and the fourth quadrant is the potential area. The system determines the matching status and triggers the corresponding strategy based on the relative position of the target object and the current service resource in the matrix.

[0038] For example, the scheduling trigger conditions include at least: the dynamic fit score is lower than the threshold of 65 points and lasts for more than 48 hours, the customer's negative feedback score is higher than the threshold of N, the service resources actively request support, or a sudden change in the customer's behavior pattern is detected.

[0039] Step S500: Invoke the preset intelligent scheduling decision algorithm, and calculate the matching degree based on the latest dynamic feature profile, the resource capability tags of the remaining service resources in the candidate resource set excluding the current service resource, and the real-time working status parameters of each remaining service resource, and select target replacement resources from the remaining service resources; wherein, the real-time working status parameters include the current load rate, emotional energy value and idle status indicator, and the matching degree calculation comprehensively considers the ability matching degree, historical collaboration success rate, professional field fit and status adaptability; the process of selecting target replacement resources includes: outputting a recommendation list and a matching reason report.

[0040] Preferably, the intelligent scheduling decision algorithm supports a collaborative takeover mode. During the takeover of target replacement resources, the original service resources are allowed to participate in subsequent interactions as auxiliary roles. The system automatically splits performance indicators based on the contributions of the two in the post-switch phase.

[0041] Step S600: Synchronize the context data of the interaction link to the target replacement resource to control the target replacement resource to take over the interaction link with the target object; wherein: the context data includes complete dialogue records, a list of answered questions, the evolution trajectory of customer focus and to-do items, the synchronization process is completed within a preset time after the switch command is confirmed, and a handover notification is automatically sent to both parties to ensure the continuity of interaction.

[0042] This invention achieves better matching and scheduling by using dynamic feature profiles of target objects and service resource capability tags, abandoning traditional scheduling modes that rely on single static rules such as regional allocation and random rotation. This significantly improves the accuracy and adaptability of resource matching. By calculating and monitoring the dynamic fit of the interaction link in real time, it can trigger scheduling warnings and replacement processes immediately when abnormal states such as fit decay or sudden changes in target object emotions occur, enabling proactive intervention in abnormal states. Compared to traditional post-event remediation methods, this effectively reduces the risk of target object churn. The intelligent scheduling decision algorithm selects and determines target replacement resources, eliminating the arbitrariness of manual subjective handover and reducing collaboration conflicts between service resources. By synchronizing and seamlessly taking over the context data of the interaction link, it achieves seamless service switching for target objects, avoiding information gaps and redundant communication. Ultimately, this improves overall resource scheduling efficiency, service conversion success rate, and target object service satisfaction.

[0043] In step S100, acquiring the initial multi-source feature data of the target object and constructing a dynamic feature profile of the target object through a preset feature extraction model includes: Step S110: Obtain the one-way behavior trajectory data of the target object generated in the public domain data source, and the initial two-way communication text stream generated in the internal business terminal; Step S120: Use the feature extraction model to perform joint feature extraction on the unidirectional behavior trajectory data and the initial bidirectional communication text stream to generate the initial object feature vector of the target object in the dimensions of decision autonomy, information processing style, risk preference and social orientation, so as to construct the dynamic feature profile.

[0044] Specifically, the one-way behavioral trajectory data includes the duration of customer browsing product detail pages online, click behavior sequences, page jump paths, and the time of inquiry initiation; the initial two-way communication text stream includes dialogue records exchanged between customers and sales personnel via telephone, instant messaging tools, or email; the decision-making autonomy dimension is quantified by measuring whether customers proactively raise clear needs and whether they quickly skip guiding content; the information processing style dimension is judged based on whether customers prefer to receive structured parameters or scenario-based cases; the risk preference dimension is scored based on their statements regarding the certainty of financial solutions, the flexibility of delivery cycles, and their tolerance for product trials; the social orientation dimension is quantified by combining the target audience's search and click frequency on third-party evaluations (such as car owner forum reviews and sales rankings) and the proportion of mentions of group identity tags or other people's opinions (such as family preferences and friend recommendations) in the communication text. Preferably, the dimensions also include a brand preference dimension and a sensitive pain point dimension. The brand preference dimension is quantified based on the frequency of customer mentions of the brand and the degree of positive evaluation, while the sensitive pain point dimension is quantified by extracting keywords related to concerns and anxieties in customer statements.

[0045] For example, the preset feature extraction model is a multimodal fusion architecture model. In the extraction process, the behavioral trajectory data is first processed by a temporal encoder to capture the decision rhythm and exploration depth. The question type, emotional tendency and information inquiry pattern in the text stream are parsed by the semantic understanding module. Then, the two types of feature vectors are aligned and weighted in the shared embedding space.

[0046] Here, the numerical values ​​of the above dimensions are combined to form a dimension vector, which serves as the initial state of the dynamic feature profile and is continuously used for subsequent matching evaluation and switching decisions. This dynamic feature profile can accurately reflect the customer's current behavioral tendencies. For example, when a customer shifts from frequently asking about technical details to focusing on current vehicle inventory and financial incentives, their decision autonomy score increases and their information processing style shifts from deep analysis to efficiency orientation, thereby triggering a dynamic reallocation of sales resources.

[0047] In step S200, obtaining the historical execution data and attribute characteristics of each service resource in the candidate resource set, and constructing resource capability tags for each service resource through preset tag generation rules, includes: Step S210: Perform in-depth mining on the historical execution data of each service resource in the candidate resource set to extract explicit professional skill parameters and implicit trait parameters; Step S220: The explicit professional skill parameters and the implicit trait parameters are structurally integrated to construct resource capability tags for each service resource, including communication mode dimension, professional field dimension, trait orientation dimension and success case mapping dimension.

[0048] Specifically, the process of in-depth mining of the historical execution data of each service resource in the candidate resource set is as follows: extracting structured and unstructured data from customer service interaction logs, call recordings, chat logs, task completion records, and customer feedback evaluations; using natural language processing technology to identify explicit professional skill parameters, including quantifiable indicators such as professional knowledge coverage areas, product (line) familiarity scores, technical problem-solving accuracy, financial solution configuration capabilities, and industry experience levels; and mining implicit trait parameters through behavioral sequence modeling and sentiment analysis algorithms, including communication rhythm preferences, emotional stability index, dominant driving tendency, patience level, frequency of empathy expression, and sensitivity to negative customer emotions.

[0049] Specifically, the structured fusion process of the explicit professional skill parameters and the implicit trait parameters is as follows: a unified representation framework is constructed based on a multidimensional feature vector space, the explicit parameters are mapped to the professional ability coordinate axis, the implicit parameters are mapped to the behavioral trait coordinate axis, and the dimensionality difference is eliminated through weighted normalization; principal component analysis or factor analysis is used to reduce the dimensionality of the high-dimensional parameters and integrate them to form a label structure with four core dimensions.

[0050] Specifically, the four core dimensions include the communication style dimension, the professional domain dimension, the trait-oriented dimension, and the success case mapping dimension. The communication style dimension reflects whether the service resource tends to lead or support services in interactions; the professional domain dimension covers the product lines, technical directions, and solution types that the service provider is proficient in; the trait-oriented dimension describes the service provider's emotional regulation ability, customer-oriented awareness, and willingness to collaborate under pressure; and the success case mapping dimension relates to specific customer scenarios with high satisfaction and high conversion rates in the service provider's history, including contextual characteristics such as the customer's level of decision-making autonomy, the intensity of information needs, and risk sensitivity.

[0051] Taking sales resources as an example, the communication mode dimension is numerically represented as a leading driving force score of 65 points (out of 100). The professional field dimension is labeled as "new energy vehicle technology expert + financial solution configuration expert". The trait-oriented dimension includes "high empathy (82 points), moderate patience (58 points), and low emotional fluctuation (standard deviation <0.3)". The success case mapping dimension is associated with three typical transaction records, corresponding to scenarios such as "conversion of technical novice customers after guidance", "efficiency-driven customers making quick decisions and closing deals", and "high-risk averse customers dispelling doubts through community case studies". This label dynamic update mechanism performs incremental training every 24 hours based on the latest interaction data to ensure the timeliness and accuracy of the labels.

[0052] Here, the resource capability tagging construction method can improve the accuracy of compatibility prediction in subsequent human-machine matching or interpersonal collaboration, and effectively support the accurate assessment and recommendation of the successor's capabilities in dynamic switching decisions.

[0053] In one embodiment of the present invention, dynamically updating the dynamic feature profile of the target object based on the interaction data sequence includes: Natural language processing and sentiment analysis are performed on the interaction data sequence to extract the instant feedback features and intent decay features of the target object in the current interaction cycle; wherein: the instant feedback features include the semantic tendency, emotional polarity score and keyword frequency changes expressed by the customer in the dialogue, and the intent decay features reflect the degree of weakening of the customer's initial consultation intent over time, such as the behavioral migration pattern from asking about technical parameters to focusing on transaction efficiency. A preset time-series decay function is introduced, and the weights of each dimension in the original dynamic feature profile of the target object are reduced or increased using the real-time feedback feature and the intent decay feature to generate the updated dynamic feature profile.

[0054] Specifically, the preset time decay function takes the interaction time interval and interaction frequency as input variables and outputs the corresponding time decay coefficient; after weighting the instant feedback feature and intent decay feature respectively, and combining them with the time decay coefficient, the weights of each dimension in the original dynamic feature profile of the target object are adjusted; if the customer reduces technical questions and increases transaction-oriented questions such as financial solutions or vehicle availability in subsequent interactions, the weight of the "information demand" dimension is reduced, while the weight of the "decision efficiency preference" dimension is increased; if the system detects that the negative feedback score exceeds the threshold and the duration meets the condition, the decay rate of the relevant intent dimension is accelerated.

[0055] The comprehensive vector representation of the dynamic feature profile is recalculated based on the adjusted weights of each dimension, generating an updated dynamic feature profile. This updated profile is used to assess the matching status between the customer and the current service personnel in real time and serves as the input for switching decisions.

[0056] In practical applications, when customer profiling evolves from a highly information-dependent type to a highly autonomous decision-making type, the updated profile identifies that the original salesperson's technical explanation style is no longer a good match, thereby triggering a successor recommendation process to ensure that the service strategy remains in sync with the customer's current intentions.

[0057] In one embodiment of the present invention, the step of calculating and monitoring the dynamic fit of the interaction link in real time based on the updated dynamic feature profile and the resource capability tags of the current service resource includes: The updated dynamic feature profile is converted into a current object feature vector, and the resource capability tags of the current service resource are converted into a current resource feature vector. Based on the vector space distance between the current object feature vector and the current resource feature vector in each corresponding dimension, calculate the sub-fit score for each dimension; Based on the topic stage of the interaction data sequence, dynamic weights are assigned to each sub-fit score, and a weighted summation algorithm is used to generate the dynamic fit that represents the state of the interaction link.

[0058] Furthermore, this invention employs a dual-layer triggering mechanism combining active and passive triggering. When the dynamic compatibility meets the preset scheduling triggering conditions, any one or more of the following triggering logics are activated: Active warning trigger: When the quantized value of the dynamic fit is continuously lower than the first preset threshold and the duration exceeds the second preset threshold within a time window; Behavioral mutation trigger: Through semantic-level analysis of the interaction data sequence, it is detected that the fluctuation variance representing the negative emotions of the target object exceeds a preset abnormal mutation threshold; Command interruption trigger: Receives a collaborative support request command initiated by the current service resource.

[0059] Among them, proactive warning triggering and behavioral change triggering are automatically and proactively triggered by the system, while instruction interruption triggering is a passive triggering initiated by service resources.

[0060] Furthermore, the real-time calculation and monitoring of the dynamic compatibility of the interaction link, and the determination of whether the preset scheduling triggering conditions are met, are based on a preset matching strategy matrix. The matching strategy matrix includes at least the first to fourth quadrant regions constructed based on the target object dimension features and the service resource dimension features. The matching strategy matrix is ​​divided into the first to fourth quadrant regions with the target object's decision autonomy as the horizontal axis and the service resource's dominance as the vertical axis. The features of the current object and the current service resource are mapped to coordinate points that fall into the matrix. For the mapping point position of the updated dynamic feature profile and the resource capability label of the current service resource in the matching strategy matrix, execute adaptive control logic: If the mapping point is located in the first quadrant region, it is determined that the scheduling triggering condition is not met, and the interaction link is maintained; If the mapping point is located in the second quadrant region, it is determined to be in a risk warning state, and service resources with buffer feature labels are marked as candidate queues in the remaining service resources; If the mapping point is located in the third quadrant region, it is determined that the scheduling triggering condition is met, and the intelligent scheduling decision algorithm is immediately invoked to perform the screening of the target replacement resource; If the mapping point is located in the fourth quadrant region, an interaction strategy adjustment instruction is issued to the service node corresponding to the current service resource.

[0061] To further explain the underlying decision logic of the "matching strategy matrix" in the embodiments of the present invention, please refer to [link / reference]. Figure 2 , Figure 2 This diagram illustrates the matching strategy matrix architecture of target objects and service resources in a multi-dimensional feature space, as shown in an embodiment of the present invention.

[0062] In this embodiment, a two-dimensional or multi-dimensional feature vector mapping space is constructed.

[0063] like Figure 2As shown, taking a two-dimensional space as an example, the horizontal axis (X-axis) of the matching strategy matrix represents specific dimensional feature parameters of the target object (e.g., decision-making autonomy, with values ​​normalized to 0 to 1; lower values ​​represent dependency on consultation, and higher values ​​represent autonomous decision-making); the vertical axis (Y-axis) represents specific dimensional resource capability label parameters of the service resources (e.g., dominant driving force, with lower values ​​representing auxiliary services and higher values ​​representing dominant driving force). Through the intersection of these two dimensions, the mapping space is divided into four preset quadrant regions.

[0064] Specifically, based on the mapping point location of the dynamic feature profile of the target object and the resource capability label of the current service resource in the matrix, the following adaptive control and calculation logic is executed.

[0065] The first quadrant region (upper right, ideal matching region): When the coordinates of the mapped points satisfy the characteristics of high autonomous decision-making and high dominant driving force (for example, the target object is efficiency-driven and the service resource is transaction-facilitating), the Euclidean distance between the two in the feature space is calculated to be within the first preset safety range. At this time, the matching state is determined to be highly coordinated, and the control logic of "maintaining the interaction link" is executed without external intervention, only maintaining continuous data collection and monitoring in the background.

[0066] The second quadrant (top left, risk warning zone): When the mapped point coordinates satisfy the characteristics of low autonomous decision-making (such as a technical novice who relies on guidance) and high dominant driving characteristics (such as an aggressive sales expert), the probability of potential conflict is calculated to be higher. At this time, the risk warning logic is triggered, and the remaining service resource set is automatically scanned and marked with buffer characteristic labels (such as high patience value, mild auxiliary type label) as a candidate queue in preparation for possible subsequent link switching.

[0067] The third quadrant (lower left, mismatch zone): When the mapped point coordinates satisfy the characteristics of low autonomous decision-making and low dominant driving force (such as hesitant objects and passively waiting resources), it is determined that the cooperative resistance of their feature vectors exceeds the preset extreme value, which is a serious deviation state. At this time, it is immediately determined that the scheduling trigger condition is met, and the intelligent scheduling decision algorithm is invoked to forcibly trigger the selection and context synchronization process of the target replacement resource (such as resources with high dominant driving force labels).

[0068] The fourth quadrant (lower right, potential adjustment zone): When the mapped point coordinates satisfy the characteristics of high autonomous decision-making and low dominant driving force, it is determined that the current state is partially matched. At this time, an "interaction strategy adjustment instruction" is issued to the service node corresponding to the current service resource (for example, prompting it through the terminal UI interface to speed up the response frequency, reduce redundant emotional interactions, and increase direct data-driven feedback) to guide the dynamic label coordinates of the current service resource to evolve towards the first quadrant.

[0069] In the specific data processing process, the intelligent scheduling decision engine encapsulates a key rule engine based on the IF-THEN architecture, and its quantitative judgment rules are exemplified below.

[0070] Rule 1 (Ideal Match Maintenance): If [Target Object. Decision Autonomy > 70] and [Target Object. Information Demand < 50], while [Service Resource. Dominance and Driving Force > 60] and [Service Resource. Professional Knowledge Score > 70], then output the status flag = "Ideal Match", and maintain the link status.

[0071] Rule 2 (Mismatch Blocking): If [Target Object.Decision Autonomy < 30] and [Target Object.Guidance Demand > 80], while [Service Resource.Dominance Driven < 30] and [Service Resource.Proactive Parameter < 40], then the output status flag = "Severe Mismatch" is triggered, and a scheduling interruption command is executed.

[0072] Example 1 (Triggering Mechanism Based on Dynamic Spatial Distance): Initially, the target object (e.g., a technical novice) has coordinates (0.2, 0.3) in the matrix, while the current service resource (e.g., a highly skilled expert) has coordinates (0.8, 0.7). Both fall into the second quadrant risk zone and are spatially far apart. During the evolution of the interactive data sequence, the target object smoothly moves its feature coordinates towards (0.6, 0.6) by acquiring information (self-learning evolution), but the coordinates of the service resource do not adaptively change. The relative distance or compatibility of the dynamic coordinates of the two is continuously calculated. When the compatibility score is detected to continuously decrease with dynamic evolution and fall below a preset threshold, the current link is precisely blocked, and a replacement resource closest to coordinates (0.6, 0.6) is rematched in the matrix (e.g., a technical explanation resource with coordinates 0.45, 0.60), thereby achieving a significant leap in dynamic matching.

[0073] Example 2 (Model Self-Correction Based on Policy Adjustment Instructions): Initially, the target object's coordinates are (0.8, 0.8), and the current service resource's coordinates are (0.3, 0.3), falling into the fourth quadrant potential area. Instead of executing a switching instruction, a feature correction instruction is issued (e.g., suggesting reducing excessive service or providing objective data support). Upon receiving the instruction, the current service resource changes its interaction mode and re-evaluates by collecting the latest communication text stream in real time. It is found that the service resource's mapped coordinates evolve towards (0.6, 0.5). As the spatial distance between the two coordinates decreases, the dynamic fit rises above the safety threshold, thus lifting the warning state and forming a closed-loop self-correction between the model and its behavior.

[0074] In step S500, the matching degree calculation based on the latest dynamic feature profile, the resource capability tags of the remaining service resources, and the real-time working status parameters of each remaining service resource, and the selection of target replacement resources from the remaining service resources, includes: Step S510: Obtain the real-time working status parameters of each remaining service resource. The real-time working status parameters include at least the current task load, the preset online identifier, and the emotional energy assessment value extracted based on recent interaction corpus. The current task load is calculated by statistically analyzing the number of customer sessions currently being processed by each service resource and the average response latency. The online identifier is determined by the service resource's proactive reporting or the system's heartbeat detection mechanism. The emotional energy assessment value is generated by performing natural language processing on the service resource's interaction text with customers in the past 24 hours, identifying the density of positive or negative emotional words, and weighting the values. The value ranges from 0 to 100, with higher values ​​indicating a more positive and stable emotional state. Step S520: Based on the current task load and the online identifier, filter and generate a candidate replacement subset from the remaining service resources; wherein, service resources whose current task load exceeds a preset load threshold such as 80% or whose online identifier is offline are excluded, and service resources whose load is lower than the preset load threshold of 80% and whose online identifier is valid are retained to form the candidate replacement subset.

[0075] Step S530: Calculate the capability matching score between the resource capability tags of each service resource in the candidate replacement subset and the latest dynamic feature profile, and dynamically correct the capability matching score using the emotional energy assessment value; wherein, the resource capability tags include dimensions such as professional field, communication style, and transaction driving force, and the latest dynamic feature profile is generated in real time from the target object's historical behavior data, current consultation intention, and decision-making style; map the resource capability tags of each service resource and the latest dynamic feature profile of the target object to the same multi-dimensional feature space to form standardized feature vectors; use the cosine similarity algorithm to quantify the spatial matching degree of the two sets of feature vectors to obtain the basic matching degree result representing the feature fit degree, and normalize it to an initial matching score within a set interval; based on the above initial matching score, combine the emotional energy assessment value of the service resource to complete the dynamic correction: use the emotional energy assessment value as an adjustment factor for the matching score, when the assessment value is lower than a preset threshold, the initial matching score is decayed and corrected according to a preset ratio rule; when the assessment value is not lower than the preset threshold, the initial matching score remains unchanged, and finally the corrected capability matching score is obtained; Step S540: Retrieve historical collaboration data of each service resource in the candidate replacement subset to calculate the collaboration success rate score. Perform weighted optimization based on the corrected capability matching score and the collaboration success rate score, and determine the service resource with the highest comprehensive score as the target replacement resource.

[0076] For example, the collaboration success rate score is based on the success conversion rate statistics of the service resource in collaborating with other service resources to handle similar customer tasks in the past 30 days. The weighted optimization adopts a linear weighted model, with the weight allocation being 70% for the adjusted capability matching score and 30% for the collaboration success rate score. The one with the highest comprehensive score is selected as the target replacement resource.

[0077] Furthermore, synchronizing the context data of the interaction link to the target replacement resource to control the target replacement resource to take over the interaction link with the target object includes: Extract the full historical interaction data sequence between the current service resource and the target object, extract key insight nodes, and extract pending task status data from the preset historical interaction database; The extracted data is packaged according to a preset structured template to generate a digital handover summary; The digital handover summary is pushed to the service-side communication node corresponding to the target replacement resource for rendering and display, and the subsequent data communication route of the target object is redirected to the target replacement resource; and a handover cooperation protocol is generated synchronously and synchronized to the relevant service resource terminal.

[0078] Specifically, the system extracts the full historical interaction data sequence between the current service resources and the target object, key insight nodes, and pending task status data from the preset historical interaction database. The full historical interaction data sequence includes all text, speech transcription, and operation logs. The key insight nodes consist of customer interest turning points, objection points, and commitment intention points automatically marked by the system. The pending task status data covers financial plans to be confirmed, test drive appointments not completed, and technical questions to be answered.

[0079] Specifically, the extracted data is packaged according to a preset structured template to generate a digital handover summary. The structured template includes a customer basic information area, an interaction timeline, a list of key concerns, risk warning prompts, and a to-do list to ensure that the information is complete and easy to understand quickly.

[0080] Specifically, the digital handover summary is pushed to the service-side communication node corresponding to the target replacement resource for rendering and display, and the subsequent data communication route of the target object is redirected to the target replacement resource. The rendering and display is presented in a card-style interface, highlighting key insights and to-do items. The route redirection is achieved by updating the session binding relationship, ensuring that the next message sent by the target object is directly received and processed by the target replacement resource.

[0081] Furthermore, after controlling the interaction link between the target replacement resource and the target object, the method further includes: Step S710: Track the subsequent interaction nodes of the target replacement resource in real time, and in response to the detection that the target replacement resource has completed the preset conversion task, obtain the total incentive amount data corresponding to the conversion task; Step S720: Extract the initial link maintenance duration feature of the current service resource and the takeover promotion node feature of the target replacement resource; Step S730: Input the initial link maintenance duration feature and the takeover advancement node feature into a preset contribution splitting algorithm model, and automatically generate and output the collaborative contribution allocation matrix data for the current service resource and the target replacement resource.

[0082] Specifically, the conversion task is defined as the customer signing a contract, paying a deposit, or clearly expressing the intention to complete the transaction. The total incentive amount is automatically calculated by the business system based on the product type and transaction amount.

[0083] Specifically, the initial link maintenance duration characteristic refers to the length of time from the first contact to the switchover trigger moment, and the takeover progress node characteristics include the time when the target replacement resource first proposes a transaction after takeover, the number of times key conversational phrases are used to facilitate customer confirmation, and the response speed to resolve core objections.

[0084] Specifically, the contribution splitting algorithm model adopts a weighted regression approach, with built-in first-contact incentive weights and transaction incentive weights, taking into account the dual contributions of initial development and final closure. The duration of the initial link maintenance corresponds to the first-contact incentive weight, and the characteristics of the takeover and promotion nodes correspond to the transaction incentive weight. The longer the initial link maintenance duration and the more complete the customer profile is established, the higher the basic contribution weight of the current service resources. The more intensive the key actions and the more timely the response in the takeover and promotion node characteristics, the higher the incremental contribution weight of the target replacement resources. The final output collaborative contribution allocation matrix data includes the incentive ratio that each party should receive, which is used for automatic commission settlement.

[0085] like Figure 3 As shown, a second aspect of the present invention provides a data processing system for resource scheduling, comprising: The profile building module is used to acquire initial multi-source feature data of the target object and construct a dynamic feature profile of the target object through a preset feature extraction model. The capability tag generation module is used to obtain the historical execution data and attribute characteristics of each service resource in the candidate resource set, and construct resource capability tags for each service resource through preset tag generation rules; An interactive data acquisition module is used to, in response to assigning the target object to a service resource in the candidate resource set as the current service resource, collect interactive data sequences in real time during the process of establishing an interactive link between the current service resource and the target object; The dynamic matching monitoring module is used to dynamically update the dynamic feature profile of the target object according to the interaction data sequence, and calculate and monitor the dynamic matching degree of the interaction link in real time based on the updated dynamic feature profile and the resource capability tag of the current service resource; when the dynamic matching degree meets the preset scheduling triggering condition, the latest dynamic feature profile of the target object is extracted. The intelligent scheduling decision engine module is used to call a preset intelligent scheduling decision algorithm to calculate the matching degree based on the latest dynamic feature profile, the resource capability tags of the remaining service resources in the candidate resource set excluding the current service resource, and the real-time working status parameters of each remaining service resource, and to select the target replacement resource from the remaining service resources. The link handover and takeover module is used to synchronize the context data of the interaction link to the target replacement resource, so as to control the target replacement resource to take over the interaction link with the target object.

[0086] To further enable those skilled in the art to clearly understand the practical application process and technical advantages of the data processing system described above, two preferred specific examples are provided below for detailed explanation, taking into account specific application scenarios in the field of automobile sales and service.

[0087] Application Scenario Example 1: Active scheduling triggering mechanism based on intent decay and interest shift.

[0088] Assuming the target audience (e.g., online lead generation customer C) generates initial multi-source feature data through external touchpoints, the system constructs an initial dynamic feature profile for them and routes them to the current service resource (e.g., salesperson A, whose resource capability tags are "deep professional knowledge" and "communication-oriented with a technical focus"). In the first interaction data sequence (e.g., the first call recording), the target audience inquires about a large number of underlying technical parameters, and the initial dynamic fit calculated by the system scores at a high level (e.g., 85 points).

[0089] As time progresses, the system collects subsequent interaction data sequences (such as instant messaging text streams) in real time. Using a feature extraction model, analysis reveals that the target object's response delays are increasing (instant feedback features are declining), and the semantic intent is shifting from "technical parameters" to transactional intents such as "vehicle availability" and "financial solutions." Based on this, the system dynamically updates its profile, increasing the feature weights of the "decision autonomy / efficiency-driven" dimension. However, the current service resource (Sales A) has not adjusted its strategy and continues to push technical documents.

[0090] At this point, the system detects that the dynamic fit of the interaction link continuously declines from 85 points to 60 points within a preset time window (e.g., within 48 hours), falling below the preset first threshold (e.g., 65 points), thus activating the "proactive early warning trigger" logic. The system then invokes the intelligent scheduling decision algorithm, based on the latest profile of the target object biased towards financial efficiency, to scan the remaining service resources and match a target replacement resource marked as idle (e.g., Salesperson B, whose capability tags include "strong transaction driving force" and "financial solution expert"). The system then generates a digital handover summary and pushes it to the service-side communication nodes of the management side and Salesperson B. After confirmation, it automatically completes the lossless synchronization of context data and the redirection and switching of the link, successfully mitigating the potential loss caused by the decline in fit.

[0091] Application Scenario Example 2: Passive Scheduling and Coordinated Takeover Mechanism Based on Negative Feedback Mutation Assessment.

[0092] Suppose a target audience (e.g., customer E) establishes an interaction link with a current service resource (e.g., salesperson D). During the interaction, the target audience repeatedly expresses concerns about "range anxiety" related to new energy vehicles. Salesperson D responds using official parameters, failing to effectively alleviate the concerns. This leads the target audience to generate negative evaluation text (e.g., "Core concern not resolved") in subsequent interaction nodes (e.g., micro-survey link).

[0093] The system's feature extraction model captures this data in real time. It detects that the variance in the fluctuation of negative emotions representing the target object exceeds a preset abnormal mutation threshold, thus activating the "behavioral mutation trigger" logic. The system deeply analyzes the latest dynamic feature profile, finding an extremely high aversion value in the "risk preference" dimension and sensitivity to "socially recognized characteristics (such as genuine car owner reviews)." Based on this, the intelligent scheduling decision algorithm optimizes and selects target replacement resources (such as Salesperson F) whose resource capability tags include "empathic communication advantages" and "community case operation advantages."

[0094] The system's collaborative takeover strategy includes: controlling the target replacement resource (Sales F) to intervene in the current interaction link and establishing a multi-party communication route with multiple nodes (such as creating a three-person exclusive group). Sales F takes the lead in handling the highly relevant task of alleviating battery anxiety, while Sales D assists in following up on routine product details. After completing the final conversion task, based on the initial link maintenance duration characteristics of Sales D and the takeover progress node characteristics of Sales F, the system calls a preset contribution splitting algorithm model to automatically generate collaborative contribution allocation matrix data (that is, allocating the overall incentive amount / commission according to the algorithm ratio), thereby achieving an internal collaborative incentive closed loop for system resource scheduling while ensuring a seamless transition in the target audience's experience.

[0095] A third aspect of the present invention provides an electronic device, comprising: at least one processor; and at least one memory communicatively connected to the processor, wherein: the memory stores program instructions executable by the processor, and the processor invokes the program instructions to perform the steps of the method as described in any of the above embodiments.

[0096] A fourth aspect of the present invention discloses a readable storage medium storing a computer program, which is executed by a processor as described in any of the above embodiments.

[0097] Computer-readable storage media can include: any entity or device capable of carrying computer programs, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory, random access memory, and software distribution media, etc. Computer programs include computer program code. Computer program code can be in the form of source code, object code, executable files, or some intermediate form, etc. Computer-readable storage media can include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory, random access memory, and software distribution media, etc.

[0098] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing a particular logical function or process, and the scope of the preferred embodiments of the invention includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as will be understood by those skilled in the art to which embodiments of the invention pertain.

[0099] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus or device (such as a computer-based system, a system including a processing module or other system that can fetch and execute instructions from, an instruction execution system, apparatus or device).

[0100] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A data processing method for resource scheduling, characterized in that, include: Acquire initial multi-source feature data of the target object, and construct a dynamic feature profile of the target object through a preset feature extraction model; Obtain historical execution data and attribute characteristics of each service resource in the candidate resource set, and construct resource capability tags for each service resource through preset tag generation rules; In response to assigning the target object to a service resource in the candidate resource set as the current service resource, during the process of establishing an interaction link between the current service resource and the target object, the interaction data sequence is collected in real time. The dynamic feature profile of the target object is dynamically updated according to the interaction data sequence, and the dynamic fit of the interaction link is calculated and monitored in real time based on the updated dynamic feature profile and the resource capability tag of the current service resource; when the dynamic fit meets the preset scheduling trigger condition, the latest dynamic feature profile of the target object is extracted. The preset intelligent scheduling decision algorithm is invoked to calculate the matching degree based on the latest dynamic feature profile, the resource capability tags of the remaining service resources in the candidate resource set excluding the current service resource, and the real-time working status parameters of each remaining service resource, and to select the target replacement resource from the remaining service resources. The context data of the interaction link is synchronized to the target replacement resource to control the target replacement resource to take over the interaction link with the target object.

2. The data processing method for resource scheduling according to claim 1, characterized in that, The process of acquiring initial multi-source feature data of the target object and constructing a dynamic feature profile of the target object using a preset feature extraction model includes: Acquire the one-way behavior trajectory data of the target object generated in the public domain data source, as well as the initial two-way communication text stream generated in the internal business terminal; The feature extraction model is used to jointly extract features from the unidirectional behavioral trajectory data and the initial bidirectional communication text stream to generate initial object feature vectors of the target object in the dimensions of decision autonomy, information processing style, risk preference, and social orientation, so as to construct the dynamic feature profile.

3. The data processing method for resource scheduling according to claim 1, characterized in that, The step of obtaining historical execution data and attribute characteristics of each service resource in the candidate resource set, and constructing resource capability tags for each service resource through preset tag generation rules, includes: Deeply mine the historical execution data of each service resource in the candidate resource set to extract explicit professional skill parameters and implicit trait parameters; The explicit professional skill parameters and the implicit trait parameters are structurally integrated to construct resource capability tags for each service resource, including communication mode dimension, professional field dimension, trait orientation dimension, and success case mapping dimension.

4. The data processing method for resource scheduling according to claim 1, characterized in that, The step of dynamically updating the dynamic feature profile of the target object based on the interaction data sequence includes: Natural language processing and sentiment analysis are performed on the interaction data sequence to extract the instantaneous feedback features of the target object and the intent decay features of the interaction intent within the current interaction cycle. A preset time-series decay function is introduced, and the weights of each dimension in the original dynamic feature profile of the target object are reduced or increased using the real-time feedback feature and the intent decay feature to generate the updated dynamic feature profile.

5. The data processing method for resource scheduling according to claim 1, characterized in that, The process of calculating and monitoring the dynamic fit of the interaction link in real time based on the updated dynamic feature profile and the resource capability tags of the current service resources includes: The updated dynamic feature profile is converted into a current object feature vector, and the resource capability tags of the current service resource are converted into a current resource feature vector. Based on the vector space distance between the current object feature vector and the current resource feature vector in each corresponding dimension, calculate the sub-fit score for each dimension; Based on the topic stage of the interaction data sequence, dynamic weights are assigned to each sub-fit score, and a weighted summation algorithm is used to generate the dynamic fit that represents the state of the interaction link.

6. The data processing method for resource scheduling according to claim 1, characterized in that, When the dynamic fit meets the preset scheduling triggering conditions, one or more of the following triggering logics are activated: Active warning trigger: When the quantized value of the dynamic fit is continuously lower than the first preset threshold and the duration exceeds the second preset threshold within a time window; Behavioral mutation trigger: Through semantic-level analysis of the interaction data sequence, it is detected that the fluctuation variance representing the negative emotions of the target object exceeds a preset abnormal mutation threshold; Command interruption trigger: Receives a collaborative support request command initiated by the current service resource.

7. The data processing method for resource scheduling according to claim 1, characterized in that, The real-time calculation and monitoring of the dynamic compatibility of the interaction link, and the determination of whether the preset scheduling triggering conditions are met, are based on a preset matching strategy matrix. The matching strategy matrix includes at least the first to fourth quadrant regions constructed based on the target object dimension features and the service resource dimension features, and the matching strategy matrix is ​​divided into the first to fourth quadrant regions with the target object decision autonomy as the horizontal axis and the service resource dominance as the vertical axis. For the mapping point position of the updated dynamic feature profile and the resource capability label of the current service resource in the matching strategy matrix, execute adaptive control logic: If the mapping point is located in the first quadrant region, it is determined that the scheduling triggering condition is not met, and the interaction link is maintained; If the mapping point is located in the second quadrant region, it is determined to be in a risk warning state, and service resources with buffer feature labels are marked as candidate queues in the remaining service resources; If the mapping point is located in the third quadrant region, it is determined that the scheduling triggering condition is met, and the intelligent scheduling decision algorithm is immediately invoked to perform the screening of the target replacement resource; If the mapping point is located in the fourth quadrant region, an interaction strategy adjustment instruction is issued to the service node corresponding to the current service resource.

8. The data processing method for resource scheduling according to claim 1, characterized in that, The matching degree calculation based on the latest dynamic feature profile, the resource capability tags of the remaining service resources, and the real-time working status parameters of each remaining service resource, and the selection of target replacement resources from the remaining service resources, includes: Obtain the real-time working status parameters of each of the remaining service resources. The real-time working status parameters include at least the current task load, the preset online identifier, and the emotional energy assessment value extracted based on recent interaction corpus. Based on the current task load and the online identifier, a candidate replacement subset is generated by filtering from the remaining service resources; Calculate the capability matching score between the resource capability tags of each service resource in the candidate replacement subset and the latest dynamic feature profile, and dynamically correct the capability matching score using the emotion energy assessment value; Historical collaboration data of each service resource in the candidate replacement subset is retrieved to calculate the collaboration success rate score. The service resource with the highest comprehensive score is selected as the target replacement resource by weighted optimization based on the corrected capability matching score and the collaboration success rate score.

9. The data processing method for resource scheduling according to claim 1, characterized in that, The step of synchronizing the context data of the interaction link to the target replacement resource, so as to control the target replacement resource to take over the interaction link with the target object, includes: Extract the full historical interaction data sequence between the current service resource and the target object, extract key insight nodes, and extract pending task status data from the preset historical interaction database; The extracted data is packaged according to a preset structured template to generate a digital handover summary; The digital handover summary is pushed to the server-side communication node corresponding to the target replacement resource for rendering and display, and the subsequent data communication route of the target object is redirected to the target replacement resource.

10. The data processing method for resource scheduling according to claim 1, characterized in that, After controlling the interaction link between the target replacement resource takeover and the target object, the following is also included: Real-time tracking of subsequent interaction nodes of the target replacement resource; in response to monitoring that the target replacement resource has completed a preset conversion task, obtaining the total incentive amount data corresponding to the conversion task. Extract the initial link maintenance duration feature of the current service resource, and the takeover promotion node feature of the target replacement resource; The initial link maintenance duration feature and the takeover advancement node feature are input into a preset contribution splitting algorithm model to automatically generate and output collaborative contribution allocation matrix data for the current service resource and the target replacement resource.

11. A data processing system oriented towards resource scheduling, characterized in that, include: The profile building module is used to acquire initial multi-source feature data of the target object and construct a dynamic feature profile of the target object through a preset feature extraction model. The capability tag generation module is used to obtain the historical execution data and attribute characteristics of each service resource in the candidate resource set, and construct resource capability tags for each service resource through preset tag generation rules; An interactive data acquisition module is used to, in response to assigning the target object to a service resource in the candidate resource set as the current service resource, collect interactive data sequences in real time during the process of establishing an interactive link between the current service resource and the target object; The dynamic matching monitoring module is used to dynamically update the dynamic feature profile of the target object according to the interaction data sequence, and calculate and monitor the dynamic matching degree of the interaction link in real time based on the updated dynamic feature profile and the resource capability tag of the current service resource; when the dynamic matching degree meets the preset scheduling triggering condition, the latest dynamic feature profile of the target object is extracted. The intelligent scheduling decision engine module is used to call a preset intelligent scheduling decision algorithm to calculate the matching degree based on the latest dynamic feature profile, the resource capability tags of the remaining service resources in the candidate resource set excluding the current service resource, and the real-time working status parameters of each remaining service resource, and to select the target replacement resource from the remaining service resources. The link handover and takeover module is used to synchronize the context data of the interaction link to the target replacement resource, so as to control the target replacement resource to take over the interaction link with the target object.

12. An electronic device, characterized in that, include: At least one processor; And at least one memory communicatively connected to the processor, wherein: the memory stores program instructions executable by the processor, and the processor invokes the program instructions to perform the steps of the resource-oriented data processing method as described in any one of claims 1-10.

13. A readable storage medium storing a computer program, characterized in that, When executed by a processor, the computer program implements the steps of the resource-oriented data processing method as described in any one of claims 1-10.

Images