Method, device and storage medium for distributing service personnel

Abstract

The application belongs to the technical field of artificial intelligence, and discloses a business personnel allocation method, device, equipment and storage medium, the allocation method comprising: extracting service characteristic variables of business service data of a business personnel; inputting the service characteristic variables into a personnel service capability time sequence attenuation model to output a business capability attenuation coefficient of the business personnel; inputting time sequence characteristic sequences in business-related data into a business demand evolution prediction model to predict business demand data of a target user for a target business; inputting a dynamic weight factor corresponding to the business demand data and the business capability attenuation coefficient into a preset matching decision model to obtain a target business personnel with the highest matching weight value with the target user. The application can be applied to financial technology, medical health and other business management systems, and solves the technical problem that the business personnel matched for the user in the prior art cannot truly meet the user's demand.

Description

Technical Field

[0001] This invention belongs to the field of artificial intelligence technology and is applied in the fields of financial technology and healthcare. In particular, it relates to a method, device, equipment and storage medium for the allocation of business personnel. Background Technology

[0002] In each business area, corresponding sales representatives are assigned to users based on their needs to provide them with complementary services. For example, in the insurance sector, such as life insurance, if a user's needs are for family risk protection and long-term savings, a life insurance planner skilled in family financial calculations can be assigned. Similarly, in the health insurance sector, if a user's needs are for medical expense insurance and convenient access to medical care, a health insurance advisor familiar with pre-existing conditions and medical resources can be assigned.

[0003] Current technology mechanically assigns a user to a salesperson. However, user needs change in real time, and the service capabilities of sales personnel also change constantly. If salespersons and users are mechanically assigned together, it is impossible to provide a salesperson with the right capabilities based on the user's real-time needs. Summary of the Invention

[0004] This invention provides a method, apparatus, device, and storage medium for allocating sales personnel, which can solve the technical problem that the sales personnel matched to users cannot truly meet the needs of users.

[0005] In a first aspect, the present invention provides a method for allocating business personnel, comprising: Obtain business service data from sales personnel; Extract the service feature variables from the business service data; The service characteristic variables are input into the personnel service capability time-series decay model, and the business capability decay coefficient of the salesperson is output. Acquire business-related data concerning the target users and the target business; Extract the time-series feature sequences from the business-related data; The time-series feature sequence is input into the business demand evolution prediction model to predict the business demand data of the target user for the target business. The business capability attenuation coefficient and the business demand data are respectively converted into dynamic weighting factors to influence the matching strategies of the salesperson and the target user; The dynamic weighting factor is used as a key parameter to adjust the matching weight value between the target user and the salesperson. It is input into a preset matching decision model so that the matching weight value between each salesperson and the target user can be corrected in real time through the dynamic weighting factor. The salesperson with the highest matching weight value with the target user is then selected as the target salesperson for the target user.

[0006] Secondly, the present invention provides a device for distributing business personnel, comprising: The first acquisition module is used to acquire the business service data of the salesperson; The first extraction module is used to extract service feature variables from the business service data; The output module is used to input the service feature variables into the personnel service capability time-series decay model and output the business capability decay coefficient of the salesperson. The second acquisition module is used to acquire business-related data of the target user and the target business. The second extraction module is used to extract the time-series feature sequences from the business-related data; The prediction module is used to input the time-series feature sequence into the business demand evolution prediction model to predict the business demand data of the target user for the target business. The conversion module is used to convert the business capability attenuation coefficient and the business demand data into dynamic weighting factors that influence the matching strategies of the salesperson and the target user, respectively. The decision module is used to input the dynamic weight factor as a key parameter for adjusting the matching weight value between the target user and the salesperson into a preset matching decision model, so as to correct the matching weight value between each salesperson and the target user in real time through the dynamic weight factor, and obtain the salesperson with the highest matching weight value with the target user as the target salesperson of the target user.

[0007] Thirdly, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described method for assigning business personnel.

[0008] Fourthly, a computer-readable storage medium is provided, which stores a computer program that, when executed by a processor, implements the steps of the above-mentioned method for assigning business personnel.

[0009] The aforementioned solution, implemented through the allocation method, apparatus, equipment, and storage media for business personnel, first collects business service data, then extracts service features, and finally calculates the attenuation coefficient using a time-series attenuation model. This logic quantifies the dynamic changes in business personnel's service capabilities in real time, overcoming the limitations of traditional static capability assessment and ensuring that capability assessment is always synchronized with actual service levels. By first collecting user business data, then extracting time-series features, and finally outputting demand data through a demand evolution prediction model, the solution captures the dynamic changes in user needs, achieving an upgrade from static demand recording to dynamic demand prediction. By converting the business personnel capability attenuation coefficient and user demand data into dynamic weighting factors, and using a matching decision model to adjust the matching weights in real time, the solution automatically increases the matching weight for high-capability business personnel when user needs escalate, and automatically decreases the matching weight for high-demand users when a business personnel's capabilities attenuate, ultimately outputting the target business personnel whose current capabilities best match the user's real-time needs. Therefore, this solution can solve the technical problem in existing technologies where matched business personnel fail to truly meet user needs. Attached Figure Description

[0010] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1 This is a flowchart illustrating a method for allocating business personnel in one embodiment of the present invention.

[0012] Figure 2 yes Figure 1 A flowchart of step S130.

[0013] Figure 3 yes Figure 1 A flowchart of step S160.

[0014] Figure 4 yes Figure 1 A flowchart of step S180.

[0015] Figure 5 This is another flowchart illustrating the method for allocating business personnel in one embodiment of the present invention.

[0016] Figure 6 This is a schematic diagram of a personnel allocation device in one embodiment of the present invention.

[0017] Figure 7 This is a schematic diagram of the structure of a computer device according to an embodiment of the present invention.

[0018] Figure 8 This is another structural schematic diagram of a computer device according to one embodiment of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Figure 1 A flowchart of the method for allocating business personnel provided in an embodiment of the present invention is shown below. Figure 1 As shown, the method for allocating business personnel provided in this embodiment of the invention includes the following steps.

[0021] Step S110: Obtain the salesperson's business service data.

[0022] Specifically, collect all data on the business services provided by salespersons within a certain period, including multi-dimensional records of sales behavior, customer service, and business processing. The data must include timestamps to reflect time sequence characteristics.

[0023] As a concrete example, in the life insurance scenario, obtaining business service data for a life insurance agent can include sales data, customer service data, and customer feedback data. Sales data includes the number of term life insurance and whole life insurance policies sold each month, the average sum assured per policy, and the new customer conversion rate (the percentage of customers who successfully sign a contract after the first communication). Customer service data can include response time to customer inquiries about policy terms and processing time for policy maintenance (such as beneficiary changes). Customer feedback data can include the number of customer complaints each month and satisfaction ratings.

[0024] As a concrete example, in the health insurance scenario, obtaining business service data for a health insurance agent can include sales data, health service data, and claims cooperation data. Sales data can include the number of critical illness insurance and multi-million dollar medical insurance policies sold each month, as well as the product matching accuracy rate. Health service data can include the completion rate of assisting customers in scheduling physical examinations and the response time for customer health consultations (such as the coverage of chronic diseases). Claims cooperation data can include the time taken to collect customer claim materials and the timeliness of claims progress updates.

[0025] Step S120: Extract the service feature variables of the business service data.

[0026] Specifically, in step S120, business service data can be preprocessed, which may include missing value handling, outlier cleaning, and data standardization.

[0027] Furthermore, service feature variables can be extracted based on the business logic and temporal characteristics of the preprocessed business service data. For example, temporal statistical features can be obtained by statistically aggregating the raw data within the sliding window, the temporal trend of a single feature can be fitted by linear regression, the slope can be extracted as a trend feature, the feature values ​​of historical moments can be extracted as inputs of the current moment as lag features to capture time dependence, and ratio indicators that reflect efficiency or quality can be constructed as ratio features.

[0028] Furthermore, among service feature variables, highly correlated features can be removed using Pearson correlation coefficients or mutual information (e.g., if the correlation between policy sales and the number of policies is 0.92, the more core sales figure can be retained) to avoid multicollinearity. Random forests can also be used to calculate feature importance scores, filtering features with scores above a preset value (e.g., in the life insurance scenario, the contract conversion rate and consultation response timeout rate have the highest scores and are prioritized for retention).

[0029] Step S130: Input the service characteristic variables into the personnel service capability time-series decay model, and output the business capability decay coefficient of the salesperson.

[0030] In some embodiments of the present invention, such as Figure 2 As shown, step S130 includes the following steps.

[0031] Step S131: Input the service feature variables into the LSTM time series model, fit the change pattern of the feature variables over time, and obtain the business capability decay coefficient of the salesperson.

[0032] Specifically, in this step, the input layer of the Long Short-Term Memory (LSTM) time series model can receive the time-series feature matrix from the service feature variables output in step S120, with a shape of (time step T, feature dimension N). For example, in a life insurance scenario, the input can be a matrix composed of 10 features (N=10) over 18 months (T=18).

[0033] More specifically, the first layer of the LSTM layer can include 32 hidden units, which can use the tanh activation function to preserve short-term temporal features (such as monthly fluctuations); the second layer of the LSTM layer can include 16 hidden units, which use the sigmoid activation function to capture long-term decay trends (such as a continuous decline in indicators for 6 months); the LSTM layer can also add a Dropout layer to prevent overfitting; the fully connected layer input to the LSTM layer can compress features through two fully connected network layers, and the final output layer uses the sigmoid activation function to output the business capability decay coefficient (with a value range of 0-1).

[0034] Specifically, when training an LSTM time series model, the time series data can be divided into a training set (the first 12 months) and a validation set (the last 6 months) in a 7:3 ratio, and time series cross-validation can be used to avoid data leakage. Furthermore, the optimizer for the LSTM time series model can be the Adam optimizer, which accelerates convergence by using a learning rate decay strategy (e.g., decaying by 10% every 5 epochs) given an initial learning rate.

[0035] Specifically, the business capability decay coefficient output by the model is a comprehensive quantification of the time-series trends of feature variables. If feature variables (such as contract conversion rate and service efficiency) show an upward or stable trend, the business capability decay coefficient approaches 1 (no capability decay). If feature variables show a continuous downward trend, and the rate of decline accelerates (e.g., the recommendation accuracy rate of health insurance agents drops from 90% to 70%, with the monthly decline increasing from 2% to 5%), the coefficient decreases as the rate of decline increases (e.g., 0.55). For example, in the life insurance scenario, the model calculates a decay coefficient of 0.62 by analyzing the 18-month trends of 10 features. Its breakdown logic can be: sales capability contribution 0.25 (weight 30%), service efficiency contribution 0.20 (weight 25%), and service quality contribution 0.17 (weight 45%).

[0036] Understandably, step S131 uses the LSTM model to perform time-series modeling of multi-dimensional service features, achieving for the first time accurate quantification and forward-looking prediction of the dynamic decay trend of salesperson's service capabilities, the comprehensive contribution of multiple features, and early decay signals, breaking through the limitations of traditional static evaluation or single indicator analysis.

[0037] In some embodiments of the present invention, after step S130, the following steps are further included.

[0038] Acquire new business service data, and extract service feature variables whose values ​​have changed from the new business service data as change feature vectors; Based on the changing feature vector and the change value of the changing feature vector, update the parameters related to the changing feature vector in the time-series decay model of personnel service capacity.

[0039] Specifically, when new business data (such as data from the current month) enters the system, feature changes can be detected, such as calculating the Euclidean distance between the new features and the features from the previous month. Features exceeding a threshold (such as 0.3) are marked as changed feature vectors. Further, parameter fine-tuning can be performed, such as freezing the parameters of the first layer of the LSTM and only updating the weights of the second layer and the fully connected layer (such as using the new data for 5 rounds of fine-tuning) to prevent the model from forgetting historical knowledge. Further, model fusion can be performed: a sliding window model fusion strategy is adopted to fuse the model outputs before and after the update according to a certain proportion, balancing timeliness and stability.

[0040] Understandably, the above steps achieve lightweight incremental iteration of the model by focusing on changing feature vectors to update parameters in a targeted manner. This avoids the high cost of retraining with full data and can capture the latest dynamics of service capabilities in real time, ensuring that the model always accurately adapts to the dynamic decay process of business personnel's capabilities. This breaks through the lag limitations of traditional static training and batch updates.

[0041] Step S140: Obtain business-related data related to the target user and the target business.

[0042] Specifically, in this step, the target business can be determined based on the specific scenario at the time of application. For example, in life insurance, business-related data could include basic user information (age, occupation, income), historical insurance data (type of policies purchased, coverage amount, payment period), interaction behavior data (customer service consultation content, policy inquiry frequency), and external data (credit records, asset certificates). In health insurance, business-related data could include user health records (medical examination reports, past medical history), claims records (claim amount, disease type, claim time), health behavior data (data synchronized from fitness apps, frequency of regular medical examinations), and product interaction data (health insurance product comparison records, coverage consultation logs).

[0043] Step S150: Extract the time-series feature sequence from the business-related data.

[0044] Specifically, in this step, the time granularity of the time series feature sequence can be determined based on the business cycle. For example, in the life insurance scenario, where users have a long insurance decision cycle, a week can be used as the time granularity, and the sequence length can be set to 12 (i.e., data from the past 12 weeks). In the health insurance scenario, where users' health behavior and claims needs have short-term fluctuations, a day can be used as the time granularity, and the sequence length can be set to 30 (i.e., data from the past 30 days).

[0045] Specifically, time-series feature sequences can include basic time-series features and advanced time-series features. Basic time-series features can include count features, interval features, and trend features. For example, count features could be the number of times health insurance products were consulted in the past 7 days and the frequency of life insurance policy inquiries in the past 30 days (the sum of daily inquiries). Interval features could be the time interval (in days) between the last insurance purchase and the current purchase, and the number of consecutive days without logging into the insurance software. Trend features could be the weekly average growth rate of the insurance intention score (1-10 points) obtained from the past 4 weeks, calculated through linear fitting. Advanced time-series features can include periodic features, abrupt change features, and embedding features. Periodic features could be the periodic components of user behavior extracted using Fourier transform (e.g., health insurance users habitually consult about claims at the beginning of each month, with a cycle of 30 days). Abrupt change features could be identified using a sliding t-test to identify the time point of feature abrupt change (e.g., after a user's medical examination results are abnormal, the frequency of health insurance inquiries suddenly increases from once a week to five times), and marked as binary features (1 = abrupt change exists, 0 = none). Embedded features can convert user consultation text (such as the coverage of critical illness insurance) into a 768-dimensional vector through the BERT model, and take the daily average vector as the text semantic temporal feature.

[0046] Step S160: Input the time-series feature sequence into the business demand evolution prediction model to predict the business demand data of the target user for the target business; In some embodiments of the present invention, such as Figure 3 As shown, step S160 includes the following steps.

[0047] Step S161: Input the time-series feature sequence into the Transformer model, capture the long-term dependencies in the time-series feature sequence, and predict the business demand tags of the target user for the target business. Step S162: Map the business requirement tag to service resource priority, which serves as the final business requirement data for the target user regarding the target business.

[0048] Specifically, in step S161, the input layer of the Transformer model can receive the temporal feature sequence output from step S150, with a shape of (sequence length T, feature dimension M) (e.g., T=30, M=20 in the health insurance scenario). Further, sine and cosine positional encoding can be used to add positional information to each time step, addressing the Transformer's insensitivity to temporal order. Further, multi-head self-attention in the encoder layer of the Transformer model (e.g., 8 attention heads, each with a dimension of 256) can be used to calculate the dependency weights between time steps through scaling dot product attention (e.g., the correlation weight between a health insurance user's abnormal physical examination on day 10 and a claims consultation on day 20). Further, the feedforward network in the Transformer model can be used to enhance the non-linear representation of features. Finally, the business requirement label can be output through the output layer of the Transformer model via a fully connected network and Softmax.

[0049] Furthermore, a training dataset for the Transformer model can be constructed. When constructing the dataset, historical user data can be used as samples, with the presence or absence of business needs in the next 7 days as labels. The dataset can then be divided chronologically into a training set (80%), a validation set (10%), and a test set (10%). Additionally, a weighted cross-entropy loss function can be used as the loss function for the Transformer model, with higher weights assigned to high-demand labels (e.g., three times that of no-demand labels) to address the imbalanced sample problem.

[0050] Specifically, in step S162, the demand labels output by the model can be converted into service resource priorities (0-100 points, with higher scores indicating higher priorities). A mapping function can be constructed based on historical demand-resource matching results. For example, in a life insurance scenario, label 0 can be mapped to priority 10, label 1 to priority 30, label 2 to priority 60, and label 3 to priority 90.

[0051] Understandably, through steps S161 and S162, the long-term temporal dependencies of user business needs are captured by the Transformer model. Combined with the mapping from demand tags to service resource priorities, the end-to-end accurate transformation from demand forecasting to resource allocation instructions is achieved, breaking through the limitation of traditional forecasting which only focuses on demand trend judgment and cannot directly guide the dynamic allocation of resources.

[0052] Step S170: Convert the business capability attenuation coefficient and the business demand data into dynamic weighting factors for influencing the matching strategy of the salesperson and the target user, respectively. In some embodiments of the present invention, step S170 includes: Based on the business capability attenuation coefficient and the business demand data, a two-dimensional weight matrix is ​​constructed, such that the salesperson dimension weight factor is associated with the business capability attenuation coefficient, and the user dimension weight factor is associated with the business demand data. The value of the business capability attenuation coefficient is inversely proportional to the value of the salesperson dimension weight factor, and the degree of demand of the business demand data is directly proportional to the value of the user dimension weight factor.

[0053] Specifically, the sales force can be grouped together as follows: The user set is Then the two-dimensional weight matrix , .in, For salesperson The dimensional weights are given by the formula: , To adjust the coefficient, ensure that the lower the attenuation coefficient, the faster the weight decreases; For users Dimension weights (and requirement priorities) (positive correlation), the formula is This is to map [0, 100] to [0.5, 0.9]. It represents the basic matching degree between salespersons and users, obtained by calculating historical interaction data vectors using cosine similarity.

[0054] Specifically, during dynamic weight updates, the updates can be triggered on a schedule, such as a full update of the weight matrix every hour (based on the latest decay coefficient and demand data); or they can be triggered by events, such as an incremental update triggered via a message queue when the salesperson's decay coefficient changes by more than 0.1 or the user's demand priority changes by more than 20 points.

[0055] Understandably, the above steps use a two-dimensional weight matrix to structure and quantify the dynamic relationship between the decline of a salesperson's ability and the degree of user demand. This allows the salesperson's weight to decrease in real time as their ability declines, while the user's weight to increase in real time as the urgency of their demand increases. This breaks through the limitations of traditional single-dimensional weighting or static rule matching, achieving precise quantitative matching of ability and demand. It provides an interpretable and dynamically adjustable mathematical basis for subsequent matching decisions.

[0056] Step S180: The dynamic weighting factor is used as a key parameter to adjust the matching weight value between the target user and the salesperson. It is input into the preset matching decision model so that the matching weight value between each salesperson and the target user is corrected in real time through the dynamic weighting factor, and the salesperson with the highest matching weight value with the target user is selected as the target salesperson of the target user.

[0057] It is understood that this invention, through its logic of first collecting business service data, then extracting service features, and finally calculating the attenuation coefficient using a time-series attenuation model, quantifies the dynamic changes in salesperson service capabilities in real time, breaking through the limitations of traditional static capability assessment and ensuring that capability assessment is always synchronized with actual service levels. By first collecting user business data, then extracting time-series features, and finally outputting demand data through a demand evolution prediction model, it captures the dynamic changes in user needs, achieving an upgrade from static demand recording to dynamic demand prediction. By converting the salesperson capability attenuation coefficient and user demand data into dynamic weighting factors, and using a matching decision model to adjust the matching weights in real time, when user needs escalate, the matching weight for high-capability salespersons is automatically increased; when a salesperson's capabilities attenuate, their matching weight with high-demand users is automatically decreased, ultimately outputting the target salesperson whose current capabilities best match the user's real-time needs. Therefore, this invention can solve the technical problem in the prior art where the salesperson matched to the user cannot truly meet the user's needs.

[0058] In some embodiments of the present invention, such as Figure 4 As shown, step S180 includes the following steps.

[0059] Step S181: Input the dynamic weight factor into the matching decision model, and adjust the matching weight values ​​between each salesperson and the target user through the attention mechanism in the matching decision model, so as to obtain the salesperson with the highest matching weight value with the target user as the target salesperson of the target user.

[0060] Specifically, the model input for the matching decision model can be a dynamic weight matrix. Salesperson-user historical matching records (feature vectors) and real-time business constraints (such as the current workload of the salesperson).

[0061] The attention mechanism in the model can employ a cross-attention layer, using user demand features as queries and salesperson capability features as keys and values ​​to calculate attention weights. Four attention heads can be set, focusing on aspects such as sales capability matching, service efficiency matching, user preference matching, and historical cooperation matching, respectively. Finally, a concatenated multi-dimensional matching result is output through a splicing header.

[0062] When matching the target user with the target salesperson, the weight vector output by the attention layer can be weighted and summed with the salesperson's comprehensive score (including historical service ratings), and heap sort can be used to select the top 3 salespersons with the highest scores from n salespersons (the specific number can be set as needed).

[0063] Understandably, step S181 transforms the dynamic weighting factor into an adaptive focus on the key dimensions of matching salespersons and users through the attention mechanism in the matching decision model. This can automatically strengthen the core matching features of high-demand users and low-decay salespersons, breaking through the mechanical nature of fixed weight allocation or average weighting in traditional matching, and realizing intelligent and precise adjustment of matching weights as both parties dynamically change.

[0064] In some embodiments of the present invention, such as Figure 5 As shown, after step S180, the following steps are also included.

[0065] Step S191: Assign the target salesperson to serve the target user; Step S192: Obtain the business feedback results from the target user regarding the target service; Step S193: Compare the deviation value between the business feedback result and the preset expected business result; Step S194: Based on the deviation value, adjust the relevant parameters of the personnel service capacity time-series decay model and the relevant parameters of the business demand evolution prediction model in reverse until the deviation value is reduced to within a preset threshold range.

[0066] Specifically, in step S192, in the life insurance scenario, the business feedback results can be the policy conversion rate, user satisfaction with the explanation of the terms and conditions, and the service cycle (the number of days from allocation to completion); in the health insurance scenario, the business feedback results can be the completeness of the claims materials submitted, the user's adoption rate of health advice, and the satisfaction with the timeliness of claims settlement.

[0067] Specifically, in step S194, for the personnel service capability decay model, if the deviation mainly stems from the assessment of the salesperson's capabilities (such as the actual service efficiency being lower than the model's prediction), the hidden layer weights of the personnel service capability decay model can be adjusted through gradient descent, focusing on optimizing parameters related to service efficiency features; if the deviation stems from inaccurate demand predictions in the business demand prediction model (such as the actual user demand being lower than the prediction), the attention head weights of the business demand prediction model can be fine-tuned to increase the attention ratio of recent user behavior features.

[0068] It is understandable that through steps S191 to S194, the personnel capability model and demand prediction model are adjusted synchronously in reverse through business feedback deviations, enabling the two core models to continuously iterate and adapt based on actual service effects. This breaks through the limitations of traditional solutions where model training and actual application are disconnected and single model optimization is isolated, and realizes the leap in the ability of the entire matching system from passive execution to proactive evolution.

[0069] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention. Software tools or components not belonging to this company that appear in the embodiments of this application are merely illustrative examples and do not represent actual use.

[0070] In one embodiment, a personnel allocation device is provided, which corresponds one-to-one with the personnel allocation method described in the above embodiments. For example... Figure 6 As shown, the distribution device includes a first acquisition module 610, a first extraction module 620, an output module 630, a second acquisition module 640, a second extraction module 650, a prediction module 660, a conversion module 670, and a decision module 680. Detailed descriptions of each functional module are as follows: The first acquisition module 610 is used to acquire the business service data of the salesperson; The first extraction module 620 is used to extract service feature variables from the business service data; Output module 630 is used to input the service feature variables into the personnel service capability time-series decay model and output the business capability decay coefficient of the salesperson. The second acquisition module 640 is used to acquire business-related data of the target user and the target business. The second extraction module 650 is used to extract the time-series feature sequence from the business-related data; Prediction module 660 is used to input the time-series feature sequence into the business demand evolution prediction model to predict the business demand data of the target user for the target business; The conversion module 670 is used to convert the business capability attenuation coefficient and the business demand data into dynamic weighting factors for influencing the matching strategy of the salesperson and the target user, respectively. The decision module 680 is used to input the dynamic weight factor as a key parameter for adjusting the matching weight value between the target user and the salesperson into a preset matching decision model, so as to correct the matching weight value between each salesperson and the target user in real time through the dynamic weight factor, and obtain the salesperson with the highest matching weight value with the target user as the target salesperson of the target user.

[0071] In one embodiment, the output module 630 is specifically used for: The service feature variables are input into an LSTM time series model to fit the change pattern of the feature variables over time, thereby obtaining the salesperson's business capability decay coefficient.

[0072] In one embodiment, the prediction module 660 is specifically used for: The time-series feature sequence is input into the Transformer model to capture the long-term dependencies in the time-series feature sequence and predict the business demand tags of the target user for the target business. The business requirement tags are mapped to service resource priorities, which serve as the final business requirement data for the target user regarding the target business.

[0073] In one embodiment, the decision module 680 is specifically used for: The dynamic weighting factor is input into the matching decision model, and the matching weight values ​​between each salesperson and the target user are adjusted through the attention mechanism in the matching decision model. The salesperson with the highest matching weight value with the target user is then selected as the target salesperson for the target user.

[0074] In one embodiment, the conversion module 670 is specifically used for: Based on the business capability attenuation coefficient and the business demand data, a two-dimensional weight matrix is ​​constructed, such that the salesperson dimension weight factor is associated with the business capability attenuation coefficient, and the user dimension weight factor is associated with the business demand data. The value of the business capability attenuation coefficient is inversely proportional to the value of the salesperson dimension weight factor, and the degree of demand of the business demand data is directly proportional to the value of the user dimension weight factor.

[0075] In one embodiment, the decision module 680 is further configured to: Assign the target salesperson to serve the target user; Obtain the business feedback results from the target user regarding the target service; Compare the deviation between the business feedback result and the preset expected business result; Based on the deviation value, the relevant parameters of the personnel service capacity time-series decay model and the relevant parameters of the business demand evolution prediction model are adjusted in reverse until the deviation value is reduced to within a preset threshold range.

[0076] In one embodiment, the output module 630 is further configured to: Acquire new business service data, and extract service feature variables whose values ​​have changed from the new business service data as change feature vectors; Based on the changing feature vector and the change value of the changing feature vector, update the parameters related to the changing feature vector in the time-series decay model of personnel service capacity.

[0077] This invention provides a device for allocating sales personnel. It first collects sales service data, then extracts service features, and finally calculates the attenuation coefficient using a time-series attenuation model. This logic quantifies the dynamic changes in sales personnel's service capabilities in real time, overcoming the limitations of traditional static capability assessments and ensuring that capability assessments are always synchronized with actual service levels. By first collecting user business data, then extracting time-series features, and finally outputting demand data through a demand evolution prediction model, it captures the dynamic changes in user needs, achieving an upgrade from static demand recording to dynamic demand prediction. By converting the sales personnel capability attenuation coefficient and user demand data into dynamic weighting factors, and using a matching decision model to adjust the matching weights in real time, when user needs escalate, the matching weight for high-capability sales personnel is automatically increased; when a sales personnel's capabilities attenuate, their matching weight with high-demand users is automatically decreased. Ultimately, it outputs the target sales personnel whose current capabilities best match the user's real-time needs. Therefore, this invention solves the technical problem in the prior art where the sales personnel matched to users cannot truly meet user needs.

[0078] Specific limitations regarding the personnel allocation device can be found in the above description of the personnel allocation method, and will not be repeated here. Each module in the aforementioned personnel allocation device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the computer device's memory as software, so that the processor can call and execute the corresponding operations of each module.

[0079] Based on the above method of allocating business personnel, such as Figure 7 As shown in the diagram, this embodiment of the invention also provides a structural schematic of a device for assigning business personnel. The device includes a processor 71 and a memory 72 coupled to the processor 71. The memory 72 stores a computer program, which, when executed by the processor 71, causes the processor 71 to perform the steps of the business personnel assignment method described in the above embodiment.

[0080] For further details regarding the implementation of the above technical solution by the processor 71 in the device for the above-mentioned personnel allocation steps, please refer to the description of the personnel allocation method provided in the above-mentioned embodiments of the invention, which will not be repeated here.

[0081] The processor 71 can also be called a CPU (Central Processing Unit). The processor 71 may be an integrated circuit chip with signal processing capabilities. The processor 71 can also be a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. The general-purpose processor can be a microprocessor, or the processor 71 can be any conventional processor.

[0082] like Figure 8 As shown in the diagram, this embodiment of the invention also provides a schematic diagram of a computer-readable storage medium, on which a readable computer program 81 is stored. The computer program 81 can be stored in the storage medium in the form of a software product, including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods described in various embodiments of the invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, magnetic disks or optical disks, ROM (Read-Only Memory), RAM (Random Access Memory), or terminal devices such as computers, servers, mobile phones, and tablets.

[0083] In the embodiments provided by this invention, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, apparatuses, or modules, and may be electrical, mechanical, or other forms.

[0084] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0085] Furthermore, the functional modules in the various embodiments of this invention can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can be stored in a computer-readable storage medium.

[0086] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product.

[0087] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can store or a data storage device such as a server or data center that integrates one or more available media. The available medium may be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium, or a semiconductor medium (e.g., SSD (solid state disk)).

[0088] The technical solution provided by the present invention has been described in detail above. Specific examples have been used to illustrate the principle and implementation of the present invention. The description of the above embodiments is only for the purpose of helping to understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation and application scope based on the idea of ​​the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

[0089] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, optical storage, etc.) containing computer-usable program code.

[0090] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0091] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0092] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0093] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for allocating sales personnel, characterized in that, include: Obtain business service data from sales personnel; Extract the service feature variables from the business service data; The service characteristic variables are input into the personnel service capability time-series decay model, and the business capability decay coefficient of the salesperson is output. Acquire business-related data concerning the target users and the target business; Extract the time-series feature sequences from the business-related data; The time-series feature sequence is input into the business demand evolution prediction model to predict the business demand data of the target user for the target business. The business capability attenuation coefficient and the business demand data are respectively converted into dynamic weighting factors to influence the matching strategies of the salesperson and the target user; The dynamic weighting factor is used as a key parameter to adjust the matching weight value between the target user and the salesperson. It is input into a preset matching decision model so that the matching weight value between each salesperson and the target user can be corrected in real time through the dynamic weighting factor. The salesperson with the highest matching weight value with the target user is then selected as the target salesperson for the target user.

2. The method for allocating business personnel according to claim 1, characterized in that, The step of inputting the service characteristic variables into the personnel service capability time-series decay model and outputting the salesperson's service capability decay coefficient includes: The service feature variables are input into an LSTM time series model to fit the change pattern of the feature variables over time, thereby obtaining the salesperson's business capability decay coefficient.

3. The method for allocating business personnel according to claim 1, characterized in that, The step of inputting the time-series feature sequence into the business demand evolution prediction model to predict the business demand data of the target user for the target business includes: The time-series feature sequence is input into the Transformer model to capture the long-term dependencies in the time-series feature sequence and predict the business demand tags of the target user for the target business. The business requirement tags are mapped to service resource priorities, which serve as the final business requirement data for the target user regarding the target business.

4. The method for allocating business personnel according to claim 1, characterized in that, The step of using the dynamic weighting factor as a key parameter to adjust the matching weight value between the target user and the salesperson, and inputting it into a preset matching decision model, so as to correct the matching weight value between each salesperson and the target user in real time through the dynamic weighting factor, and obtaining the salesperson with the highest matching weight value with the target user as the target salesperson of the target user, includes: The dynamic weighting factor is input into the matching decision model, and the matching weight values ​​between each salesperson and the target user are adjusted through the attention mechanism in the matching decision model. The salesperson with the highest matching weight value with the target user is then selected as the target salesperson for the target user.

5. The method for allocating business personnel according to claim 1, characterized in that, The step of converting the business capability attenuation coefficient and the business demand data into dynamic weighting factors for influencing the matching strategy for the salesperson and the target user includes: Based on the business capability attenuation coefficient and the business demand data, a two-dimensional weight matrix is ​​constructed, such that the salesperson dimension weight factor is associated with the business capability attenuation coefficient, and the user dimension weight factor is associated with the business demand data. The value of the business capability attenuation coefficient is inversely proportional to the value of the salesperson dimension weight factor, and the degree of demand of the business demand data is directly proportional to the value of the user dimension weight factor.

6. The method for allocating business personnel according to claim 1, characterized in that, After inputting the dynamic weighting factor as a key parameter for adjusting the matching weight value between the target user and the salesperson into a preset matching decision model, so as to correct the matching weight value between each salesperson and the target user in real time through the dynamic weighting factor, and obtaining the salesperson with the highest matching weight value with the target user as the target user's target salesperson, the method further includes: Assign the target salesperson to serve the target user; Obtain the business feedback results from the target user regarding the target service; Compare the deviation between the business feedback result and the preset expected business result; Based on the deviation value, the relevant parameters of the personnel service capacity time-series decay model and the relevant parameters of the business demand evolution prediction model are adjusted in reverse until the deviation value is reduced to within a preset threshold range.

7. The method for allocating business personnel according to claim 1, characterized in that, After inputting the service characteristic variables into the personnel service capability time-series decay model and outputting the salesperson's service capability decay coefficient, the method further includes: Acquire new business service data, and extract service feature variables whose values ​​have changed from the new business service data as change feature vectors; Based on the changing feature vector and the change value of the changing feature vector, update the parameters related to the changing feature vector in the time-series decay model of personnel service capacity.

8. A device for distributing sales personnel, characterized in that, include: The first acquisition module is used to acquire the business service data of the salesperson; The first extraction module is used to extract service feature variables from the business service data; The output module is used to input the service feature variables into the personnel service capability time-series decay model and output the business capability decay coefficient of the salesperson. The second acquisition module is used to acquire business-related data of the target user and the target business. The second extraction module is used to extract the time-series feature sequences from the business-related data; The prediction module is used to input the time-series feature sequence into the business demand evolution prediction model to predict the business demand data of the target user for the target business. The conversion module is used to convert the business capability attenuation coefficient and the business demand data into dynamic weighting factors that influence the matching strategies of the salesperson and the target user, respectively. The decision module is used to input the dynamic weight factor as a key parameter for adjusting the matching weight value between the target user and the salesperson into a preset matching decision model, so as to correct the matching weight value between each salesperson and the target user in real time through the dynamic weight factor, and obtain the salesperson with the highest matching weight value with the target user as the target salesperson of the target user.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the personnel allocation method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the personnel allocation method as described in any one of claims 1 to 7.

Images