Method, device and equipment for supporting courier to mine potential customers and storage medium

Abstract

The invention relates to the technical field of computers, in particular to a method, device and equipment for supporting a courier to mine potential customers and a storage medium. The method for supporting the courier to mine the potential customers comprises the following steps: screening out peripheral customers of the current courier according to the real-time position of the courier, and constructing a peripheral customer distribution map; then, a big data mixed recommendation model is adopted to obtain a final recommendation score of each client; calculating a local optimal mining range in the area range, and calculating a recommended customer visit route in the local optimal mining range; and establishing a mining recommendation table according to the final recommendation score, generating a follow-up plan according to the mining recommendation table, and displaying the mining recommendation table, the follow-up plan and the local optimal mining range in a peripheral customer distribution map. Through the technical scheme of the invention, a courier does not need to depend on off-line blind popularization, peripheral high-value customers and personalized matching customers can be quickly positioned, and the time cost of customer mining is reduced.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and in particular to a method, apparatus, device, and storage medium for supporting couriers in identifying potential customers. Background Technology

[0002] The express delivery industry is currently facing a lack of technological tools in the potential customer development stage, resulting in couriers' low efficiency in customer acquisition and resource mining capabilities. In terms of customer acquisition methods, couriers still heavily rely on traditional models such as offline promotion and referrals from acquaintances. Offline promotion mainly involves "street canvassing" by distributing flyers and asking around at each shop, which not only has limited coverage but also fails to accurately target high-value customers (such as e-commerce companies with large estimated shipment volumes or service industry customers with high-frequency shipments), often consuming a lot of time but only yielding sporadic inquiries. Referrals from acquaintances rely on personal networks, which are unstable and difficult to scale.

[0003] Meanwhile, the customer information accumulated by delivery personnel daily often becomes ineffective data due to a lack of technological means to integrate and analyze it. If existing customer information could be used to identify nearby customers with similar needs, then the customer resources held by delivery companies could be transformed. Summary of the Invention

[0004] In view of the shortcomings of the prior art, the purpose of this invention is to provide a method, apparatus, equipment and storage medium to support couriers in identifying potential customers, aiming to solve the problems of couriers lacking efficient technical means to identify potential customers, and having traditional and inefficient promotion models in the prior art.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: The first aspect of this invention provides a method for supporting couriers in identifying potential customers, comprising the following steps: obtaining the real-time location of the courier; calculating the area that the courier can quickly reach based on the real-time location; filtering out the courier's surrounding customers from the courier's internal customer data based on the area, obtaining surrounding customer data, and constructing a surrounding customer distribution map; inputting the surrounding customer data into a pre-constructed big data hybrid recommendation model to obtain a final recommendation score for each customer, wherein the big data hybrid recommendation model includes a recommendation sub-model and a collaborative filtering sub-model; calculating the locally optimal mining range within the area based on the area, the geographical location of the surrounding customer data, and the final recommendation score for each customer, and calculating the recommended customer visit route within the locally optimal mining range; establishing a mining recommendation table based on the final recommendation score, generating a follow-up plan based on the mining recommendation table, and displaying the mining recommendation list, follow-up plan, and locally optimal mining range on the surrounding customer distribution map.

[0006] Optionally, in a first implementation of the first aspect of the present invention, the step of obtaining the real-time location of the courier, calculating the area range that the courier can quickly reach based on the real-time location, filtering out the surrounding customers of the current courier from the customer data within the courier company based on the area range, obtaining surrounding customer data, and constructing a surrounding customer distribution map specifically includes: obtaining the real-time location of the courier; using a distance algorithm, calculating the area range that the courier can quickly reach based on the real-time location of the courier and a preset distance range; using a matching algorithm, filtering out customers within the area range from the customer data within the courier company to obtain surrounding customer data; mapping the surrounding customer data to the area range that the courier can quickly reach, and outputting a surrounding customer distribution map corresponding to the area range.

[0007] Optionally, in a second implementation of the first aspect of the present invention, the step of inputting surrounding customer data into a pre-built big data hybrid recommendation model to obtain a final recommendation score for each customer, wherein the big data hybrid recommendation model includes a recommendation sub-model and a collaborative filtering sub-model, specifically including: inputting surrounding customer data into the pre-built recommendation sub-model to calculate the cosine similarity between the feature vector of each customer and the courier's preference vector to obtain a first score; inputting surrounding customer data into the collaborative filtering sub-model to calculate the courier's success rate in developing each customer to obtain a second score; and calculating the final recommendation score based on the first score and the second score through weighted fusion.

[0008] Optionally, in the third implementation of the first aspect of the present invention, the step of inputting surrounding customer data into a pre-constructed recommendation sub-model, calculating the cosine similarity between the feature vector of each customer and the courier's preference vector, and obtaining a first score specifically includes: obtaining the courier's service records, which include service type, service industry, and parcel volume; cleaning and preprocessing the courier's service records and surrounding customer data to obtain courier-side data and potential customer data; uniformly converting the courier-side data and potential customer data into a numerical format that the model can calculate, and keeping the vector dimensions of the client and courier ends consistent to construct the feature vector of each customer and the courier's preference vector; and using the recommendation sub-model, calculating the similarity between a single customer vector and a single courier vector using the cosine similarity formula to obtain a first score.

[0009] Optionally, in the fourth implementation of the first aspect of the present invention, the step of inputting surrounding customer data into the collaborative filtering sub-model to calculate the success rate of the courier in developing each customer and obtain a second score specifically includes: obtaining customer development records of multiple couriers, wherein the customer development records include basic customer information and development results; combining the surrounding customer data with the courier's customer development records and cleaning the data, and converting the cleaned data into a matrix format that can be calculated by collaborative filtering; using the collaborative filtering sub-model, filtering similar customers through the Pearson correlation coefficient and calculating the customer development success rate to obtain the second score.

[0010] Optionally, in the fifth implementation of the first aspect of the present invention, the step of calculating the locally optimal mining range within the region based on the regional range, the geographical location of surrounding customer data, and the final recommendation score of each customer, and calculating the recommended customer visit route within the locally optimal mining range, specifically includes: marking each customer within the regional range based on the geographical location of surrounding customer data, and associating each customer with their final recommendation score; constructing a sliding window, setting the shape and sliding step size of the sliding window; forming multiple local regions within the regional range through the sliding window, calculating the total recommendation score window by window according to the calculation rules, sorting the multiple local regions according to the score size to obtain the locally optimal mining range; and using a weighted TSP algorithm to calculate the recommended customer visit route within the locally optimal mining range based on the customer's geographical location and final recommendation score.

[0011] Optionally, in the sixth implementation of the first aspect of the present invention, the step of establishing a mining recommendation table based on the final recommendation score, generating a follow-up plan based on the mining recommendation table, and displaying the mining recommendation list, follow-up plan, and locally optimal mining range on the surrounding customer distribution map specifically includes: establishing a mining recommendation table based on the final recommendation score, and sorting customers according to the size of the final recommendation score; generating a follow-up plan based on the customer sorting in the mining recommendation table, the follow-up plan including follow-up time and follow-up method, outputting the follow-up plan to the courier to obtain follow-up data fed back by the courier; forming a follow-up analysis report based on the follow-up data fed back by the courier, displaying the mining recommendation list, follow-up plan, follow-up analysis report, and locally optimal mining range on the surrounding customer distribution map, and outputting the surrounding customer distribution map to the courier.

[0012] A second aspect of the present invention provides a device for supporting couriers in identifying potential customers, comprising: a filtering module for acquiring the real-time location of the courier, calculating the area that the courier can quickly reach based on the real-time location, filtering out the courier's surrounding customers from internal customer data of the courier company based on the area, obtaining surrounding customer data, and constructing a surrounding customer distribution map; a calculation module for inputting the surrounding customer data into a pre-constructed big data hybrid recommendation model to obtain a final recommendation score for each customer, wherein the big data hybrid recommendation model includes a recommendation sub-model and a collaborative filtering sub-model; a regional recommendation module for calculating a locally optimal mining range within the regional range based on the regional range, the geographical location of the surrounding customer data, and the final recommendation score for each customer, and calculating a recommended customer visit route within the locally optimal mining range; and a display module for establishing a mining recommendation table based on the final recommendation score, generating a follow-up plan based on the mining recommendation table, and displaying the mining recommendation list, follow-up plan, and locally optimal mining range on the surrounding customer distribution map.

[0013] Optionally, in a first implementation of the second aspect of the present invention, the filtering module includes: a range calculation unit, used to obtain the real-time location of the courier, and using a distance algorithm to calculate the area range that the courier can quickly reach based on the real-time location of the courier and a preset distance range; a filtering unit, used to use a matching algorithm to filter out customers within the area range from the customer data inside the courier company, and obtain surrounding customer data; and a distribution unit, used to map the surrounding customer data to the area range that the courier can quickly reach, and output a surrounding customer distribution map corresponding to the area range.

[0014] Optionally, in a second implementation of the second aspect of the present invention, the calculation module includes: a first calculation unit, used to input surrounding customer data into a pre-built recommendation sub-model, calculate the cosine similarity between the feature vector of each customer and the courier's preference vector, and obtain a first score; a second calculation unit, used to input surrounding customer data into a collaborative filtering sub-model, calculate the courier's success rate in developing each customer, and obtain a second score; and a weighted calculation unit, used to calculate the final recommendation score based on the first score and the second score through weighted fusion.

[0015] Optionally, in a third implementation of the second aspect of the present invention, the first calculation unit includes: a first acquisition subunit, used to acquire the courier's service records, the service records including service type, service industry and parcel volume, and to clean and preprocess the courier's service records and surrounding customer data to obtain courier-side data and potential customer data; a first transformation subunit, used to uniformly transform the courier-side data and potential customer data into a numerical format that the model can calculate, and to keep the vector dimensions of the customer and courier sides consistent, so as to construct the feature vector of each customer and the courier's preference vector; and a first calculation subunit, used to use a recommendation sub-model to calculate the similarity between a single customer vector and a single courier vector using the cosine similarity formula to obtain a first score.

[0016] Optionally, in a fourth implementation of the second aspect of the present invention, the second calculation unit includes: a second acquisition subunit, used to acquire customer development records of multiple couriers, the customer development records including basic customer information and development results; a second conversion subunit, used to combine surrounding customer data with the customer development records of couriers and perform data cleaning, and after cleaning, convert the data into a matrix format that can be calculated by collaborative filtering; and a second calculation subunit, used to use a collaborative filtering sub-model to filter similar customers through Pearson correlation coefficient and calculate the customer development success rate to obtain a second score.

[0017] Optionally, in a fifth implementation of the second aspect of the present invention, the region recommendation module includes: a labeling unit, used to label each customer within a region based on the geographical location of surrounding customer data, and associate each customer's final recommendation score; a window construction unit, used to construct a sliding window, and set the shape and sliding step size of the sliding window; a local calculation unit, used to form multiple local regions within the region through the sliding window, calculate the total recommendation score window by window according to calculation rules, sort the multiple local regions according to the score size, and obtain the local optimal mining range; and a route generation unit, used to calculate the recommended customer visit route within the local optimal mining range using a weighted TSP algorithm, based on the customer's geographical location and final recommendation score.

[0018] Optionally, in a sixth implementation of the second aspect of the present invention, the display module includes: a sorting unit, configured to establish a mining recommendation table based on the final recommendation score and sort customers according to the size of the final recommendation score; a planning unit, configured to generate a follow-up plan based on the customer sorting in the mining recommendation table, the follow-up plan including follow-up time and follow-up method, and output the follow-up plan to the courier to obtain follow-up data fed back by the courier; and a display unit, configured to generate a follow-up analysis report based on the follow-up data fed back by the courier, display the mining recommendation list, follow-up plan, follow-up analysis report, and local optimal mining range in the surrounding customer distribution map, and output the surrounding customer distribution map to the courier.

[0019] A third aspect of the present invention provides a device for supporting couriers in identifying potential customers, comprising a memory and at least one processor, wherein the memory stores computer-readable instructions; the at least one processor invokes the computer-readable instructions in the memory to perform the various steps of the method for supporting couriers in identifying potential customers as described above.

[0020] A fourth aspect of the present invention provides a computer-readable storage medium storing computer-readable instructions, which, when executed by a processor, implement the steps described above for supporting the method of a courier in identifying potential customers.

[0021] Beneficial Effects: This invention provides a method to support couriers in identifying potential customers. The method first obtains the courier's real-time location, calculates the area the courier can quickly reach based on that location, and then filters out nearby customers from the courier company's internal customer data based on that area. Next, a distribution map of nearby customers is constructed, allowing couriers to visually see the distribution of potential customers in their vicinity, which helps them plan their business operations. Then, the nearby customer data is input into a pre-built big data hybrid recommendation model that includes recommendation sub-models and collaborative filtering sub-models to obtain a final recommendation score for each customer. This big data hybrid recommendation model comprehensively considers both customer and courier information and uses data to intuitively... The method displays the final recommendation score of potential customers, improving the success rate of business operations. Then, based on the regional scope, the geographical location of surrounding customer data, and the final recommendation score of each customer, it calculates the locally optimal mining range within the region. This further recommends smaller areas for business operations within a larger selectable range, improving feasibility. The method also calculates recommended customer visit routes within the locally optimal mining range, improving efficiency during business operations. Finally, a mining recommendation table is established based on the final recommendation score, and a follow-up plan is generated from the mining recommendation table. The mining recommendation list, follow-up plan, and locally optimal mining range are displayed on the surrounding customer distribution map, allowing couriers to intuitively see the nearby customer resources and plan arrangements, providing valuable support for mining potential customers. Attached Figure Description

[0022] Figure 1 This is a first flowchart of a method for supporting couriers in identifying potential customers, provided by an embodiment of the present invention. Figure 2 This is a second flowchart illustrating a method for supporting couriers in identifying potential customers, as provided in an embodiment of the present invention. Figure 3 This is a third flowchart of a method for supporting couriers in identifying potential customers, provided by an embodiment of the present invention. Figure 4 This is a fourth flowchart of a method for supporting couriers in identifying potential customers, provided by an embodiment of the present invention. Figure 5 A fifth flowchart for a method to support couriers in identifying potential customers, provided by an embodiment of the present invention; Figure 6 A sixth flowchart for a method to support couriers in identifying potential customers, provided in an embodiment of the present invention; Figure 7 A seventh flowchart for a method to support couriers in identifying potential customers, provided in an embodiment of the present invention; Figure 8A schematic diagram of a device for supporting couriers in identifying potential customers, provided in an embodiment of the present invention; Figure 9 Another structural schematic diagram of the device for supporting couriers in identifying potential customers provided in an embodiment of the present invention; Figure 10 This is a schematic diagram of the structure of a device that supports couriers in identifying potential customers, provided as an embodiment of the present invention. Detailed Implementation

[0023] This invention provides a method, apparatus, device, and storage medium to support couriers in identifying potential customers. The invention first acquires the courier's real-time location, calculates the area the courier can quickly reach based on that location, then filters out nearby customers from the courier company's internal customer data based on this area, and further constructs a nearby customer distribution map to provide the courier with an intuitive spatial distribution view of potential customers in the vicinity, effectively helping them to scientifically plan the pace and priority of their business operations. Subsequently, the nearby customer data is input into a pre-constructed big data hybrid recommendation model that includes a recommendation sub-model and a collaborative filtering sub-model to generate a final recommendation score for each customer. This model comprehensively considers the matching degree between customer characteristics and courier capabilities, quantifying the success rate of potential customer cooperation in a data-driven form, significantly improving the accuracy of the courier's business operations. Finally, based on the aforementioned area range, the geographical location of nearby customers, and the final recommendation score for each customer... The recommendation score calculates the locally optimal mining range within a broad selectable area, focusing on high-value niches to significantly reduce the randomness of business expansion and improve feasibility. Simultaneously, this method further calculates recommended customer visit routes within the locally optimal mining range, optimizing route planning to reduce ineffective courier trips and significantly improve time efficiency and resource utilization. Finally, a mining recommendation table is established based on the final recommendation score, and a customer follow-up plan is generated from this table. The mining recommendation list, follow-up plan, and locally optimal mining range are simultaneously displayed on the surrounding customer distribution map. Courier receives comprehensive information including customer resource distribution, business priorities, and execution plans, enabling them to intuitively grasp the value of surrounding customers and implementation arrangements, providing accurate and feasible technical support for efficiently mining potential customers.

[0024] The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0025] For ease of understanding, the specific process of the embodiments of the present invention is described below. Please refer to [link / reference]. Figure 1 The first embodiment of the method for supporting couriers to discover potential customers in this invention includes: S101. Obtain the real-time location of the courier, calculate the area that the courier can quickly reach based on the real-time location, filter out the surrounding customers of the current courier from the customer data of the courier company based on the area, obtain the surrounding customer data, and construct a surrounding customer distribution map. S102. Input the surrounding customer data into the pre-built big data hybrid recommendation model to obtain the final recommendation score for each customer. The big data hybrid recommendation model includes a recommendation sub-model and a collaborative filtering sub-model. S103. Based on the area range, the geographical location of surrounding customer data, and the final recommendation score of each customer, calculate the local optimal mining range within the area range, and calculate the recommended customer visit route within the local optimal mining range; S104. Based on the final recommendation score, establish a mining recommendation table, and generate a follow-up plan based on the mining recommendation table. Display the mining recommendation list, follow-up plan, and locally optimal mining range on the surrounding customer distribution map.

[0026] In this embodiment, addressing the current lack of effective tools for couriers to identify potential customers, the system acquires the courier's real-time location and uses this location as the center to construct a search range. It then filters out nearby customers from the courier company's internal customer data, obtaining surrounding customer data and constructing a surrounding customer distribution map. This allows couriers to know the number and location of potential customers in their vicinity, providing basic information support for potential customer discovery. Specifically, the courier's real-time location is collected in coordinate form to establish a coordinate range. The courier company's internal customer data records the location information of potential customers. Based on this location information, the geographical coordinates of each potential customer can be determined, thus identifying which potential customers fall within the coordinate range. Potential customers are marked within the area based on their geographical coordinates, and a surrounding customer distribution map is generated. This allows couriers to intuitively see where potential customers are located and in which directions the density of potential customers is higher. When conducting business, they can choose areas with higher potential customer density, saving travel time.

[0027] After acquiring surrounding customer data, to improve the success rate of business negotiations, this embodiment further uses a pre-built hybrid recommendation model to analyze potential customers and obtain a final recommendation score for each customer. The higher the score, the higher the success rate of cooperation with that customer, thus improving accuracy. Specifically, the hybrid recommendation model includes a recommendation sub-model and a collaborative filtering sub-model, analyzing potential customers from multiple perspectives and then summarizing the analysis results to avoid analytical bias problems that may occur from a single model analysis perspective, making the prediction results more consistent with reality.

[0028] To further improve the feasibility of the recommended content, this embodiment further analyzes and recommends data within a specific area. For example, if the radius of the area is 5 kilometers, the system gives the courier's surrounding area as approximately 78.5 square kilometers, which is actually a very large business scope, containing multiple potential customers. During normal business development, it's difficult to visit all potential customers within a day. Therefore, to improve business development efficiency and reduce travel time, this application further calculates a locally optimal development range within the area. This locally optimal development range has a comprehensive score, which is obtained by comprehensively considering the number of customers and the final recommendation score for each customer. After recommending highly feasible business development areas, to further improve business development efficiency, this embodiment also recommends customer visit routes. These routes are calculated with the goal of minimizing distance traveled, reducing the courier's travel time.

[0029] When displaying relevant data on the system, this invention shows the established recommendation list, follow-up plan, and locally optimal mining range on the surrounding customer distribution map. This allows delivery personnel to clearly understand the progress of potential customer mining and to conduct customer mining in a planned manner.

[0030] For acquiring potential customer information, data can be collected through courier uploads to the system. For example, a multi-entry design enables efficient data entry, ensuring couriers can easily integrate customer resources from various channels. Not only are prominent "Add to Customer List" buttons placed on the search results page for potential customers, the competitor's customer acquisition page, and the big data recommendation page, allowing couriers to easily synchronize customer information to the app's customer management system, but a separate manual entry function is also provided in the customer management module. This facilitates recording customer information acquired offline through visits and communications, preventing information omissions. Required fields for data entry include customer name, contact information, address, and customer type, ensuring the completeness of core information. Optional fields cover industry type, company size, shipping frequency, and remarks, meeting diverse information management needs. Furthermore, to ensure accuracy, key contact methods such as mobile phone numbers and landlines undergo real-time format verification. If an incorrect format is entered, a prompt will immediately appear to guide the courier to correct it. For already entered customer information, flexible editing and deletion operations are supported: After the courier clicks on the target customer in the customer management list to enter the details page, they can add or modify the information through the edit button. After the modification is completed and the courier clicks save, the system will pop up a window to indicate that the customer information has been updated successfully, ensuring that the operation is perceptible. The deletion function is equipped with double protection. After clicking delete on the details page or list page, the deletion operation will only be executed after the courier confirms it, effectively preventing information loss due to accidental deletion, and comprehensively ensuring the convenience, accuracy and security of customer information management.

[0031] Please see Figure 2 The second embodiment of the method for couriers to identify potential customers in this invention includes: S201. Obtain the real-time location of the courier, and use a distance algorithm to calculate the area that the courier can quickly reach based on the courier's real-time location and a preset distance range; S202. Using a matching algorithm, customers within the specified area are selected from the courier company's internal customer data to obtain surrounding customer data; S203. Map the surrounding customer data to the area that the courier can quickly reach, and output the surrounding customer distribution map corresponding to the area.

[0032] In this embodiment, the Haversine distance algorithm can be used to filter customers near the courier by combining the courier company's internal customer data. The Haversine distance algorithm is a classic algorithm for calculating the shortest distance (great circle distance) between two points on the Earth's surface. Its core is to derive the distance based on the principle of spherical trigonometry using the latitude and longitude coordinates of the two points.

[0033] Specifically, assuming a courier named Li's real-time location is determined by GPS as 30.25°N, 120.18°E, and the preset "fast arrival distance range" is 5 kilometers, the boundary of a circular area is calculated using the spherical distance formula with a radius of 5 kilometers. Then, a spatial coordinate matching algorithm is used to calculate the Haversine distance between each customer's coordinates and Li's real-time location, and the surrounding customer data is obtained after filtering. The latitude and longitude of the filtered customer data are mapped onto the electronic map corresponding to the calculated "5-kilometer circular area." Further adjustments and more accurate positioning of customers on the electronic map can be made using their address information.

[0034] By following the steps above, the boundaries of the area can be defined quantitatively and accurately, ensuring that all customers who are easily accessible in the surrounding area are included, without omitting potential customers or including invalid customers who are beyond our capacity.

[0035] Please see Figure 3 The third embodiment of the method for couriers to identify potential customers in this invention includes: S301. Input the surrounding customer data into the pre-built recommendation sub-model, calculate the cosine similarity between the feature vector of each customer and the courier's preference vector, and obtain the first score; S302. Input the surrounding customer data into the collaborative filtering sub-model, calculate the success rate of the courier in developing each customer, and obtain the second score; S303. The final recommended score is calculated based on the first and second scores through weighted fusion.

[0036] In this embodiment, a big data recommendation model can be constructed based on the historical business data of couriers, customer behavior data, and market industry data collected by the APP. The model analyzes factors such as the courier's business preferences, the characteristics of successfully acquired customers, and the market demand in the surrounding area to provide personalized recommendations of potential customers to the courier. A big data recommendation section is set up on the potential customer mining page to display the potential customers recommended by the system to the courier. Each recommended customer displays basic information along with the reason for the recommendation and a matching score (0-100 points). A higher matching score indicates a greater fit between the customer and the courier's business needs. Couriers can then conduct targeted customer expansion work based on the recommendation results.

[0037] Big data hybrid recommendation can specifically include: feature engineering, content-based recommendation sub-model, collaborative filtering sub-model, weighted fusion, and feedback optimization. Among them, the content-based recommendation sub-model calculates the cosine similarity between the customer feature vector and the courier preference vector; the collaborative filtering sub-model selects similar couriers and calculates the customer development success rate through the Pearson correlation coefficient; the weighted fusion combines the scores of the two sub-models into the final recommendation score through dynamic weights; and the feedback optimization adjusts the weights through gradient descent to minimize the mean square error between the recommendation score and the feedback label.

[0038] Specifically, in feature engineering, the features of both the client and the courier are first standardized to provide usable vectors for subsequent models. This includes categorical feature encoding, numerical feature normalization, and final vector construction. Traditional single models (such as those relying solely on experience) cannot quantify the feature matching degree between customers and couriers (e.g., the matching degree between "e-commerce customers and couriers skilled in e-commerce"). This step, however, uses cosine similarity to accurately measure the matching degree. Even for new customers, as long as they have features such as industry and shipment volume, a first score can be calculated, effectively solving the "cold start problem for new customers." The collaborative filtering sub-model uses Pearson correlation coefficient to associate the experience of similar customers, expanding "individual experience" into "group experience," making success rate prediction more aligned with actual business scenarios. Furthermore, single models have inherent biases. Weighted fusion uses dynamic weights (e.g., W1=0.55, W2=0.45) to balance the advantages of the two types of models, ensuring that the final score reflects both feature matching degree and incorporates historical success rate experience. In addition, feedback optimization continuously adjusts weights through gradient descent, transforming actual development results into the basis for model iteration. For example, when customer A is successfully identified, the model adjusts weights by reducing the mean squared error, resulting in more accurate recommendation scores for similar e-commerce customers, thus achieving a virtuous cycle.

[0039] Please see Figure 4 The fourth embodiment of the method for couriers to identify potential customers in this invention includes: S401. Obtain the courier's service records, which include service type, service industry, and parcel volume. Clean and preprocess the courier's service records and surrounding customer data to obtain courier-side data and potential customer data. S402. Transform the data from the courier end and the data from potential clients into a numerical format that the model can calculate, and keep the vector dimensions of the clients and couriers consistent, so as to construct the feature vector of each customer and the preference vector of the courier. S403. Using the recommendation sub-model, the similarity between a single customer vector and a single courier vector is calculated using the cosine similarity formula to obtain the first score.

[0040] In this embodiment, by cleaning and preprocessing the courier's service records and surrounding customer data, interfering data is removed to ensure accurate courier capability assessment. Simultaneously, core information is aggregated to reduce subsequent processing complexity. For example, service records are aggregated into core data composed of service type of expertise, main industry, and average shipment volume, while irrelevant fields such as "contact information" are removed, allowing subsequent vector construction to focus on only three key dimensions. Then, "non-numerical features" are converted into "numerical codes," making features computable through encoding (e.g., door-to-door pickup = 1, e-commerce = 1), transforming textual features into numerical values ​​that the model can recognize. Furthermore, normalization can prevent large shipment volume values ​​from masking the influence of other features, and subsequent dimension unification ensures fair feature comparison. For example, both customer and courier vectors are fixed to three dimensions: "[service type code, industry code, and normalized shipment volume value]", ensuring that similarity can be directly calculated between the two. Finally, by calculating the cosine similarity, the "matching degree between customer and courier" can be transformed into an intuitive score (such as customer A=1.0, customer B=0.93, customer C=0.76), so that the courier can clearly know that customer A is the most compatible with him.

[0041] Please see Figure 5 The fifth embodiment of the method for couriers to identify potential customers in this invention includes: S501. Obtain customer development records from multiple couriers, wherein the customer development records include basic customer information and development results; S502. Combine surrounding customer data with the customer development records of couriers and clean the data. After cleaning, convert it into a matrix format that can be computed through collaborative filtering. S503. A collaborative filtering sub-model is used to screen similar customers and calculate the customer development success rate through the Pearson correlation coefficient to obtain the second score.

[0042] In this embodiment, by collecting customer development records from multiple couriers, a richer range of customer types (such as customers from different industries and with different shipment volumes) and development results can be covered, expanding the data scope and encompassing more customer scenarios. The customer development records must include the development results (e.g., customer A was successfully developed by Xiao Li, customer D was unsuccessfully developed by Xiao Wang). This is the core of collaborative filtering; without the "whether development was successful" label, it cannot be used for subsequent judgments. The "development result" label or data directly links "customer characteristics" and "success probability," providing real data support for subsequent predictions.

[0043] Next, by combining surrounding customer data with the deliveryman's customer development records, data from different sources are correlated, leading to a more comprehensive assessment. After data cleaning, the data format is ensured to be standardized and noise-free, allowing for immediate conversion into a "deliveryman-customer" interaction matrix. By using Pearson correlation coefficients to filter for "customers similar to the target customer," and then examining the success rate of developing these similar customers, even if the deliveryman has never worked with similar clients before, the success rate can be judged based on "group validation patterns," avoiding blind development that could lead to failure.

[0044] The first score (cosine similarity) is calculated based solely on the "feature matching degree between the customer and the courier," but it does not consider the actual success rate of developing such highly matched customers. The second score (development success rate) directly quantifies the probability of successfully developing such customers, complementing the first score and ensuring that the final recommendation score reflects both the feature matching degree and historical success experience, making it closer to the actual development results.

[0045] Please see Figure 6 The sixth embodiment of the method for couriers to identify potential customers in this invention includes: S601. Based on the geographical location of surrounding customer data, each customer is labeled within a region and associated with the final recommendation score of each customer; S602. Construct a sliding window, and set the shape and sliding step size of the sliding window; S603. Multiple local regions are formed within the area by sliding windows. According to the calculation rules, the recommended total score is calculated window by window. The multiple local regions are sorted according to the score to obtain the local optimal mining range. S604. Using the weighted TSP algorithm, a recommended customer visit route is calculated within the locally optimal mining range based on the customer's geographical location and final recommendation score.

[0046] In this embodiment, the system allows couriers to directly mark customers within the area of ​​the electronic map, enabling them to see the location of each customer simply by opening the system, eliminating the need to manually search the map using address lists. To improve business visit efficiency, this embodiment employs a sliding window, whose shape and step size can be flexibly configured to meet different needs. By calculating the total score and sorting to find the optimal range using the sliding window, the problem of couriers blindly running around large areas can be solved, helping them focus on high-value, smaller areas and improving resource utilization. Specifically, when calculating the total score window by window, after dividing the area into multiple smaller regions, the system calculates the sum of the final recommended scores for all customers within each window. For example, the total score for window 1 = customer A + customer D = 1.59, and the total score for window 2 = customer B + customer E = 1.14. The total score directly measures which smaller region is more worthwhile to focus on, avoiding the problem of only considering the number of customers without considering their quality. Other improved calculation rules can also limit the number of customers calculated, for example, only calculating the total score of the top 5 customers within a local area. Because when the customer density in certain areas is too high, even if the customer scores are low, their total score will still be high. However, limiting the calculation to the total score of the five highest-scoring customers in a local area can better reflect the development value of that area.

[0047] Please see Figure 7 The seventh embodiment of the method for couriers to identify potential customers in this invention includes: S701. Build a recommendation table based on the final recommendation score, and sort customers according to the size of the final recommendation score; S702. Generate a follow-up plan based on the customer ranking in the recommendation table. The follow-up plan includes the follow-up time and follow-up method. Output the follow-up plan to the courier to obtain follow-up data from the courier. S703. Based on the follow-up data provided by the courier, generate a follow-up analysis report, display the mining recommendation list, follow-up plan, follow-up analysis report, and local optimal mining range on the surrounding customer distribution map, and output the surrounding customer distribution map to the courier.

[0048] To help couriers quickly find the most valuable potential customers, this example displays search results in a list format. Each result includes information such as customer name, address, contact information, customer type, industry type, and estimated shipment volume. For enterprise customers, additional information such as company size and main business can be displayed. The list supports sorting by fields such as distance and estimated shipment volume.

[0049] Setting up follow-up plans allows couriers to better manage their time and provides system reminders. In practical applications, follow-up plan settings can be accessed via a button on the page. The pop-up settings window initially displays potential customers sorted according to the customer ranking in the recommendation table. The window also allows couriers to configure key information as needed: they can choose the follow-up time, method, and purpose, and record key follow-up points in detail through the remarks column. It also supports creating multiple follow-up plans at once, and the system automatically generates a structured follow-up task list based on the set time order, helping couriers prioritize their work. To avoid missing follow-up opportunities, each plan in this embodiment is equipped with a flexible reminder mechanism: reminder methods include in-app push notifications and SMS reminders, and the reminder time can be selected from options such as 15 minutes, 30 minutes, or 1 hour before the follow-up time, ensuring that couriers can prepare for communication in advance. Regarding the completion of follow-up records, the entry design balances timeliness and flexibility. When the reminder time arrives, the courier can directly access the completion page by clicking the reminder message. If additional follow-up information is needed, it can also be manually added in the follow-up record section of the customer details page. The record content must include the follow-up time, follow-up method, follow-up result, communication content, and next steps.

[0050] To facilitate subsequent data analysis, this embodiment will also generate follow-up analysis reports from the feedback data, comprehensively displaying various follow-up work indicators for couriers. These include: daily / weekly / monthly statistics on the total number of follow-up tasks, detailed data on completed follow-up tasks, categorized statistics of incomplete follow-up tasks, the specific number of successfully followed-up customers, and detailed records of failed follow-up customers. Simultaneously, the system supports multi-dimensional intelligent analysis functions, enabling cross-statistics and in-depth analysis based on multiple key dimensions such as follow-up date, time period distribution, customer industry classification, and customer level. Data comparisons visually demonstrate the differences in follow-up effectiveness across different time periods and customer types, providing data support for management decisions. Advanced chart visualization technology is employed to transform complex follow-up data into intuitive and easy-to-understand graphical displays. This includes: using comparative bar charts to show the comparison of follow-up task completion across different time periods; using trend line charts to dynamically display the phased changes in the number of successfully followed-up customers; using pie charts to clearly present the follow-up percentage distribution for various customer types; and introducing heatmaps to analyze the distribution characteristics of follow-up time periods. With these professional data visualization tools, couriers and managers can clearly grasp the overall effectiveness of follow-up work, promptly identify weaknesses and potential problems in the work, and thus optimize follow-up strategies and work methods in a targeted manner.

[0051] The above describes the method for supporting couriers in identifying potential customers in embodiments of the present invention. The following describes the device for supporting couriers in identifying potential customers in embodiments of the present invention. Please refer to [link / reference]. Figure 8One embodiment of the device supporting couriers in identifying potential customers in this invention includes: The filtering module 10 is used to obtain the real-time location of the courier, calculate the area range that the courier can quickly reach based on the real-time location of the courier, filter out the surrounding customers of the current courier from the customer data of the courier company based on the area range, obtain the surrounding customer data, and construct the surrounding customer distribution map. The calculation module 20 is used to input surrounding customer data into a pre-built big data hybrid recommendation model to obtain the final recommendation score for each customer. The big data hybrid recommendation model includes a recommendation sub-model and a collaborative filtering sub-model. The regional recommendation module 30 is used to calculate the local optimal mining range within the regional range based on the regional range, the geographical location of surrounding customer data and the final recommendation score of each customer, and to calculate the recommended customer visit route within the local optimal mining range. The display module 40 is used to create a mining recommendation table based on the final recommendation score, generate a follow-up plan based on the mining recommendation table, and display the mining recommendation list, follow-up plan, and local optimal mining range on the surrounding customer distribution map.

[0052] Please see Figure 9 One embodiment of the device supporting couriers in identifying potential customers in this invention includes: The filtering module 10 is used to obtain the real-time location of the courier, calculate the area range that the courier can quickly reach based on the real-time location of the courier, filter out the surrounding customers of the current courier from the customer data of the courier company based on the area range, obtain the surrounding customer data, and construct the surrounding customer distribution map. The calculation module 20 is used to input surrounding customer data into a pre-built big data hybrid recommendation model to obtain the final recommendation score for each customer. The big data hybrid recommendation model includes a recommendation sub-model and a collaborative filtering sub-model. The regional recommendation module 30 is used to calculate the local optimal mining range within the regional range based on the regional range, the geographical location of surrounding customer data and the final recommendation score of each customer, and to calculate the recommended customer visit route within the local optimal mining range. The display module 40 is used to create a mining recommendation table based on the final recommendation score, generate a follow-up plan based on the mining recommendation table, and display the mining recommendation list, follow-up plan, and local optimal mining range on the surrounding customer distribution map. In this embodiment, the filtering module 10 includes: The range calculation unit 11 is used to obtain the real-time location of the courier and, using a distance algorithm, calculate the area range that the courier can quickly reach based on the real-time location of the courier and a preset distance range. The filtering unit 12 is used to use a matching algorithm to filter out customers within the area from the customer data inside the courier company, and obtain surrounding customer data. Distribution unit 13 is used to map surrounding customer data to an area that the courier can quickly reach, and output a surrounding customer distribution map corresponding to the area. In this embodiment, the computing module 20 includes: The first calculation unit 21 is used to input surrounding customer data into a pre-built recommendation sub-model, calculate the cosine similarity between the feature vector of each customer and the courier's preference vector, and obtain the first score. The second calculation unit 22 is used to input surrounding customer data into the collaborative filtering sub-model, calculate the success rate of the courier in developing each customer, and obtain the second score. The weighted calculation unit 23 is used to calculate the final recommendation score based on the first score and the second score through weighted fusion. In this embodiment, the first computing unit 21 includes: The first acquisition subunit 211 is used to acquire the courier's service records, which include service type, service industry and parcel volume, and to clean and preprocess the courier's service records and surrounding customer data to obtain courier-end data and potential client data. The first transformation subunit 212 is used to uniformly transform the data of the courier end and the data of potential customers into a numerical format that the model can calculate, and to keep the vector dimensions of the client end and the courier end consistent, so as to construct the feature vector of each customer and the courier preference vector. The first calculation subunit 213 is used to use the recommendation submodel to calculate the similarity between a single customer vector and a single courier vector using the cosine similarity formula, and obtain the first score. In this embodiment, the second computing unit 22 includes: The second acquisition subunit 221 is used to acquire customer development records of multiple couriers, wherein the customer development records include basic customer information and development results. The second transformation subunit 222 is used to combine the surrounding customer data with the customer development records of the courier and clean the data, and then transform it into a matrix format that can be computed by collaborative filtering. The second calculation subunit 223 is used to use a collaborative filtering sub-model to screen similar customers through the Pearson correlation coefficient and calculate the customer development success rate to obtain a second score. In this embodiment, the region recommendation module 30 includes: Labeling unit 31 is used to label each customer within a region based on the geographical location of surrounding customer data and associate each customer's final recommendation score; Window building unit 32 is used to build a sliding window and set the shape and sliding step of the sliding window; Local calculation unit 33 is used to form multiple local regions within the area by sliding window, calculate the recommended total score window by window according to the calculation rules, sort the multiple local regions according to the score size, and obtain the local optimal mining range; The route generation unit 34 is used to calculate the recommended customer visit route within the local optimal mining range using the weighted TSP algorithm, based on the customer's geographical location and the final recommendation score. In this embodiment, the display module 40 includes: The sorting unit 41 is used to build a mining recommendation table based on the final recommendation score and sort customers according to the size of the final recommendation score; Planning unit 42 is used to generate a follow-up plan based on the customer ranking in the recommendation table. The follow-up plan includes follow-up time and follow-up method. The follow-up plan is output to the courier to obtain follow-up data fed back by the courier. Display unit 43 is used to generate follow-up analysis reports based on the follow-up data provided by the courier. It displays the mining recommendation list, follow-up plan, follow-up analysis report, and local optimal mining range on the surrounding customer distribution map, and outputs the surrounding customer distribution map to the courier.

[0053] The device for enabling couriers to discover potential customers, as described in this invention, first acquires the courier's geographical location in real time and, combined with a preset distance threshold, quantifies and calculates the courier's effective service area using a spatial distance algorithm. Then, based on this area, it filters eligible potential customers from the courier company's internal customer database and integrates this customer data with regional geographic information to construct a visualized distribution map of surrounding customers. On this basis, the filtered surrounding customer data is input into a pre-trained big data hybrid recommendation model, which includes a content-based recommendation sub-model and a collaborative filtering sub-model. The recommendation sub-model calculates the matching degree between customer feature vectors and courier preference vectors using cosine similarity, while the collaborative filtering sub-model uses Pearson correlation coefficients to mine historical development patterns of similar customers. The results of both are dynamically weighted and fused to output a final recommendation score for each potential customer. This model quantifies the compatibility and success rate of customer-courier cooperation in a data-driven manner, significantly improving the accuracy of courier business expansion and avoiding blindly pursuing low-value customers. Subsequently, combining the aforementioned effective service area, customer geographic location, and final recommendation score, a sliding window algorithm is used to divide the broad service area into multiple local regions. The sum of customer recommendation scores is calculated and sorted window by window to determine the locally optimal mining range. This significantly reduces the randomness of business expansion, allowing resources to be concentrated on areas with higher output efficiency and improving implementation feasibility. Simultaneously, for customers within the locally optimal mining range, a weighted TSP algorithm is used to plan recommended visit routes, reducing ineffective courier trips and time waste, avoiding detours and missed visits, and significantly increasing daily customer visit volume and time utilization. Finally, a structured mining recommendation table is generated based on the customer's final recommendation score, and a phased follow-up plan is automatically generated in conjunction with the recommendation table. Couriers can grasp customer value, regional priorities, and implementation steps in one stop without switching between multiple systems, providing a complete technical solution from data support to actionable implementation for efficiently mining potential customers.

[0054] The above describes in detail the device for supporting couriers to discover potential customers in the embodiments of the present invention from the perspective of modular functional entities. The following describes in detail the device for supporting couriers to discover potential customers in the embodiments of the present invention from the perspective of hardware processing.

[0055] Figure 10This is a schematic diagram of a device 900 for supporting couriers in identifying potential customers, provided as an embodiment of the present invention. This device 900 can vary significantly due to different configurations or performance characteristics. It may include one or more central processing units (CPUs) 910 (e.g., one or more processors) and a memory 920, and one or more storage media 930 (e.g., one or more mass storage devices) for storing applications 933 or data 932. The memory 920 and storage media 930 can be temporary or persistent storage. The program stored in the storage media 930 may include one or more modules (not shown in the diagram), each module including a series of instruction operations on the device 900. Furthermore, the processor 910 may be configured to communicate with the storage media 930 and execute the series of instruction operations in the storage media 930 on the device 900 to implement the steps of the methods for supporting couriers in identifying potential customers provided in the above-described embodiments.

[0056] The device 900 supporting delivery personnel in identifying potential customers may also include one or more power supplies 940, one or more wired or wireless network interfaces 950, one or more input / output interfaces 960, and / or one or more operating systems 931, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. Those skilled in the art will understand that... Figure 10 The illustrated structure of the device for supporting couriers in identifying potential customers does not constitute a limitation on the device itself, and may include more or fewer components than illustrated, or combine certain components, or have different component arrangements.

[0057] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when executed on a computer, cause the computer to perform steps supporting a method for couriers to identify potential customers.

[0058] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the device or apparatus described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0059] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0060] It is understood that those skilled in the art can make equivalent substitutions or modifications to the technical solution and inventive concept of the present invention, and all such substitutions or modifications should fall within the protection scope of the appended claims.

Claims

1. A method for supporting delivery personnel in identifying potential customers, characterized in that, Includes the following steps: Get the real-time location of the courier, calculate the area that the courier can quickly reach based on the real-time location, filter out the surrounding customers of the current courier from the customer data of the courier company based on the area, obtain the surrounding customer data, and build a surrounding customer distribution map. The surrounding customer data is input into a pre-built big data hybrid recommendation model to obtain the final recommendation score for each customer. The big data hybrid recommendation model includes a recommendation sub-model and a collaborative filtering sub-model. Based on the aforementioned regional scope, the geographical location of surrounding customer data, and the final recommendation score for each customer, the local optimal mining range within the regional scope is calculated, and the recommended customer visit route within the local optimal mining range is calculated. A mining recommendation table is created based on the final recommendation score, and a follow-up plan is generated based on the mining recommendation table. The mining recommendation list, follow-up plan, and locally optimal mining range are displayed on the surrounding customer distribution map.

2. The method for supporting couriers in identifying potential customers according to claim 1, characterized in that, The process of obtaining the courier's real-time location, calculating the area the courier can quickly reach based on the courier's real-time location, filtering out nearby customers from the courier company's internal customer data based on the area, obtaining nearby customer data, and constructing a nearby customer distribution map specifically includes: The system obtains the real-time location of the courier and uses a distance algorithm to calculate the area that the courier can quickly reach based on the courier's real-time location and a preset distance range. A matching algorithm is used to filter customers within the specified area from the courier company's internal customer data to obtain surrounding customer data; The system maps surrounding customer data to areas that delivery personnel can quickly reach, and outputs a distribution map of surrounding customers corresponding to those areas.

3. The method for supporting couriers in identifying potential customers according to claim 1, characterized in that, The process involves inputting surrounding customer data into a pre-built big data hybrid recommendation model to obtain a final recommendation score for each customer. This big data hybrid recommendation model includes a recommendation sub-model and a collaborative filtering sub-model, specifically comprising: The surrounding customer data is input into the pre-built recommendation sub-model, and the cosine similarity between the feature vector of each customer and the courier's preference vector is calculated to obtain the first score; Input the surrounding customer data into the collaborative filtering sub-model to calculate the success rate of the courier in developing each customer and obtain the second score; The final recommended score is calculated by weighted fusion based on the first and second scores.

4. The method for supporting couriers in identifying potential customers according to claim 3, characterized in that, The step of inputting surrounding customer data into a pre-built recommendation sub-model, calculating the cosine similarity between each customer's feature vector and the courier's preference vector, and obtaining a first score specifically includes: The service records of couriers are obtained, including service type, service industry and parcel volume. The service records of couriers and surrounding customer data are cleaned and preprocessed to obtain courier-side data and potential customer data. The data from both the courier and potential customer sides are uniformly converted into a numerical format that the model can compute, and the vector dimensions of the customer and courier sides are kept consistent in order to construct the feature vectors of each customer and the preference vectors of the courier. The recommendation sub-model is used to calculate the similarity between a single customer vector and a single courier vector using the cosine similarity formula, and obtain the first score.

5. The method for supporting couriers in identifying potential customers according to claim 3, characterized in that, The step of inputting surrounding customer data into the collaborative filtering sub-model to calculate the success rate of the courier in developing each customer and obtain a second score specifically includes: Obtain customer development records from multiple couriers, including basic customer information and development results; The data is combined with the customer development records of couriers and cleaned. After cleaning, it is transformed into a matrix format that can be computed through collaborative filtering. A collaborative filtering sub-model is used to screen similar customers and calculate the customer development success rate by using the Pearson correlation coefficient to obtain the second score.

6. The method for supporting couriers in identifying potential customers according to claim 1, characterized in that, The process involves calculating the locally optimal mining range within the region based on the geographical location of surrounding customer data and the final recommendation score for each customer, and then calculating the recommended customer visit route within that locally optimal mining range. Specifically, this includes: Based on the geographical location of surrounding customers, each customer is tagged within a region and associated with their final recommendation score; Create a sliding window and set its shape and sliding step size; Multiple local regions are formed within the area by sliding windows. According to the calculation rules, the recommended total score is calculated window by window. The multiple local regions are sorted according to the score to obtain the local optimal mining range. The weighted TSP algorithm is used to calculate the recommended customer visit route within the local optimal mining range based on the customer's geographical location and final recommendation score.

7. The method for supporting couriers in identifying potential customers according to claim 1, characterized in that, The process of establishing a mining recommendation table based on the final recommendation score, generating a follow-up plan based on the mining recommendation table, and displaying the mining recommendation list, follow-up plan, and locally optimal mining range on the surrounding customer distribution map specifically includes: A recommendation table is created based on the final recommendation score, and customers are ranked according to the size of the final recommendation score. A follow-up plan is generated based on the customer ranking in the recommendation table. The follow-up plan includes the follow-up time and follow-up method. The follow-up plan is then output to the courier to obtain follow-up data from the courier. Based on the follow-up data provided by the couriers, a follow-up analysis report is generated. The recommended mining list, follow-up plan, follow-up analysis report, and local optimal mining range are displayed on the surrounding customer distribution map, and the surrounding customer distribution map is output to the couriers.

8. A device for supporting couriers in identifying potential customers, characterized in that, include: The filtering module is used to obtain the real-time location of the courier, calculate the area that the courier can quickly reach based on the real-time location, filter out the surrounding customers of the current courier from the customer data of the courier company based on the area, obtain surrounding customer data, and build a surrounding customer distribution map. The calculation module is used to input surrounding customer data into a pre-built big data hybrid recommendation model to obtain the final recommendation score for each customer. The big data hybrid recommendation model includes a recommendation sub-model and a collaborative filtering sub-model. The regional recommendation module is used to calculate the local optimal mining range within the regional range based on the regional range, the geographical location of surrounding customer data, and the final recommendation score of each customer, and to calculate the recommended customer visit route within the local optimal mining range. The display module is used to create a mining recommendation table based on the final recommendation score, generate a follow-up plan based on the mining recommendation table, and display the mining recommendation list, follow-up plan, and local optimal mining range on the surrounding customer distribution map.

9. A device for supporting couriers in identifying potential customers, characterized in that, It includes a memory and at least one processor, wherein the memory stores computer-readable instructions; The at least one processor invokes the computer-readable instructions in the memory to perform the steps of the method for supporting couriers to discover potential customers as described in any one of claims 1-7.

10. A computer-readable storage medium storing computer-readable instructions thereon, characterized in that, When the computer-readable instructions are executed by a processor, they implement the steps of the method for supporting couriers to discover potential customers as described in any one of claims 1-7.

Images