A multi-source whitelist-based intelligent routing method and system
By acquiring multi-source heterogeneous datasets and performing structured and labeled processing to generate multi-dimensional feature vectors, the problem of accurately and intelligently matching customer groups in existing technologies has been solved. This has enabled efficient parallel scheduling and load balancing, improving the success rate and return on investment of outbound call tasks.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- JINGFAYUN DIGITAL TECHNOLOGY (JIANGXI) CO LTD
- Filing Date
- 2026-03-24
- Publication Date
- 2026-06-09
Smart Images
Figure CN122179511A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of intelligent communication technology, and in particular to an intelligent routing method and system based on a multi-source whitelist. Background Technology
[0002] With the development of artificial intelligence technology, AI outbound calling systems are being used more and more widely in scenarios such as telemarketing and customer follow-up.
[0003] Existing outbound calling systems typically rely on multiple line providers to execute call tasks. However, in actual operation, customer groups have multi-dimensional attribute differences, such as industry (e.g., tobacco retailers, e-commerce merchants), region, and purchasing power. Different outbound calling lines show significant differences in connection rates and cost performance among different customer groups with different attributes.
[0004] Currently, most outbound calling systems rely on manual experience to assign routes to specific lists for routing selection. This approach has several drawbacks: 1. Inability to accurately and intelligently match customer groups: It cannot select the optimal route for a given customer group based on its precise attributes (e.g., tobacco merchants, e-commerce businesses), resulting in low overall outbound calling success rates and high marketing costs; 2. Data silos: Customer lists are often multi-sourced, heterogeneous, and managed in a decentralized manner, creating data silos that hinder unified analysis and value extraction; 3. Low efficiency: Route selection heavily relies on individual experience, leading to slow response times. Furthermore, when managing multiple partners, efficient parallel scheduling and load balancing are difficult, resulting in high operational complexity and poor stability. Summary of the Invention
[0005] To address the shortcomings of existing technologies, the present invention aims to provide an intelligent routing method and system based on a multi-source whitelist. This method addresses the technical problems of existing outbound call routing methods that rely on manual experience to assign routes to specific lists, which cannot accurately and intelligently match customer groups, have data barriers between multiple data sources, and are inefficient.
[0006] To achieve the above objectives, in a first aspect, embodiments of this application provide an intelligent routing method based on a multi-source whitelist, comprising the following steps: A multi-source heterogeneous dataset including several customer whitelist data is obtained, and the multi-source heterogeneous dataset is structured to obtain a multi-source whitelist dataset. The multi-source whitelist dataset is tagged to generate a multi-dimensional feature vector corresponding to each customer. The multi-dimensional feature vector includes industry attribute tags, regional feature tags, and consumption level tags. Based on the multidimensional feature vector, the optimal routing path is calculated for each customer, and the outbound call task is distributed to the outbound call partner according to the optimal routing path, and the task execution result of the outbound call partner is received.
[0007] Furthermore, the step of performing structured processing on the multi-source heterogeneous dataset to obtain a multi-source whitelist dataset includes: Data cleaning is performed on several of the aforementioned customer whitelist data to obtain comprehensive whitelist data; The multi-source whitelist dataset is extracted from the comprehensive whitelist data based on the mapped field set.
[0008] Furthermore, the mapping field set includes a customer identifier field, an industry type field, a geographic location field, and a historical consumption amount field. The multi-source whitelist dataset includes several mapping value sets corresponding to customers, and the mapping value sets include customer identifier codes, industry type codes, geographic location codes, and historical consumption values.
[0009] Furthermore, the step of tagging the multi-source whitelist dataset to generate a multi-dimensional feature vector corresponding to each customer, wherein the multi-dimensional feature vector includes industry attribute tags, regional feature tags, and consumption level tags, includes: Based on the industry type code, generate industry attribute tags corresponding to the customer; The administrative region code is extracted from the geographic location code, and the administrative region code is compared with the regional economic level comparison table to generate a regional feature label corresponding to the customer. The historical consumption values are converted into consumption level labels using a binning algorithm; The industry attribute label, the regional feature label, and the consumption level label are vectorized and concatenated into a multi-dimensional feature vector.
[0010] Furthermore, the step of calculating the optimal routing path for each customer based on the multidimensional feature vector includes: Construct a rule management library, which includes several combinations of judgment conditions and a set of candidate routing paths corresponding to the combinations of judgment conditions; The multidimensional feature vector is compared with the judgment conditions to select a target route path set from a plurality of candidate route path sets, wherein the target route path set includes a plurality of candidate route paths; Obtain the priority score of the candidate routing paths, and select the candidate routing path with the highest priority score as the optimal routing path.
[0011] Furthermore, the combination of judgment conditions includes industry condition labels, regional condition labels, and consumption condition labels.
[0012] Furthermore, the formula for obtaining the priority score is: , in, This represents the priority score of the i-th candidate route. This represents the real-time call cost of the i-th candidate routing path. This represents the sum of real-time call costs for all candidate routing paths. This represents the historical success rate of the i-th candidate routing path for customers with the same multidimensional feature vector. This represents the current availability status of the i-th candidate route path, and The value can be 0 or 1. , , All of these represent weighting coefficients.
[0013] Secondly, embodiments of this application provide an intelligent routing system based on a multi-source whitelist, applied to the intelligent routing method based on a multi-source whitelist as described in the first aspect above, the system comprising: The processing module is used to acquire a multi-source heterogeneous dataset including several customer whitelist data, and to perform structured processing on the multi-source heterogeneous dataset to obtain a multi-source whitelist dataset. The conversion module is used to perform tagging processing on the multi-source whitelist dataset to generate a multi-dimensional feature vector corresponding to each customer. The multi-dimensional feature vector includes industry attribute tags, regional feature tags, and consumption level tags. The execution module is used to calculate the optimal routing path for each customer based on the multidimensional feature vector, distribute the outbound call task to the outbound call partner according to the optimal routing path, and receive the task execution results from the outbound call partner.
[0014] Thirdly, embodiments of this application provide a computer, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the intelligent routing method based on a multi-source whitelist as described in the first aspect above.
[0015] Fourthly, embodiments of this application provide a storage medium storing a computer program thereon, which, when executed by a processor, implements the intelligent routing method based on a multi-source whitelist as described in the first aspect above.
[0016] Compared with existing technologies, the beneficial effects of the present invention are as follows: by accessing and structuring the multi-source heterogeneous dataset, the scattered data is aggregated and unified, breaking down data barriers; by obtaining the multi-dimensional feature vector and obtaining the optimal routing path based on the multi-dimensional feature vector, on the basis of data unification, the line selection can be completed automatically and intelligently based on the customer's precise attributes, realizing efficient parallel scheduling and load balancing, replacing the inefficient manual line selection mode, and improving the overall success rate and return on investment of outbound call tasks. Attached Figure Description
[0017] Figure 1 This is a flowchart of the intelligent routing method based on a multi-source whitelist in the first embodiment of the present invention; Figure 2 This is a structural block diagram of the intelligent routing system based on a multi-source whitelist in the second embodiment of the present invention; The following detailed description, in conjunction with the accompanying drawings, will further illustrate the present invention. Detailed Implementation
[0018] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Several embodiments of the invention are illustrated in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.
[0019] It should be noted that when a component is said to be "fixed to" another component, it can be directly on the other component or there may be an intervening component. When a component is said to be "connected to" another component, it can be directly connected to the other component or there may be an intervening component. The terms "vertical," "horizontal," "left," "right," and similar expressions used in this document are for illustrative purposes only.
[0020] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.
[0021] Please see Figure 1 The intelligent routing method based on a multi-source whitelist provided in the first embodiment of the present invention includes the following steps: S10: Obtain a multi-source heterogeneous dataset including several customer whitelist data, and perform structured processing on the multi-source heterogeneous dataset to obtain a multi-source whitelist dataset; Understandably, the various customer whitelist data sets are collected from different data sources. In this embodiment, the multi-source heterogeneous dataset includes four customer whitelist data sets: a tobacco merchant whitelist, an e-commerce whitelist, a store whitelist, and an agricultural whitelist. Specifically, multiple data access interfaces are configured to acquire these various customer whitelist data sets.
[0022] Step S10 includes: S110: Perform data cleaning on several customer whitelist data to obtain comprehensive whitelist data; In this embodiment, the data cleaning includes duplicate data removal, invalid data filtering, and erroneous data correction; Specifically, the duplicate data removal refers to: within a single customer whitelist, deduplication is performed through precise matching and fuzzy matching. For example, within the tobacco merchant whitelist, duplicate data with identical mobile phone numbers are removed. Deduplication is also performed across several customer whitelists. For instance, a customer may appear in both the e-commerce whitelist and the store whitelist. By using preset matching rules (such as identical mobile phone numbers or identical "company name + legal representative"), overlapping duplicate records are identified and merged or marked according to the data source priority strategy (such as prioritizing e-commerce over stores) to prevent subsequent routing tasks from making duplicate outbound calls to the same customer.
[0023] The invalid data filtering refers to filtering data whose corresponding subsequent mapping fields are empty, and filtering illegal values defined by business logic, such as filtering out numbers that are not 11 digits or that start with invalid number segments such as "000" or "111"; filtering out obvious abnormal values such as negative historical consumption amounts.
[0024] The error data correction refers to: unifying the format of data that is inconsistent in format but valid in content, such as removing spaces, hyphens, parentheses, etc. from numbers; unifying the expression format of different dates; and removing spaces or garbled characters.
[0025] S120: Extract a multi-source whitelist dataset from the comprehensive whitelist data based on the mapping field set.
[0026] The mapping field set includes a customer identifier field, an industry type field, a geographic location field, and a historical consumption amount field. For each individual sub-data in the whitelist comprehensive data (each sub-data corresponds to one customer), a customer identifier code, an industry type code, a geographic location code, and a historical consumption value corresponding to the sub-data are generated, which is the mapping value set. Several mapping value sets are combined to form the multi-source whitelist dataset.
[0027] S20: The multi-source whitelist dataset is labeled to generate a multi-dimensional feature vector corresponding to each customer. The multi-dimensional feature vector includes industry attribute labels, regional feature labels, and consumption level labels. Step S20 includes: S210: Generate industry attribute tags corresponding to the customer based on the industry type code; A pre-defined tag library is constructed, which stores tags corresponding to different industry type codes. The obtained industry type codes are then matched with the tag library to generate industry attribute tags.
[0028] S220: Extract the administrative region code from the geographic location code, compare the administrative region code with the regional economic level comparison table, and generate a regional feature label corresponding to the customer. The geographic location code can be converted into a standard administrative region code through the map API, and then regional feature labels can be generated through the regional economic level comparison table (e.g., 110105 → "core area of first-tier city"; 330102 → "main urban area of second-tier city"; 422822 → "county economic zone" etc.).
[0029] S230: Convert the historical consumption values into consumption level labels using a binning algorithm; After sorting all historical consumption data, it is divided into several consecutive amount ranges. A consumption level label (such as Level1, Level2, etc.) is set for each amount range, and then the consumption level label corresponding to the amount range in which the historical consumption data is located is obtained.
[0030] S240: Vectorize and concatenate the industry attribute label, the regional feature label, and the consumption level label into a multi-dimensional feature vector; The industry attribute label, the regional feature label, and the consumption level label are respectively converted into numerical vectors through an embedding function, and then concatenated into the multidimensional feature vector.
[0031] S30: Calculate the optimal routing path for each customer based on the multidimensional feature vector, distribute the outbound call task to the outbound call partner according to the optimal routing path, and receive the task execution result from the outbound call partner; S310: Construct a rule management library, which includes several combinations of judgment conditions and a set of candidate routing paths corresponding to the combinations of judgment conditions; The judgment condition combination includes industry condition tags, regional condition tags, and consumption condition tags. The industry condition tags, regional condition tags, and consumption condition tags are substantially the same as the industry attribute tags, regional feature tags, and consumption level tags, respectively. Assuming the set of industry condition tags is {Tobacco, E-commerce, Store, Agriculture}, the set of regional condition tags is {Tier1_Core, Tier1_Suburb, Tier2_City, Rural}, and the set of consumption condition tags is {Level1, Level2, Level3, Level4}, then the combination of these three can form several judgment condition combinations, such as {Tobacco, Tier1_Core, Level1}. Each judgment condition combination corresponds to a different set of candidate routing paths, such as one set of candidate routing paths being {Route A, Route C, Route D}, where Route A, Route C, and Route D are all considered.
[0032] S320: Compare the multidimensional feature vector with the judgment condition combination to select a target route path set from a plurality of candidate route path sets, wherein the target route path set includes a plurality of candidate route paths; The candidate routing paths refer to lines A, C, and D in the aforementioned example. When the multidimensional feature vector is the same as a certain combination of judgment conditions, the set of candidate routing paths corresponding to that combination of judgment conditions is obtained.
[0033] S330: Obtain the priority score of the candidate routing paths, and select the candidate routing path with the highest priority score as the optimal routing path; The formula for obtaining the priority score is: , in, This represents the priority score of the i-th candidate route. This represents the real-time call cost of the i-th candidate routing path. This represents the sum of real-time call costs for all candidate routing paths. This represents the historical success rate of the i-th candidate routing path for customers with the same multidimensional feature vector. This represents the current availability status of the i-th candidate route path, and The value can be 0 or 1. , , All of these represent weighting coefficients. It should be noted that... , , The sum of is 1.
[0034] Real-time call cost can be obtained by calling the price query interface provided by each line provider or querying the real-time updated price list to obtain the unit cost of making an outbound call to the customer at the current moment; historical success rate can be obtained by calculating the cosine similarity between the current customer's multidimensional feature vector and the multidimensional feature vectors of previous customers in each historical record, and then extracting historical records with similarity exceeding the similarity threshold, and calculating the proportion of successful outbound calls in this part of the historical records; current availability status can be determined by the heartbeat detection or health check interface to determine the current status of the line. If the line is normal and schedulable, the value is 1; if the line is faulty, congested, or unavailable, the value is 0.
[0035] By accessing and structuring the multi-source heterogeneous dataset, the scattered data is aggregated and unified, breaking down data barriers. By obtaining the multi-dimensional feature vector and obtaining the optimal routing path based on the multi-dimensional feature vector, on the basis of data unification, the route selection can be completed automatically and intelligently based on the customer's precise attributes, realizing efficient parallel scheduling and load balancing, replacing the inefficient manual route selection mode, and improving the overall success rate and return on investment of outbound call tasks.
[0036] Please see Figure 2 The second embodiment of the present invention provides an intelligent routing system based on a multi-source whitelist. This system is applied to the intelligent routing method based on a multi-source whitelist described in the above embodiments, and will not be repeated hereafter. As used below, the terms "module," "unit," "subunit," etc., can refer to a combination of software and / or hardware that performs a predetermined function. Although the apparatus described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.
[0037] The system includes: Processing module 10 is used to acquire a multi-source heterogeneous dataset including several customer whitelist data, and to perform structured processing on the multi-source heterogeneous dataset to obtain a multi-source whitelist dataset. The processing module 10 includes: The first unit is used to perform data cleaning on several customer whitelist data to obtain comprehensive whitelist data. The second unit is used to extract a multi-source whitelist dataset from the comprehensive whitelist data based on the mapping field set. The conversion module 20 is used to perform tagging processing on the multi-source whitelist dataset to generate a multi-dimensional feature vector corresponding to each customer. The multi-dimensional feature vector includes industry attribute tags, regional feature tags, and consumption level tags. The conversion module 20 includes: The third unit is used to generate industry attribute tags corresponding to customers based on the industry type code; The fourth unit is used to extract the administrative region code from the geographic location code, compare the administrative region code with the regional economic level comparison table, and generate a regional feature label corresponding to the customer. The fifth unit is used to convert the historical consumption values into consumption level tags using a binning algorithm; The sixth unit is used to vectorize and concatenate the industry attribute label, the regional feature label, and the consumption level label into a multi-dimensional feature vector; The execution module 30 is used to calculate the optimal routing path for each customer based on the multidimensional feature vector, distribute the outbound call task to the outbound call partner according to the optimal routing path, and receive the task execution result from the outbound call partner. The execution module 30 includes: The seventh unit is used to construct a rule management library, which includes several combinations of judgment conditions and a set of candidate routing paths corresponding to the combinations of judgment conditions. The eighth unit is used to compare the multidimensional feature vector with the judgment condition combination to select a target route path set from a plurality of candidate route path sets, wherein the target route path set includes a plurality of candidate route paths; The ninth unit is used to obtain the priority score of the candidate routing paths and select the candidate routing path with the highest priority score as the optimal routing path.
[0038] The present invention also provides a computer, 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 intelligent routing method based on a multi-source whitelist as described in the above technical solutions.
[0039] The present invention also provides a storage medium storing a computer program thereon, which, when executed by a processor, implements the intelligent routing method based on a multi-source whitelist as described in the above technical solution.
[0040] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.
[0041] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the appended claims.
Claims
1. A smart routing method based on a multi-source whitelist, characterized in that, Includes the following steps: A multi-source heterogeneous dataset including several customer whitelist data is obtained, and the multi-source heterogeneous dataset is structured to obtain a multi-source whitelist dataset. The multi-source whitelist dataset is tagged to generate a multi-dimensional feature vector corresponding to each customer. The multi-dimensional feature vector includes industry attribute tags, regional feature tags, and consumption level tags. Based on the multidimensional feature vector, the optimal routing path is calculated for each customer, and the outbound call task is distributed to the outbound call partner according to the optimal routing path, and the task execution result of the outbound call partner is received.
2. The intelligent routing method based on multi-source whitelist according to claim 1, characterized in that, The step of performing structured processing on the multi-source heterogeneous dataset to obtain a multi-source whitelist dataset includes: Data cleaning is performed on several of the aforementioned customer whitelist data to obtain comprehensive whitelist data; The multi-source whitelist dataset is extracted from the comprehensive whitelist data based on the mapped field set.
3. The intelligent routing method based on multi-source whitelist according to claim 2, characterized in that, The mapping field set includes a customer identifier field, an industry type field, a geographic location field, and a historical consumption amount field. The multi-source whitelist dataset includes several mapping value sets corresponding to customers, and the mapping value sets include customer identifier codes, industry type codes, geographic location codes, and historical consumption values.
4. The intelligent routing method based on multi-source whitelist according to claim 3, characterized in that, The step of tagging the multi-source whitelist dataset to generate a multi-dimensional feature vector corresponding to each customer, wherein the multi-dimensional feature vector includes industry attribute tags, regional feature tags, and consumption level tags, includes: Based on the industry type code, generate industry attribute tags corresponding to the customer; The administrative region code is extracted from the geographic location code, and the administrative region code is compared with the regional economic level comparison table to generate a regional feature label corresponding to the customer. The historical consumption values are converted into consumption level labels using a binning algorithm; The industry attribute label, the regional feature label, and the consumption level label are vectorized and concatenated into a multi-dimensional feature vector.
5. The intelligent routing method based on multi-source whitelist according to claim 1, characterized in that, The step of calculating the optimal routing path for each customer based on the multidimensional feature vector includes: Construct a rule management library, which includes several combinations of judgment conditions and a set of candidate routing paths corresponding to the combinations of judgment conditions; The multidimensional feature vector is compared with the judgment conditions to select a target route path set from a plurality of candidate route path sets, wherein the target route path set includes a plurality of candidate route paths; Obtain the priority score of the candidate routing paths, and select the candidate routing path with the highest priority score as the optimal routing path.
6. The intelligent routing method based on multi-source whitelist according to claim 5, characterized in that, The combination of judgment conditions includes industry condition labels, regional condition labels, and consumption condition labels.
7. The intelligent routing method based on multi-source whitelist according to claim 5, characterized in that, The formula for obtaining the priority score is: ， in, This represents the priority score of the i-th candidate route. This represents the real-time call cost of the i-th candidate routing path. This represents the sum of real-time call costs for all candidate routing paths. This represents the historical success rate of the i-th candidate routing path for customers with the same multidimensional feature vector. This represents the current availability status of the i-th candidate route path, and The value can be 0 or 1. , , All of these represent weighting coefficients.
8. A smart routing system based on a multi-source whitelist, applied to the smart routing method based on a multi-source whitelist as described in any one of claims 1 to 7, characterized in that, The system includes: The processing module is used to acquire a multi-source heterogeneous dataset including several customer whitelist data, and to perform structured processing on the multi-source heterogeneous dataset to obtain a multi-source whitelist dataset. The conversion module is used to perform tagging processing on the multi-source whitelist dataset to generate a multi-dimensional feature vector corresponding to each customer. The multi-dimensional feature vector includes industry attribute tags, regional feature tags, and consumption level tags. The execution module is used to calculate the optimal routing path for each customer based on the multidimensional feature vector, distribute the outbound call task to the outbound call partner according to the optimal routing path, and receive the task execution results from the outbound call partner.
9. A computer 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 intelligent routing method based on a multi-source whitelist as described in any one of claims 1 to 7.
10. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the intelligent routing method based on a multi-source whitelist as described in any one of claims 1 to 7.