Multi-channel inquiry information processing system based on asynchronous distribution architecture

By using a multi-channel inquiry information processing system based on an asynchronous distribution architecture, and leveraging intelligent hierarchical distribution and elastic processing clusters, the problems of resource mismatch and low resource utilization have been solved, achieving a balance between high throughput and low latency, thus improving system performance.

CN123123343APending Publication Date: 2026-11-17郑州悉知信息科技股份有限公司
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611252018.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-18
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing multi-channel inquiry information processing systems suffer from resource mismatch due to indiscriminate traffic processing. High-priority messages are blocked by long-duration messages, the system cannot balance high throughput and low latency, and resource utilization is low.

Method used

A multi-channel inquiry information processing system based on an asynchronous distribution architecture is adopted. The intelligent hierarchical distribution module performs multi-dimensional attribute parsing on messages to generate processing attribute tuples, and routes messages to differentiated virtual queues. Combined with an elastic processing cluster, adaptive correlation aggregation batch processing is performed to achieve differentiated isolation distribution of messages and multiple reuse of common data.

Benefits of technology

The system achieves a balance between high throughput and low latency in the multi-channel inquiry information processing system, improves resource utilization, reduces redundant calculations and I/O overhead, and enhances the overall performance of the system.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123123343A_ABST
    Figure CN123123343A_ABST
Patent Text Reader

Abstract

This invention discloses a multi-channel inquiry information processing system based on an asynchronous distribution architecture, relating to the field of computer communication technology. The system includes: a multi-channel access module that receives inquiry messages from multiple channels and converts each message into a unified standard format; a message middleware cluster module that asynchronously receives and persistently stores the standard format messages; an intelligent hierarchical distribution module that pulls the standard format messages from the message middleware cluster module, performs multi-dimensional attribute parsing to generate processing attribute tuples, and routes the standard format messages to corresponding virtual queues in a pre-built differentiated virtual queue system based on the processing attribute tuples; and an elastic processing cluster module that performs adaptive associative aggregation batch processing on the standard format messages pulled from the differentiated virtual queue system. This invention achieves a balance between high throughput and low latency in multi-channel inquiry scenarios, as well as efficient utilization of system resources.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of computer communication technology, specifically to a multi-channel inquiry information processing system based on an asynchronous distribution architecture. Background Technology

[0002] With the rapid development of global cross-border e-commerce and foreign trade, businesses' customer touchpoints have shifted from a single official website inquiry portal to a multi-channel parallel model encompassing Facebook private messages, WhatsApp, official website online chat, corporate email, LinkedIn, and more. This trend has created an urgent need for unified access and efficient processing of inquiries from multiple channels. To address this challenge, the industry has gradually evolved from traditional monolithic architectures to distributed system architectures based on a "unified access layer + message middleware + asynchronous processing." These systems deploy standardized adapters at the access layer to convert messages from heterogeneous protocols across channels into an internally unified format and introduce message queue clusters to achieve asynchronous buffering and peak shaving of traffic, thereby solving fundamental problems such as link congestion and high system coupling inherent in synchronous architectures.

[0003] However, the aforementioned existing technologies still have the following significant drawbacks: On the one hand, their distribution strategies generally adopt a "non-discriminatory traffic processing" model, where messages from all channels and all business types are mixed and delivered to a single or homogeneous message queue, and are then pulled and processed by a unified consumer cluster. This mechanism ignores the essential differences in business priority, processing complexity, and real-time requirements among different messages, resulting in high-priority, simple query messages requiring immediate response being severely blocked by complex work order messages that consume long time, causing resource mismatch and preventing the system from achieving a balance between high throughput and low latency. On the other hand, existing consumer processing clusters mostly adopt a single-message serial retrieval and processing model, generating a large amount of network I / O overhead in high-concurrency scenarios. Moreover, for related messages of the same customer or with the same business intent, their common user profile data, CRM records, and protocol parsing results are executed independently and repeatedly in the processing flow of each message, resulting in a huge waste of computing and I / O resources, making it difficult to linearly increase system throughput and resulting in low overall resource utilization. Therefore, how to overcome the resource mismatch problem caused by the crude distribution strategy in the existing technology, and the low resource utilization problem caused by the inefficient processing mechanism, and achieve asynchronous distribution processing of multi-channel inquiry information with high throughput, low latency and efficient resource utilization is the problem we need to solve. To this end, we now provide a multi-channel inquiry information processing system based on asynchronous distribution architecture. Summary of the Invention

[0004] The purpose of this invention is to provide a multi-channel inquiry information processing system based on an asynchronous distribution architecture.

[0005] The objective of this invention can be achieved through the following technical solution: a multi-channel inquiry information processing system based on an asynchronous distribution architecture, comprising: The multi-channel access module is used to receive inquiry messages from multiple channels and convert each inquiry message into a unified standard format message; The message middleware cluster module is used to asynchronously receive and persistently store the standard format messages; The intelligent hierarchical distribution module is used to pull the standard format message from the message middleware cluster module, perform multi-dimensional attribute parsing on the standard format message to generate a processing attribute tuple, and route the standard format message to the corresponding virtual queue in the pre-built differentiated virtual queue system according to the processing attribute tuple. The elastic processing cluster module is used to perform adaptive correlation aggregation batch processing on standard format messages pulled from the differentiated virtual queue system to complete business logic processing and generate response messages.

[0006] Furthermore, the process by which the intelligent hierarchical distribution module performs multi-dimensional attribute parsing on the standard format message to generate processing attribute tuples includes: The standard format message is marked with a business priority to generate a priority tag; The standard format message is processed with a complexity prediction label to generate a complexity tag; the processing complexity prediction label is obtained by classifying the standard format message by intent through an intent recognition model; The standard format messages are marked with real-time requirements to generate real-time tags; the real-time requirement marking is determined based on the message channel type and user interaction status. The priority label, complexity label, and real-time label are combined into a triple, which is used as the processing attribute tuple.

[0007] Furthermore, the intelligent hierarchical distribution module also includes: A differentiated virtual queue routing controller is used to map standard format messages to multiple differentiated virtual queues based on the processing attribute tuples; the differentiated virtual queues include an immediate high-priority queue, a general standard queue, and a batch low-priority queue.

[0008] Furthermore, the intelligent hierarchical distribution module also includes: Real-time collection of CPU utilization and number of unfinished tasks from the elastic processing cluster module; For messages routed to the instant high priority queue, a low-latency push mode is adopted, that is, the message is pushed to the corresponding consumer cluster immediately when it arrives at the queue. For messages routed to the batch low-priority queue, an authorization signal is sent to the corresponding consumer cluster only when the CPU utilization of the corresponding consumer cluster is detected to be lower than a preset threshold and the number of incomplete tasks is lower than a preset number, allowing it to pull messages in batches.

[0009] Furthermore, the elastic processing cluster module includes: An adaptive batch aggregation processor is used to perform adaptive associative aggregation batch processing on messages pulled from the general standard queue or the batch low-priority queue.

[0010] Furthermore, the process of performing adaptive correlation aggregation batch processing includes: Based on the current message backlog in the queue, dynamically adjust the upper limit of the number of messages to be fetched at one time to obtain the batch of messages to be processed. For the batch of messages to be processed, within a preset time window or quantity window, multiple messages belonging to the same customer identifier or the same business intent are aggregated into a composite task unit. The common data query and parsing operation is performed on the composite task unit, and the obtained common data is cached in the memory cache of the composite task unit for sharing by all message processing logic within the composite task unit; The common data includes user profile data, customer relationship management records, and message protocol parsing results.

[0011] Furthermore, the elastic processing cluster module corresponds to different virtual queues in the differentiated virtual queue system and deploys consumer clusters with differentiated configurations.

[0012] Furthermore, the differentiated consumer cluster configuration includes: The consumer cluster corresponding to the real-time high-priority queue is configured with a local memory cache for storing hot data, which is the frequently accessed product inventory status or logistics query template. The consumer cluster corresponding to the batch low-optimal queue is deployed on a preemptive computing instance and automatically scales horizontally based on the queue message backlog and resource utilization.

[0013] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention uses an intelligent hierarchical distribution engine to perform multi-dimensional attribute parsing on standard format messages to generate processing attribute tuples, and routes messages to corresponding virtual queues in a pre-built differentiated virtual queue system based on these processing attribute tuples. This achieves differentiated isolation and distribution of messages with different priorities, complexities, and real-time requirements, solving the problems of resource mismatch and the inability to balance high throughput and low latency caused by indiscriminate traffic processing in existing technologies. Furthermore, by using an elastic and scalable backend processing cluster to pull messages from the differentiated virtual queue system and performing adaptive association aggregation batch processing on the pulled message batches, this invention achieves one-time loading and multiple reuse of common data in associated messages, eliminating redundant calculations and I / O overhead in single-message processing mode. This solves the problems of low resource utilization and difficulty in linearly increasing system throughput caused by inefficient processing mechanisms in existing technologies. This invention achieves a balance between high throughput and low latency in multi-channel inquiry scenarios and efficient utilization of system resources, solving the problems of resource mismatch caused by indiscriminate traffic processing and low resource utilization caused by single-message serial processing in existing technologies. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0015] Figure 1 This is a schematic diagram of the present invention. Detailed Implementation

[0016] like Figure 1 As shown, the multi-channel inquiry information processing system based on an asynchronous distribution architecture includes: The multi-channel access module is used to receive inquiry messages from multiple external channels and convert each inquiry message into a unified standard format message; The message middleware cluster module is used to asynchronously receive and persistently store the standard format messages; The intelligent hierarchical distribution module is used to pull the standard format message from the message middleware cluster module, perform multi-dimensional attribute parsing on the standard format message to generate a processing attribute tuple, and route the standard format message to the corresponding virtual queue in the pre-built differentiated virtual queue system according to the processing attribute tuple. The elastic processing cluster module is used to pull messages from the differentiated virtual queue system and perform adaptive batch processing on the pulled message batches to complete business logic processing and generate response messages.

[0017] It should be further explained that, in the specific implementation process, the multi-channel access module receives inquiry messages from multiple channels and converts each inquiry message into a unified standard format message, including: The multi-channel access module is used to connect to multiple external channels such as Facebook private messages, WhatsApp, official website online chat, and corporate email, to receive original inquiry messages, and to uniformly convert original inquiry messages of different protocols and formats into standard format messages within the system. Specifically, the multi-channel access module contains multiple channel adapters. Each adapter encapsulates the protocol parsing, encryption / decryption, and authentication logic for the corresponding channel. It extracts the key metadata of the original inquiry message and encapsulates it into a unified JSON format message. The JSON message includes fields such as: message unique identifier, channel source type, customer unique identity identifier, message body content, message attachment list, and timestamp. After conversion, the multi-channel access module asynchronously writes the standard format message to the message middleware cluster module.

[0018] It should be further explained that, in the specific implementation process, the message middleware cluster module includes: The message middleware cluster module consists of a hybrid high-throughput cluster and a low-latency cluster. The high-throughput cluster is built using Apache Kafka and is used to handle non-real-time, batch-processable messages. The low-latency cluster is built using RabbitMQ and is used to handle messages that require immediate response. After receiving a standard format message, the message middleware cluster persists the message to disk and performs multi-replica synchronization to ensure that the message is not lost. At the same time, it returns a successful write confirmation to the upstream.

[0019] It should be further explained that, in the specific implementation process, the intelligent hierarchical distribution module, as the core scheduling component, pulls standard format messages from the message middleware cluster module in the form of an independent consumer process, and performs multi-dimensional attribute parsing and differentiated routing; wherein, the process of performing multi-dimensional attribute parsing on the standard format messages to generate processing attribute tuples includes: The standard format message is marked with a business priority to generate a priority tag; The standard format message is processed with a complexity prediction label to generate a complexity tag; the processing complexity prediction label is obtained by classifying the standard format message by intent through an intent recognition model; The standard format messages are marked with real-time requirements to generate real-time tags; the real-time requirement marking is determined based on the message channel type and user interaction status. The priority label, complexity label, and real-time label are combined into a triplet, which is used as the processing attribute tuple. Specifically, the intelligent hierarchical distribution module performs three-dimensional attribute parsing on the retrieved standard format messages using multi-dimensional message attribute parsing and a hierarchical marker: The first dimension is business priority marking. The system has a pre-set high-value keyword library, such as "bulk purchase", "OEM", "CEO", etc., and maintains a cache of customer historical transaction volume levels. This cache is updated regularly by synchronizing with the CRM system. At the same time, a weight coefficient is preset for each channel, for example, the weight of WhatsApp channel is 0.9 and the weight of email channel is 0.6. The grade marker matches the message body with the high-value keyword library to obtain the keyword matching score; queries the customer's historical transaction volume level and maps it to the level score; reads the channel source of the message to obtain the channel weight score; calculates the comprehensive priority score by summing the above three scores according to the preset weighted summation formula, and generates a priority label after comparing it with the threshold. The priority label value set is {high, medium, low}. The second dimension is the processing complexity prediction label. The hierarchical labeler is equipped with a lightweight intent recognition model based on BERT. This model classifies the message body into two intent categories: "simple query" or "complex work order" and generates corresponding complexity labels. The third dimension is the real-time requirement label. The graded labeler determines the requirement based on the message channel type and user interaction status: if the channel is the official website online chat or WhatsApp and the user is in an active session, it is labeled as "instant response"; if the channel is corporate email or LinkedIn and there is no active session label, it is labeled as "allow delay". Finally, the generated priority label, complexity label, and real-time label are combined into a triple, denoted as the processing attribute tuple. ,in For priority tags, For complexity labels, This is a real-time tag.

[0020] It should be further noted that, in the specific implementation process, the intelligent hierarchical distribution module also includes: A differentiated virtual queue routing controller is used to map standard format messages to multiple differentiated virtual queues based on the processing attribute tuples; the differentiated virtual queues include an immediate high-priority queue, a general standard queue, and a batch low-priority queue; The process of constructing a differentiated virtual queue routing controller includes: The differentiated virtual queue routing controller creates multiple topics in a high-throughput Kafka cluster, each serving as a different virtual queue. Its routing rules are based on strict matching of the processing attribute tuple T: when T satisfies the conditions P=high, C=simple query, and R=immediate response, the message is routed to the immediate high-priority queue; when T satisfies the conditions (P=medium and C=simple query) or (P=high and C=complex work order), the message is routed to the general standard queue; when T satisfies the conditions P=low or R=allowed delay, the message is routed to the batch low-priority queue. This queue partitioning achieves strict physical isolation of messages with different processing needs, avoiding resource contention.

[0021] It should be further explained that, in the specific implementation process, the intelligent hierarchical distribution module also includes a load-aware dynamic distribution executor, the working process of which includes: Real-time collection of CPU utilization and number of unfinished tasks from the elastic processing cluster module; For messages routed to the instant high-priority queue, a low-latency push mode is adopted, and the message is immediately pushed to the corresponding consumer cluster through a long connection when it arrives at the queue. For messages routed to the batch low-priority queue, an authorization signal is sent to the corresponding consumer cluster to allow it to pull messages in batch pull mode only when the CPU utilization of the corresponding consumer cluster is detected to be lower than a preset threshold and the number of unfinished tasks is lower than a preset number. Specifically, the load-aware dynamic dispatch executor collects the CPU utilization of each consumer group in the elastic processing cluster module through the Kubernetes API, and collects the message backlog of the consumer group as the number of unfinished tasks by calling the metric interface of the message middleware. For messages routed to the immediate high-priority queue, the dynamic dispatch executor adopts a low-latency push mode. That is, in the consumer connection channel maintained internally by the dynamic dispatch executor, once a message arrives at the immediate high-priority queue, it is immediately pushed to the corresponding consumer cluster of the immediate high-priority queue through a long connection, ensuring that the end-to-end latency is less than 50 milliseconds. For messages routed to the batch low-priority queue, the dynamic dispatch executor only sends an authorization signal to the cluster when it detects that the CPU utilization of the corresponding batch low-priority queue consumer cluster is lower than a preset threshold of 40% and the number of unfinished tasks is lower than a preset number of 100. This allows the cluster to pull up to 500 messages at a time from the batch low-priority queue for processing in batch pull mode. If the load exceeds the threshold, the pull is suspended to avoid impacting the overall system.

[0022] It should be further explained that, in the specific implementation process, the elastic processing cluster module receives messages distributed by the intelligent hierarchical distribution module, completes the actual business processing, and sends the generated response message back to the multi-channel access module through the response queue in the message middleware cluster module, and finally replies to the end user; the elastic processing cluster module specifically includes: An adaptive batch aggregation processor is used to perform adaptive associative aggregation batch processing on messages pulled from a general standard queue or a low-priority batch queue. The adaptive correlation aggregation batch processing specifically includes the following processing steps: First, dynamic batch retrieval; based on the current message backlog in the queue, the upper limit of the number of messages retrieved at one time is dynamically adjusted to obtain a batch of messages to be processed; specifically, the retrieval controller in the adaptive batch aggregation processor continuously monitors the current message backlog in the subscribed queue, denoted as... The system has a preset dynamic adjustment strategy table. At that time, the maximum number of items that can be retrieved at one time is set to 50; when At that time, the upper limit increases linearly to 200; when At this time, the upper limit is further increased to 500 messages; the pull controller performs message pull according to this dynamic upper limit to reduce the number of network round trips; Second, the relevance window aggregation: For the batch of messages to be processed, within a preset time window or quantity window, multiple messages belonging to the same customer identifier or the same business intent are aggregated into a composite task unit; specifically, the dynamically batch-fetched message batch is sent to the aggregation window buffer, which maintains a time window of 200 milliseconds and a quantity window with a maximum capacity of 200 messages, and the aggregation is triggered based on the first arrival; within the window, the adaptive batch aggregation processor groups the messages: first, it hashes the messages by the customer's unique identifier, and aggregates the messages of the same customer into a sub-unit; then, it compares the intent tags of the messages in each sub-unit, and further aggregates messages belonging to the same business intent, such as all being "inquiries about product A", ultimately forming at least one composite task unit; the composite task unit maintains a message list and a shared context pointer; Third, common resources are preloaded and reused in a unified manner. The common data is queried and parsed for each composite task unit, and the obtained common data is cached in the memory cache of that composite task unit for sharing by all message processing logic within the unit. Specifically, for each composite task unit, before executing specific business logic, the processor first performs a unified preloading step: based on the customer identifier carried in the first message within the unit, it initiates only one query request to the customer relationship management system to obtain user profile data and historical order records; simultaneously, it performs only one complete parsing of the message protocol to extract all necessary structure data; the obtained common data is stored in the memory cache of the composite task unit and marked as read-only. This common data includes user profile data, customer relationship management records, and message protocol parsing results. When each sub-message within the unit enters the business processing pipeline sequentially, the required common data is read directly from this memory cache without initiating network calls or repeating parsing, thereby greatly reducing I / O overhead and redundant CPU computation.

[0023] It should be further explained that, in the specific implementation process, the elastic processing cluster module corresponds to different virtual queues in the differentiated virtual queue system and is deployed with consumer clusters with differentiated configurations. The consumer cluster corresponding to the real-time high-priority queue is configured with a local memory cache for storing hot data, which is the frequently accessed product inventory status or logistics query template. The consumer cluster corresponding to the aforementioned low-optimal batch queues is deployed on preemptive computing instances and performs horizontal autoscaling based on the queue message backlog and resource utilization. Specifically, the consumer cluster corresponding to the immediate high-priority queue is deployed on high-performance bare metal or dedicated cloud server instances, internally configured with a local memory cache based on Caffeine to store frequently accessed hot data. This hot data consists of product inventory status or logistics query templates that have been queried more than 100 times in the past hour. The cached data uses a Redis publish / subscribe mechanism to monitor data change events to ensure eventual consistency. The consumer cluster corresponding to the batch low-priority queue is deployed on preemptive computing instances provided by cloud service providers, such as AWS Spot Instances or Alibaba Cloud preemptive instances, to reduce computing costs. The horizontal autoscaling strategy of this cluster is implemented through Kubernetes HPA (Horizontal Pod Autoscaler), with the scaling metric set to the message backlog of the batch low-priority queue. The combination of cluster average CPU utilization, when Expansion is triggered when the value is greater than 2000 and CPU utilization is higher than 70%. When the number of CPUs is less than 500 and the CPU utilization is below 30%, a scaling down mechanism is triggered to maximize resource utilization while ensuring processing power.

[0024] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications or equivalent substitutions made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A multi-channel inquiry information processing system based on an asynchronous distribution architecture, characterized in that, include: The multi-channel access module is used to receive inquiry messages from multiple channels and convert each inquiry message into a unified standard format message; The message middleware cluster module is used to asynchronously receive and persistently store the standard format messages; The intelligent hierarchical distribution module is used to pull the standard format message from the message middleware cluster module, perform multi-dimensional attribute parsing on the standard format message to generate a processing attribute tuple, and route the standard format message to the corresponding virtual queue in the pre-built differentiated virtual queue system according to the processing attribute tuple. The elastic processing cluster module is used to perform adaptive correlation aggregation batch processing on standard format messages pulled from the differentiated virtual queue system to complete business logic processing and generate response messages.

2. The multi-channel inquiry information processing system based on asynchronous distribution architecture according to claim 1, characterized in that, The intelligent hierarchical distribution module performs multi-dimensional attribute parsing on the standard format message to generate processing attribute tuples, including: The standard format message is marked with a business priority to generate a priority tag; The standard format message is processed with a complexity prediction label to generate a complexity tag; the processing complexity prediction label is obtained by classifying the standard format message by intent through an intent recognition model; The standard format messages are marked with real-time requirements to generate real-time tags; the real-time requirement marking is determined based on the message channel type and user interaction status. The priority label, complexity label, and real-time label are combined into a triple, which is used as the processing attribute tuple.

3. The multi-channel inquiry information processing system based on asynchronous distribution architecture according to claim 1, characterized in that, The intelligent hierarchical distribution module also includes: A differentiated virtual queue routing controller is used to map standard format messages to multiple differentiated virtual queues based on the processing attribute tuples; the differentiated virtual queues include an immediate high-priority queue, a general standard queue, and a batch low-priority queue.

4. The multi-channel inquiry information processing system based on asynchronous distribution architecture according to claim 3, characterized in that, The intelligent hierarchical distribution module also includes: Real-time collection of CPU utilization and number of unfinished tasks from the elastic processing cluster module; For messages routed to the instant high priority queue, a low-latency push mode is adopted, that is, the message is pushed to the corresponding consumer cluster immediately when it arrives at the queue. For messages routed to the batch low-priority queue, an authorization signal is sent to the corresponding consumer cluster only when the CPU utilization of the corresponding consumer cluster is detected to be lower than a preset threshold and the number of incomplete tasks is lower than a preset number, allowing it to pull messages in batches.

5. The multi-channel inquiry information processing system based on asynchronous distribution architecture according to claim 3, characterized in that, The elastic processing cluster module includes: An adaptive batch aggregation processor is used to perform adaptive associative aggregation batch processing on messages pulled from the general standard queue or the batch low-priority queue.

6. The multi-channel inquiry information processing system based on an asynchronous distribution architecture according to claim 5, characterized in that, The process of performing adaptive correlation aggregation batch processing includes: Based on the current message backlog in the queue, dynamically adjust the upper limit of the number of messages to be fetched at one time to obtain the batch of messages to be processed. For the batch of messages to be processed, within a preset time window or quantity window, multiple messages belonging to the same customer identifier or the same business intent are aggregated into a composite task unit. The common data query and parsing operation is performed on the composite task unit, and the obtained common data is cached in the memory cache of the composite task unit for sharing by all message processing logic within the composite task unit; The common data includes user profile data, customer relationship management records, and message protocol parsing results.

7. The multi-channel inquiry information processing system based on asynchronous distribution architecture according to claim 1, characterized in that, The elastic processing cluster module corresponds to different virtual queues in the differentiated virtual queue system and deploys consumer clusters with differentiated configurations.

8. The multi-channel inquiry information processing system based on asynchronous distribution architecture according to claim 7, characterized in that, The differentiated consumer clusters include: The consumer cluster corresponding to the real-time high-priority queue is configured with a local memory cache for storing hot data, which is the frequently accessed product inventory status or logistics query template. The consumer cluster corresponding to the batch low-optimal queue is deployed on a preemptive computing instance and automatically scales horizontally based on the queue message backlog and resource utilization.