Task execution method and device applied to unmanned convenience store, equipment and medium

Abstract

The embodiment of the invention discloses a task execution method and device applied to an unmanned convenience store, equipment and a medium. A specific embodiment of the method comprises the following steps: obtaining dialogue records of switching to manual service after a user dials an intelligent customer service telephone, and obtaining a dialogue record set; performing label matching processing on each dialogue record in the dialogue record set by using a pre-trained large language model to obtain a dialogue label set; allocating a corresponding task queue to each dialogue tag in the dialogue tag set to obtain an incoming call waiting queue and an operation class queue; performing sequential optimization on the incoming call waiting queue to obtain a waiting queue; and sending the waiting queue and the operation class queue to a task processing terminal corresponding to a manual customer service so as to control unmanned convenience store equipment to execute the operation class task. According to the embodiment, the waiting time of customers can be remarkably shortened, and the service efficiency of manual customer service is improved.

Description

Technical Field

[0001] Embodiments of the present disclosure relate to the field of computer technology, and more particularly to a task execution method, apparatus, device, and medium applied to unmanned convenience stores. Background Art

[0002] In unmanned convenience store scenarios, customers encounter shopping issues and need to call human customer service. While intelligent customer service can resolve some issues before human customer service intervenes, there are still issues that cannot be resolved, such as equipment responsiveness issues in unmanned convenience stores. These issues will still be transferred to human customer service. For human customer service to handle issues, the typical approach is to transfer the user's questions and calls to online human customer service representatives and handle them in chronological order.

[0003] However, when using the above approach to handle customer issues, the following technical issues often arise:

[0004] First, unmanned convenience store equipment suffers from slow mechanical response and insufficient sensor accuracy, leading to long queues during peak usage periods and frequent transfers to human service. Human customer service handles issues chronologically, deprioritizing urgent or short-term tasks, resulting in low resource utilization. Frequent intervention by customer service personnel to troubleshoot related equipment can lead to significant wear and tear.

[0005] Second, manual customer service provides answers based on knowledge bases or experience. The knowledge base is often updated with a lag, and unmanned convenience store equipment lacks the ability to detect and repair. This results in customer service being unable to respond to all issues. Some situations caused by equipment repairs may need to be transferred to professional departments on-site to handle the repair of related equipment.

[0006] The above information disclosed in this Background section is only for enhancement of understanding of the background of the inventive concept and therefore it may contain information that does not form the prior art that is already known in this country to a person of ordinary skill in the art. Summary of the Invention

[0007] The content of this disclosure is used to briefly introduce concepts that will be described in detail in the detailed description section below. The content of this disclosure is not intended to identify key features or essential features of the claimed technical solution, nor is it intended to limit the scope of the claimed technical solution.

[0008] Some embodiments of the present disclosure propose task execution methods, devices, equipment, and media applied to unmanned convenience stores to solve one or more of the technical problems mentioned in the above background technology section.

[0009] In a first aspect, some embodiments of the present disclosure provide a task execution method applied to unmanned convenience stores, including: obtaining conversation records when a user calls an intelligent customer service phone and is transferred to manual service to obtain a conversation record set; using a preset tag library, using a pre-trained large language model to perform tag matching processing on each conversation record in the above conversation record set to generate a conversation tag to obtain a conversation tag set; using the task queue corresponding to the manual customer service, assigning a corresponding task queue to each conversation tag in the above conversation tag set to obtain an incoming waiting queue and an operation queue; performing sequential optimization on the above incoming waiting queue to obtain a waiting queue; sending the above waiting queue and the above operation queue to the task processing terminal corresponding to the manual customer service to control the unmanned convenience store equipment to perform operation tasks, wherein the above operation tasks include: product distribution and door opening.

[0010] In a second aspect, some embodiments of the present disclosure provide a task execution device for an unmanned convenience store, including: a conversation acquisition unit, configured to acquire conversation records when a user dials an intelligent customer service phone and is transferred to manual service, to obtain a conversation record set; a label matching unit, configured to use a preset label library and a pre-trained large language model to perform label matching processing on each conversation record in the above conversation record set to generate a conversation label and obtain a conversation label set; a queue allocation unit, configured to assign a corresponding task queue to each conversation label in the above conversation label set through the task queue corresponding to the manual customer service, to obtain an incoming waiting queue and an operation queue; a queue optimization unit, configured to perform sequential optimization on the above incoming waiting queue to obtain a waiting queue; an operation processing unit, configured to send the above waiting queue and the above operation queue to the task processing terminal corresponding to the manual customer service, to control the unmanned convenience store equipment to perform operation tasks, wherein the above operation tasks include: product distribution and door opening.

[0011] In a third aspect, some embodiments of the present disclosure provide an electronic device comprising: one or more processors; a storage device on which one or more programs are stored, and when the one or more programs are executed by one or more processors, the one or more processors implement the method described in any implementation manner in the first aspect.

[0012] In a fourth aspect, some embodiments of the present disclosure provide a computer-readable medium having a computer program stored thereon, wherein when the program is executed by a processor, the method described in any implementation manner in the first aspect is implemented.

[0013] The aforementioned embodiments of the present disclosure have the following beneficial effects: The task execution methods applied to unmanned convenience stores, as described in some embodiments of the present disclosure, reduce the time it takes for human customer service staff to handle issues, significantly shorten customer waiting times, and improve the efficiency of human customer service. Specifically, the long wait times for customer service staff to handle issues are due to the fact that, in unmanned convenience stores, a single human customer service representative must handle call requests from multiple stores. During peak hours, customers may queue up and wait for service for extended periods, degrading the user shopping experience. Based on this, the task execution methods applied to unmanned convenience stores, as described in some embodiments of the present disclosure, first obtain conversation records from user calls to an intelligent customer service hotline that are transferred to human service, thereby generating a conversation record set. This provides complete user problem information, providing a data foundation for subsequent intent classification and task priority determination. Then, using a pre-trained large language model, label matching is performed on each conversation record in the conversation record set to generate conversation labels, resulting in a conversation label set. This automatically identifies the type of user issue, resolving the issue of delayed updates to the human-dependent database. Secondly, using the task queue corresponding to the human customer service representative, each conversation label in the conversation label set is assigned a corresponding task queue, resulting in an incoming call waiting queue and an operation queue. This separates operational tasks from inbound call tasks, preventing long-duration tasks from blocking the processing of shorter tasks and improving customer service resource utilization. Next, the incoming call waiting queue is sequentially optimized to generate a waiting queue. By interrupting and retracting short-duration tasks, urgent tasks or those that are shorter than other tasks can be optimized, reducing queue times during peak hours. Finally, the waiting queue and operational queue are sent to the corresponding task processing terminal for human customer service to control the unmanned convenience store equipment to execute operational tasks, including product distribution and door unlocking. This allows customer service to process inbound call tasks and one-click operational tasks (such as remote door unlocking) in parallel, shortening problem resolution time, reducing the frequency of referrals to specialized departments, and improving overall service efficiency. In summary, by analyzing user call records and leveraging a large model, label matching and processing time estimation are achieved. The human customer service task queue is prioritized and sorted, enabling multi-task parallel processing. This shortens user wait times and improves human customer service efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] The above and other features, advantages, and aspects of the various embodiments of the present disclosure will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. Throughout the drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic and that components and elements are not necessarily drawn to scale.

[0015] Figure 1is a flowchart of some embodiments of a task execution method applied to an unmanned convenience store according to the present disclosure;

[0016] Figure 2 is a schematic structural diagram of some embodiments of a task execution device applied to an unmanned convenience store according to the present disclosure;

[0017] Figure 3 It is a structural diagram of an electronic device suitable for implementing some embodiments of the present disclosure. DETAILED DESCRIPTION

[0018] Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as being limited to the embodiments described herein. On the contrary, these embodiments are provided to provide a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are for illustrative purposes only and are not intended to limit the scope of protection of the present disclosure.

[0019] It should also be noted that, for ease of description, only the parts related to the invention are shown in the drawings. In the absence of conflict, the embodiments and features in the embodiments of the present disclosure may be combined with each other.

[0020] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0021] It should be noted that the modifications of "one" and "multiple" mentioned in the present disclosure are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, they should be understood as "one or more".

[0022] The names of the messages or information exchanged between multiple devices in the embodiments of the present disclosure are only used for illustrative purposes and are not used to limit the scope of these messages or information.

[0023] The present disclosure will be described in detail below with reference to the accompanying drawings and in conjunction with embodiments.

[0024] refer to Figure 1 , shows a process 100 of some embodiments of a task execution method applied to an unmanned convenience store according to the present disclosure. The task execution method applied to an unmanned convenience store includes the following steps:

[0025] Step 101: Obtain the conversation record of the user who dialed the intelligent customer service phone and was transferred to the manual service to obtain a conversation record set.

[0026] In some embodiments, the execution subject (for example, an electronic device) of the task execution method applied to the unmanned convenience store can be the customer service system of the unmanned convenience store. The customer service system can include a background processing part, a manual customer service task processing terminal and an unmanned convenience store equipment control interface. The user can be a customer who uses the unmanned convenience store service and encounters problems and contacts customer service by phone. The smart customer service phone can be an automated telephone answering system that can handle and answer common questions. The transfer to manual service can be when the smart customer service cannot solve the user's problem and transfers the call to a real customer service. The conversation record can be the complete interaction content (voice or text) between the user and the smart customer service, including timestamp, problem description, and transfer node information.

[0027] As an example, the voice call between the intelligent customer service representative and the user can be converted into text. Secondly, the text information can be stored in a database as a record of the conversation. For example, the user's voice call with the intelligent customer service representative might be "I need to open the door of the convenience store." The text information can be {"transfer_flag":"true","user_id":"U12345","timestamp":"2024-01-01 14:30:00","content":"I need to open the door of the convenience store"}.

[0028] In step 102, a pre-trained large language model is used to perform label matching processing on each conversation record in the conversation record set through a preset label library to generate a conversation label and obtain a conversation label set.

[0029] In some embodiments, the execution entity may use a pre-trained large language model to perform label matching on each conversation record in the conversation record set using a preset label library to generate conversation labels and obtain a conversation label set. The preset label library may be a preset classification label used to describe the question type and user intent in the conversation record. The pre-trained large language model may be a natural language processing model based on a large amount of scene data, capable of understanding text semantics and generating classification results. For example, GPT-4. The large number of scenes may be customer service conversations in unmanned convenience stores. The conversation labels may be classification labels assigned to the conversation records after matching. The conversation label set may be a collection of conversation records with matching labels, used for subsequent processing steps. The label matching process may be calling a third-party LLM (Large Language Model) interface to determine the corresponding matching label set for the conversation record.

[0030] As an example, first, the LLM interface is called to identify keywords in the conversation log. Second, similar tags in the preset tag library are matched based on the keywords. Finally, the one with the highest similarity is selected as the conversation tag. For example, the keyword in the conversation log may be "checkout, amount error." Similar tags in the preset tag library may be "checkout error" and "product out of stock." The similarity with "checkout error" may be 0.95, and the similarity with "product out of stock" may be 0.1. The conversation tag may then be "checkout error."

[0031] In some optional implementations of some embodiments, the execution entity may use a preset tag library and a pre-trained large language model to perform tag matching processing on each conversation record in the conversation record set to generate conversation tags, which may include the following steps:

[0032] In the first step, the conversation records are categorized by labeling using the large language model to obtain a first-level classification, where the first-level classification is one of the following: an operation-related task label or an incoming call-related task label. The first classification can be a coarse-grained classification of the conversation records. The operation-related task label can be an operation task that can be performed directly without having to speak with a human customer service representative. For example, the operation-related task label can be a task label corresponding to opening a shelf door or opening a convenience store door. The incoming call-related task label can be a task label corresponding to a problem that requires speaking with a human customer service representative to resolve it. For example, the incoming call-related task label can be a task label corresponding to issues with product quality or product checkout amount.

[0033] For example, the LLM interface can be called to use a large language model to identify the semantics of conversation logs, determining whether a customer service representative needs to proactively operate the device or require extended communication. For example, a user conversation log might be "opening the convenience store door." After large language analysis, this can be labeled as an operational task.

[0034] In the second step, in response to the first-level classification being in the preset tag library, the following steps are performed:

[0035] In the first sub-step, in response to the first-level classification as an operation-related task label, an operation-related label and an estimated processing time are generated based on the preset label library, and the operation-related label is used as a conversation label. The operation-related label can be a label for the conversation record, indicating a short-term task that requires manual customer service to trigger device commands to complete. The estimated processing time can be the estimated time required to complete the operation task.

[0036] As an example, first, we categorize tags from historical customer service call logs and calculate the corresponding handling times. Next, we add these tags and corresponding handling times to a pre-set tag library. Finally, we use LLM to analyze the call logs based on the pre-set tag library to generate the corresponding tag type and estimated handling time.

[0037] In the second sub-step, in response to the first-level classification of the task as an inbound call, an inbound call tag and estimated processing time are generated based on the preset tag library, and the inbound call tag is used as the conversation tag. The inbound call tag can be a label for the conversation record, indicating a long-term task that requires human customer service to resolve the issue and control the terminal to trigger device commands to complete.

[0038] For example, the conversation record may be "product checkout problem." Matching "checkout problem" in the preset tag library may yield an estimated processing time of "60 seconds."

[0039] In the third step, in response to the first-level classification not being in the preset label library, a target clustering algorithm is used to generate candidate labels corresponding to the first-level classification. The target clustering algorithm can be an unsupervised learning algorithm, such as a K-means or DBSCAN algorithm, for grouping similar conversation records and discovering new question types. The candidate labels can be temporary labels for the question types to be updated in the clustering results.

[0040] For example, if a match cannot be found in the preset tag library, key high-frequency words can be extracted from the conversation log as candidate tags. For example, the key word for clustering could be "network link failure." Then the candidate tag could be "network failure."

[0041] In the fourth step, if the candidate tag passes manual review, the candidate tag is added to the preset tag library as a conversation tag. Manual review can involve an administrator or expert confirming the rationality and necessity of the candidate tag. For example, a candidate tag might be "network failure." If it passes manual review, "network failure" is added to the preset tag library.

[0042] Step 103: assign a corresponding task queue to each dialog tag in the dialog tag set through the task queue corresponding to the manual customer service, and obtain an incoming call waiting queue and an operation queue.

[0043] In some embodiments, the execution entity can assign a corresponding task queue to each conversation tag in the conversation tag set through the task queue corresponding to the manual customer service, thereby obtaining an incoming call waiting queue and an operation queue. The task queue corresponding to the manual customer service can be divided into two types of queues based on the type of task, facilitating the management of work order tasks handled by the manual customer service. The incoming call waiting queue can be a queue for tasks that require continuous interaction between the customer service and the user. The operation queue can be a queue for tasks that can be completed by a single operation by the customer service.

[0044] As an example, tasks can be assigned to corresponding queues based on the dialog tags (operation tags / incoming tags).

[0045] In some optional implementations of some embodiments, the execution entity may assign a corresponding task queue to each dialog tag in the dialog tag set through a task queue corresponding to a human customer service representative, thereby obtaining an incoming call waiting queue and an operation queue, which may include the following steps:

[0046] In the first step, in response to the dialog tag being a customer service continuous interaction task, the dialog tag and the estimated processing time corresponding to the dialog tag are assigned to the incoming call waiting queue. The customer service continuous interaction task may be a task that requires multiple communications between the user and a human customer service representative to resolve the issue.

[0047] In the second step, in response to the dialog tag being a customer service single-time task, the dialog tag and the estimated processing time corresponding to the dialog tag are assigned to an operation queue. The customer service single-time task may be a control task that can be completed by a human customer service representative through a terminal control device with a single operation.

[0048] As an example, first, determine whether the dialog tag type is an operation class. Finally, add the dialog tag and estimated processing time to the operation class queue, and generate device instruction parameters.

[0049] Step 104: Optimize the order of the incoming call waiting queue to obtain a waiting queue.

[0050] In some embodiments, the execution entity may perform sequential optimization on the incoming call waiting queue to obtain a waiting queue, wherein the waiting queue may be a sequentially optimized incoming call waiting queue, and tasks are dynamically sorted by priority.

[0051] For example, tasks with a short estimated processing time can be placed at the front of the queue, while tasks with a long estimated processing time can be placed at the back of the queue. For example, the incoming call queue could be [A (other questions 120s), B (buy cigarettes 5s), D (checkout 60s), E (buy cigarettes 5s)]. Then the waiting queue could be [E (buy cigarettes 5s), B (buy cigarettes 5s), D (checkout 60s), A (other questions 120s)].

[0052] In some optional implementations of some embodiments, the execution entity may perform sequential optimization on the incoming call waiting queue to obtain the waiting queue, which may include the following steps:

[0053] The first step is to traverse the incoming call waiting queue from front to back, marking each short-term task to obtain a short-term task queue. These short-term tasks may be tasks with an estimated processing time less than a preset threshold. The short-term task queue may be a queue of short-term tasks selected from the incoming call waiting queue for subsequent queue insertion operations.

[0054] In the second step, for each short-term task in the short-term task queue, perform the following insertion steps:

[0055] The first sub-step is to determine the position of the short-term task in the incoming call waiting queue as the target position, wherein the target position may be the original index position of the short-term task in the original incoming call waiting queue.

[0056] In a second sub-step, in response to a long-duration task existing before the target location in the incoming call waiting queue, the waiting time and number of times the long-duration task has been inserted are determined, wherein the waiting time is the time the long-duration task has remained in the incoming call waiting queue. The long-duration task may be a task with an estimated processing time less than a preset threshold. The number of times the long-duration task has been inserted is the cumulative number of times a short-duration task has been inserted into the queue, thereby limiting excessive queue interruption.

[0057] In a third sub-step, in response to the fact that the number of times the long-term task has been inserted is less than a preset number and the waiting time does not exceed a preset threshold, the short-term task is inserted before the corresponding sequence position of the long-term task.

[0058] For example, the incoming call queue might be [A (other questions, 120 seconds), B (buy cigarettes, 5 seconds), D (checkout, 60 seconds), E (buy cigarettes, 5 seconds)]. The short-duration task might be E. The preset number of times might be 2. The number of times long-duration task A is inserted might be 1. The waiting time for long-duration task A might be 30 seconds. The preset threshold might be 40 seconds. After inserting E, the incoming call queue might be [E (buy cigarettes, 5 seconds), A (other questions, 120 seconds), B (buy cigarettes, 5 seconds), D (checkout, 60 seconds)].

[0059] The fourth sub-step is, in response to the waiting time exceeding a preset threshold, inserting the short-term task into the incoming call waiting queue after the corresponding sequence position of the long-term task.

[0060] For example, the incoming call queue might be [A (other questions, 120 seconds), B (buy cigarettes, 5 seconds), D (checkout, 60 seconds), E (buy cigarettes, 5 seconds)]. The short-duration task might be E. The preset number of times might be 2. The number of times long-duration task A is inserted might be 1. The waiting time for long-duration task A might be 70 seconds. The preset threshold might be 40 seconds. After inserting E, the incoming call queue might be [A (other questions, 120 seconds), E (buy cigarettes, 5 seconds), B (buy cigarettes, 5 seconds), D (checkout, 60 seconds)].

[0061] The third step is to determine that the incoming call waiting queue after the insert operation is the above waiting queue.

[0062] For example, the original incoming call waiting queue may be [A (other questions 120s), B (buy cigarettes 5s), D (checkout 60s), E (buy cigarettes 5s)]. Then, the incoming call waiting queue after the insertion operation may be [B (buy cigarettes 5s), E (buy cigarettes 5s), A (other questions 120s), inserted times (2), D (checkout 60s), (inserted times 1)].

[0063] In the process of adopting technical solutions to solve the above-mentioned technical problem 1, the following problems often occur: short-term tasks are frequently inserted in front of long-term tasks, resulting in the processing order of long-term tasks being continuously shifted back, causing long-term tasks to be unable to be solved and an imbalance in the fairness of queue-jumping.

[0064] Conventional solutions to these problems generally limit the number of queue jumps. When the preset number of queue jumps is reached, short-term tasks will be prohibited from queue jumping. However, the inventors consider that this will cause long-term tasks to remain for too long and still be unable to be processed. This cannot meet the dynamic queuing scenario. We can adopt the following solutions:

[0065] Optionally, in some optional implementations of some embodiments, the execution entity may perform sequential optimization on the incoming call waiting queue to obtain the waiting queue, which may include the following steps:

[0066] The first step is to traverse the incoming call waiting queue from front to back, marking each long-duration task to obtain a long-duration task queue. The incoming call waiting queue can be a task queue that has undergone preliminary optimization and may contain short-duration task queues that have already been interrupted. The long-duration task queue can be a queue composed of long-duration tasks screened from the incoming call waiting queue, which is used for subsequent retrieval operations.

[0067] The second step is to perform the following rollback steps for each long-duration task in the long-duration task queue:

[0068] The first sub-step is to determine the position of the long-duration task in the incoming call waiting queue as the target position, wherein the target position may be the current index position of the long-duration task in the incoming call waiting queue.

[0069] The second sub-step is to determine the waiting time and number of times the long-duration task has been inserted. The waiting time can be the time the long-duration task has been in the incoming call queue. The number of times it has been inserted can be the cumulative number of times it has been interrupted by short-duration tasks, which is used to determine whether to trigger a pullback.

[0070] The third sub-step is, in response to the number of insertions being non-zero, traversing the incoming call waiting queue to the target position to obtain an inserted short-time task queue, wherein the inserted short-time task queue may be all short-time tasks inserted before the incoming call waiting queue reaches the target position.

[0071] For example, the incoming call waiting queue may be [E (buy cigarettes 5s), A (other questions 120s), inserted times (1), B (buy cigarettes 5s), D (checkout 60s)]. The short-term task inserted before the long-term task A may be E.

[0072] The fourth sub-step is to decrement the number of times each short-term task has been inserted into the short-term task queue. The number of times the short-term task has been inserted may be the cumulative number of times the short-term task has been inserted into the queue. The decrement operation may be to reduce the number of times the short-term task has been inserted by one.

[0073] For example, the incoming call waiting queue can be [E (buy cigarettes 5s), inserted times (2), A (other questions 120s), inserted times (1), B (buy cigarettes 5s), D (checkout 60s)]. The short-term task inserted before the long-term task A can be E. Then the number of times E has been inserted should be 1.

[0074] The fifth sub-step is to determine the queue position of the first short-term task in the inserted short-term task queue in the incoming call waiting queue, wherein the queue position may be the original index of the first short-term task in the incoming call waiting queue.

[0075] For example, the incoming call waiting queue may be [B (buy cigarettes 5s), inserted times (1), E (buy cigarettes 5s), inserted times (2), A (other questions 120s), inserted times (2), D (checkout 60s)]. Then B's queue position may be 0.

[0076] The sixth sub-step is to insert the long-time task into the queue position of the incoming call waiting queue.

[0077] For example, the incoming call waiting queue can be [B (buy cigarettes 5s), inserted times (1), E (buy cigarettes 5s), inserted times (2), A (other questions 120s), inserted times (2), D (checkout 60s)]. B's queue position can be 0. The long-term task can be A. Then A is inserted before B. The incoming call waiting queue can be [A (other questions 120s), inserted times (2), B (buy cigarettes 5s), inserted times (0), E (buy cigarettes 5s), inserted times (1), D (checkout 60s)].

[0078] The third step is to determine that the incoming call waiting queue after the withdrawal operation is the above-mentioned waiting queue.

[0079] For example, the original incoming call waiting queue may be [B (buy cigarettes 5s), inserted times (1), E (buy cigarettes 5s), inserted times (2), A (other questions 120s), inserted times (2), H (buy water 5s), inserted times (1), D (checkout 60s), inserted times (2)]. Then, the incoming call waiting queue after the withdrawal operation may be [A (other questions 120s), inserted times (2), B (buy cigarettes 5s), inserted times (0), E (buy cigarettes 5s), inserted times (1), H (buy water 5s), inserted times (1), D (checkout 60s), inserted times (2)].

[0080] Steps 1 through 3 described above, as an inventive feature of this disclosure, address the first technical issue mentioned in the background art: "urgent or short-duration tasks are not prioritized, resulting in severe queues during peak hours and low resource utilization for manual customer service." The factors contributing to this technical issue are often as follows: during peak hours, short-duration tasks surge, and short-duration tasks are inserted before long-duration tasks and processed as urgent tasks, causing long-duration tasks to remain in the queue for extended periods, exacerbating overall queue congestion. Therefore, this disclosure designs a dynamic backoff scheme. First, the long-duration tasks in the queue are identified to provide targets for task priority adjustment. Second, the waiting time and number of insertions of long-duration tasks are monitored to monitor fairness and queue-jumping intensity. Then, a backoff mechanism is implemented based on the length of the waiting time, removing short-duration tasks that have been queued before long-duration tasks and moving the long-duration tasks to the position of the first short-duration task. The queue order is reset to prevent long-duration tasks from being continuously queued. Finally, the queue after the backoff operation is used as the customer service processing queue. Therefore, while ensuring that short-term tasks are prioritized, the problem of long-term tasks being unable to be processed due to queue jumping, which can lead to queue congestion, is resolved. Through a dynamic fallback mechanism, the customer service terminal system's risks are reduced, ensuring the continuity and overall efficiency of customer service problem-solving services.

[0081] Step 105: Send the waiting queue and the operation queue to the task processing terminal corresponding to the manual customer service to control the unmanned convenience store equipment to perform the operation task.

[0082] In some embodiments, the execution entity may send the waiting queue and the operation queue to the task processing terminal corresponding to the manual customer service to control the unmanned convenience store equipment to perform operation tasks. Among them, the operation tasks include: product distribution and access control. The task processing terminal corresponding to the manual customer service may be an interactive interface used by the manual customer service, which is used to receive and process queue tasks and support parallel processing of incoming tasks and operation tasks. The unmanned convenience store equipment may be a physical device connected via the Internet of Things (for example, access control, vending machines, and surveillance cameras), which can realize remote control of the equipment.

[0083] For example, tasks in the waiting queue are displayed so that human agents can prioritize them. For operation tasks in the operation queue, device control instructions can be generated based on the task tags, making it easier for human agents to select and execute the operation tasks.

[0084] While employing technical solutions to address the aforementioned technical issue, the following issues often arise: Confusion in task priorities: Traditional customer service terminals cannot intuitively distinguish the urgency of tasks, resulting in critical tasks (short-term emergency operations) being easily blocked by longer-term tasks. Multi-task parallel processing conflicts: While processing the main task, tasks controlling device operations are difficult to execute simultaneously, which can easily lead to operational blockages or lags in interactive interfaces.

[0085] Conventional solutions to these problems typically involve splitting operational tasks into separate windows and assigning them to human customer service representatives. However, the inventors considered that this would require frequent manual switching between terminal pages, performing tasks across multiple terminals, and potentially requiring a transfer process for long-duration tasks. Furthermore, this would be time-consuming and require coordinating multiple resources, increasing labor costs. Therefore, we can adopt the following solutions:

[0086] In some optional implementations of some embodiments, the execution entity may send the waiting queue and the operation queue to a task processing terminal corresponding to a manual customer service representative to control the unmanned convenience store device to perform the operation task, which may include the following steps:

[0087] The first step is to display the waiting queue in the main interactive area of ​​the task processing terminal. The short-term tasks in the main interactive area are located at the top of the interface as the highest priority tasks, highlighted with a red border, and the long-term tasks are located at the bottom of the interface as the second-priority tasks, highlighted with a yellow border. The main interactive area can be the core interface area of ​​the customer service terminal, used to display waiting queues that require continuous processing. The highest-priority tasks can be the short-term tasks displayed at the top of the interface, tasks that require immediate processing by human customer service. The second-priority tasks can be long-term tasks among the short-term tasks that require sequential processing by human customer service.

[0088] The second step is to display the above-mentioned operation queue in the side operation area of ​​the above-mentioned task processing terminal, wherein the tasks in the above-mentioned side operation area are used as immediate execution tasks. The above-mentioned immediate execution tasks are linked to the store equipment control interface, and the customer service can trigger the device to respond to instructions and execute the operation tasks with one click. The above-mentioned side operation area can be the sidebar area of ​​the customer service terminal, which is used to display device control tasks that can be executed with one click. The above-mentioned immediate execution tasks can be operation tasks that are linked to the device interface (such as opening a door, distributing goods), which can trigger the device to respond and process the operation tasks immediately after clicking. The above-mentioned store equipment control interface can be the HTTP / MQTT interface provided by the unmanned convenience store equipment for remote control.

[0089] In the third step, in response to the human customer service processing the highest-priority task in the main interactive area, the task processing terminal is controlled to synchronously execute the immediately executed task via an asynchronous thread. This synchronous execution can occur while the human customer service is processing the main task and triggering device operations in the background, without interfering with each other. Asynchronous threads can be threads that execute in parallel in the background to avoid blocking the main interface interaction. In practice, the same interfaces or methods used in browsers to call processing devices can be used, such as Web Workers.

[0090] Step 4: In response to the manual customer service handling of the secondary priority task in the primary interaction area, perform the following steps:

[0091] The first sub-step is to control the task processing terminal in response to the triggering of the above-mentioned immediate execution task to pause the above-mentioned secondary priority task processing interface. The pause can be to temporarily freeze the current page to prevent operation conflicts. In practice, methods for saving task progress (for example, Redux and Vuex) can be used, and a pop-up window for device operation can be switched.

[0092] The second sub-step is to control the task processing terminal to save the processing progress information of the second priority task and display the pop-up button corresponding to the immediate execution task operation instruction, so as to call the device interface to complete the immediate execution task operation. The processing progress information can be the current task status (for example, the completed form, the communicated content). In practice, the progress data can be temporarily stored locally. The pop-up button can be an operation button in a floating window, which is used to trigger the device instruction. In practice, the pop-up button can be implemented using the corresponding pop-up component. For example, React Modal and Vue Dialog components.

[0093] Step 5: In response to the completion of the instant execution task, the task processing terminal is controlled to update the operation queue status. In practice, the operation task status is marked and removed from the operation queue based on the execution result (success / failure) of the device.

[0094] The first to fifth steps mentioned above, as an inventive point of the present disclosure, solve the technical problem one mentioned in the background technology, "urgent or short-time processing tasks cannot be given priority, there are serious queues during peak hours, and the resource utilization rate of manual customer service is low." The factors that lead to the above technical problems are often as follows: splitting the operation tasks into independent windows and assigning relevant manual customer service processing. In the process of executing tasks, cross-terminal execution is required. When necessary, tasks need to be transferred, which increases labor costs and may also lead to incomplete information transmission and reduce the processing efficiency of manual customer service. Therefore, the present disclosure designs a single-terminal integration solution. Visualization of different task priorities is achieved on a unified interface. Asynchronous parallel processing of multiple tasks and one-click control linkage of operation tasks calling related equipment improve the efficiency of manual customer service in processing tasks.

[0095] The aforementioned embodiments of the present disclosure have the following beneficial effects: The task execution methods applied to unmanned convenience stores, as described in some embodiments of the present disclosure, reduce the time it takes for human customer service staff to handle issues, significantly shorten customer waiting times, and improve the efficiency of human customer service. Specifically, the long wait times for customer service staff to handle issues are due to the fact that, in unmanned convenience stores, a single human customer service representative must handle call requests from multiple stores. During peak hours, customers may queue up and wait for service for extended periods, degrading the user shopping experience. Based on this, the task execution methods applied to unmanned convenience stores, as described in some embodiments of the present disclosure, first obtain conversation records from user calls to an intelligent customer service hotline that are transferred to human service, thereby generating a conversation record set. This provides complete user problem information, providing a data foundation for subsequent intent classification and task priority determination. Then, using a pre-trained large language model, label matching is performed on each conversation record in the conversation record set to generate conversation labels, resulting in a conversation label set. This automatically identifies the type of user issue, resolving the issue of delayed updates to the human-dependent database. Secondly, using the task queue corresponding to the human customer service representative, each conversation label in the conversation label set is assigned a corresponding task queue, resulting in an incoming call waiting queue and an operation queue. This separates operational tasks from inbound call tasks, preventing long-duration tasks from blocking the processing of shorter tasks and improving customer service resource utilization. Next, the incoming call waiting queue is sequentially optimized to generate a waiting queue. By interrupting and retracting short-duration tasks, urgent tasks or those that are shorter than other tasks can be optimized, reducing queue times during peak hours. Finally, the waiting queue and operational queue are sent to the corresponding task processing terminal for human customer service to control the unmanned convenience store equipment to execute operational tasks, including product distribution and door unlocking. This allows customer service to process inbound call tasks and one-click operational tasks (such as remote door unlocking) in parallel, shortening problem resolution time, reducing the frequency of referrals to specialized departments, and improving overall service efficiency. In summary, by analyzing user call records and leveraging a large model, label matching and processing time estimation are achieved. The human customer service task queue is prioritized and sorted, enabling multi-task parallel processing. This shortens user wait times and improves human customer service efficiency.

[0096] Further references Figure 2 As an implementation of the methods shown in the above figures, the present disclosure provides some embodiments of a task execution device for an unmanned convenience store. These device embodiments are similar to Figure 1 Corresponding to the method embodiments shown, the task execution device applied to unmanned convenience stores can be specifically applied to various electronic devices.

[0097] like Figure 2As shown, a task execution device 200 for use in unmanned convenience stores includes: a conversation acquisition unit 201, a label matching unit 202, a queue allocation unit 203, a queue optimization unit 204, and an operation processing unit 205. The conversation acquisition unit 201 is configured to acquire conversation records when a user dials an intelligent customer service line and is transferred to manual service, thereby obtaining a conversation record set. The label matching unit 202 is configured to use a pre-trained large language model to perform label matching on each conversation record in the conversation record set using a preset label library to generate conversation labels, thereby obtaining a conversation label set. The queue allocation unit 203 is configured to use the task queue corresponding to the manual customer service to assign a corresponding task queue to each conversation label in the conversation label set, thereby obtaining an incoming call waiting queue and an operation queue. The queue optimization unit 204 is configured to perform sequential optimization on the incoming call waiting queue to obtain a waiting queue. The operation processing unit 205 is configured to: send the above-mentioned waiting queue and the above-mentioned operation queue to the task processing terminal corresponding to the manual customer service to control the unmanned convenience store equipment to perform operation tasks, wherein the above-mentioned operation tasks include: product distribution and access control opening.

[0098] It is understandable that the units described in the task execution device 200 for unmanned convenience stores are similar to those described in the reference Figure 1 Therefore, the operations, features and beneficial effects described above for the method are also applicable to the task execution device 200 used in unmanned convenience stores and the units contained therein, and will not be repeated here.

[0099] Reference below Figure 3 , which shows a structural schematic diagram of an electronic device (eg, an electronic device) 300 suitable for implementing some embodiments of the present disclosure. Figure 3 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0100] like Figure 3 As shown, the electronic device 300 may include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 301, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 302 or a program loaded from a storage device 308 into a random access memory (RAM) 303. Various programs and data required for the operation of the electronic device 300 are also stored in the RAM 303. The processing device 301, the ROM 302, and the RAM 303 are connected to each other via a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0101] Typically, the following devices may be connected to the I / O interface 305: an input device 306 including, for example, a touch screen, a touchpad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 307 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 308 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 309. The communication device 309 may allow the electronic device 300 to communicate with other devices wirelessly or by wire to exchange data. Figure 3 The electronic device 300 is shown with various devices, but it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed instead. Figure 3 Each block shown in the figure may represent one device, or may represent multiple devices as needed.

[0102] In particular, according to some embodiments of the present disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, some embodiments of the present disclosure include a computer program product comprising a computer program carried on a computer-readable medium, the computer program comprising program code for executing the method shown in the flowchart. In some such embodiments, the computer program can be downloaded and installed from a network via the communication device 309, or installed from the storage device 308, or installed from the ROM 302. When the computer program is executed by the processing device 301, the above-mentioned functions defined in the method of some embodiments of the present disclosure are performed.

[0103] It should be noted that in some embodiments of the present disclosure, the computer-readable medium mentioned above may be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In some embodiments of the present disclosure, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, device, or device. In some embodiments of the present disclosure, the computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium may be transmitted using any suitable medium, including but not limited to wires, optical cables, RF (radio frequency), etc., or any suitable combination thereof.

[0104] In some embodiments, the client and server can communicate using any currently known or future developed network protocol, such as HTTP (HyperText Transfer Protocol), and can be interconnected with any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), an internet (e.g., the Internet), and a peer-to-peer network (e.g., an ad hoc peer-to-peer network), as well as any currently known or future developed network.

[0105] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device. The computer-readable medium carries one or more programs. When the one or more programs are executed by the electronic device, the electronic device: obtains the conversation record of the user who dials the intelligent customer service phone and is transferred to the manual service, thereby obtaining a conversation record set; uses a pre-trained large language model through a preset tag library to perform tag matching processing on each conversation record in the conversation record set to generate a conversation tag, thereby obtaining a conversation tag set; assigns a corresponding task queue to each conversation tag in the conversation tag set through the task queue corresponding to the manual customer service, thereby obtaining an incoming call waiting queue and an operation queue; performs sequential optimization on the incoming call waiting queue to obtain a waiting queue; and sends the waiting queue and the operation queue to the task processing terminal corresponding to the manual customer service to control the unmanned convenience store equipment to perform operation tasks, wherein the operation tasks include: product distribution and door opening.

[0106] Computer program code for performing the operations of some embodiments of the present disclosure may be written in one or more programming languages, or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0107] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, program segment, or a part of code, and the module, program segment, or a part of code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order than that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of the boxes in the block diagram and / or flowchart, can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0108] The units described in some embodiments of the present disclosure may be implemented in software or in hardware. The described units may also be provided in a processor. For example, they may be described as: a processor including a conversation acquisition unit, a label matching unit, a queue allocation unit, a queue optimization unit, and an operation processing unit. The names of these units do not, in some cases, constitute limitations on the units themselves. For example, the conversation acquisition unit may also be described as a "unit that obtains conversation records when a user dials an intelligent customer service phone and is transferred to manual service, and obtains a conversation record set."

[0109] The functions described above herein may be performed, at least in part, by one or more hardware logic components. For example, and without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chip (SOCs), complex programmable logic devices (CPLDs), and the like.

[0110] The above description is only an illustration of some preferred embodiments of the present disclosure and the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present disclosure is not limited to the technical solutions formed by the specific combination of the above-mentioned technical features, but should also cover other technical solutions formed by any combination of the above-mentioned technical features or their equivalent features without departing from the above-mentioned inventive concept. For example, the above-mentioned features are replaced with (but not limited to) technical features with similar functions disclosed in the embodiments of the present disclosure.

Claims

1. A task execution method applied to an unmanned convenience store, comprising: Obtain the conversation record of the user who dialed the intelligent customer service phone and was transferred to the manual service, and obtain the conversation record set; Using a pre-trained large language model, a tag matching process is performed on each conversation record in the conversation record set using a preset tag library to generate a conversation tag, thereby obtaining a conversation tag set; By using the task queue corresponding to the manual customer service, the corresponding task queue is assigned to each dialogue tag in the dialogue tag set to obtain the incoming call waiting queue and the operation queue; Optimizing the order of the incoming call waiting queue to obtain a waiting queue; The waiting queue and the operation queue are sent to the task processing terminal corresponding to the manual customer service to control the unmanned convenience store equipment to perform operation tasks, wherein the operation tasks include: product distribution and door opening.

2. The method according to claim 1, wherein The method of performing label matching processing on each conversation record in the conversation record set by using a pre-trained large language model through a preset label library to generate a conversation label includes: Using the large language model, the conversation record is subjected to label classification to obtain a first-level classification, wherein the first-level classification is one of the following: an operation task label, an incoming call task label; In response to the first-level classification being in the preset tag library, the following steps are performed: In response to the first-level classification as an operation-type task tag, generating an operation-type tag and an estimated processing time according to the preset tag library, and using the operation-type tag as a conversation tag; In response to the first-level classification as an incoming call task tag, generating an incoming call tag and an estimated processing time according to the preset tag library, and using the incoming call tag as a conversation tag; In response to the first-level classification not being in the preset label library, generating candidate labels corresponding to the first-level classification by using a target clustering algorithm; In response to the candidate tag passing the manual review, the candidate tag is used as a conversation tag and updated to the preset tag library.

3. The method according to claim 1, wherein The task queue corresponding to the manual customer service is used to assign a corresponding task queue to each dialogue tag in the dialogue tag set, thereby obtaining an incoming call waiting queue and an operation queue, including: In response to the dialog tag being a customer service continuous interaction task, allocating the dialog tag and the estimated processing time corresponding to the dialog tag to an incoming call waiting queue; In response to the conversation tag being a task that can be completed by the customer service in one time, the conversation tag and the estimated processing time corresponding to the conversation tag are allocated to an operation queue.

4. The method according to claim 1, wherein The step of sequentially optimizing the incoming call waiting queue to obtain the waiting queue includes: Traverse the incoming call waiting queue from front to back, mark each short-term task, and obtain a short-term task queue; For each short-term task in the short-term task queue, perform the following insertion steps: Determine the position of the short-term task in the incoming call waiting queue as a target position; In response to a long-duration task existing before the target position in the incoming call waiting queue, determining the waiting time and the number of times the long-duration task has been inserted, wherein the waiting time is a time during which the long-duration task remains in the incoming call waiting queue; In response to the number of times the long task has been inserted being less than a preset number and the waiting time not exceeding a preset threshold, inserting the short task before the corresponding sequence position of the long task; In response to the waiting time exceeding a preset threshold, inserting the short-term task into the incoming call waiting queue after the corresponding sequence position of the long-term task; The incoming call waiting queue after the insert operation is determined to be the waiting queue.

5. A task execution device for an unmanned convenience store, comprising: A conversation acquisition unit is configured to acquire conversation records when a user dials an intelligent customer service number and is transferred to a manual service, thereby obtaining a conversation record set; a label matching unit configured to perform label matching processing on each conversation record in the conversation record set using a pre-trained large language model through a preset label library to generate a conversation label to obtain a conversation label set; The queue allocation unit is configured to allocate a corresponding task queue to each dialog tag in the dialog tag set through the task queue corresponding to the manual customer service, thereby obtaining an incoming call waiting queue and an operation queue; a queue optimization unit configured to sequentially optimize the incoming call waiting queue to obtain a waiting queue; The operation processing unit is configured to send the waiting queue and the operation queue to the task processing terminal corresponding to the manual customer service to control the unmanned convenience store equipment to perform operation tasks, wherein the operation tasks include: product distribution and access control opening.

6. An electronic device comprising: one or more processors; a storage device having one or more programs stored thereon, When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 4.

7. A computer-readable medium having a computer program stored thereon, wherein: When the program is executed by a processor, the method according to any one of claims 1 to 4 is implemented.

Images