Customer demand work order processing method, device and equipment based on AI voice interaction

Abstract

The invention provides a customer demand work order processing method, device and equipment based on AI voice interaction, and the method comprises the steps: transmitting a voice stream to an ASR engine in real time, and outputting text data with a timestamp; performing intention classification on the text data, identifying a user core request, and generating a to-be-processed work order according to the user core request; determining the priority level of the to-be-processed work order according to a level calculation formula and the work order information of the to-be-processed work order; adding the to-be-processed work orders into a global work order queue, and sorting the to-be-processed work orders according to the priorities to obtain a corresponding sorting result; and after entering the dispatching process, if there is no guest room service robot which can be scheduled and is adaptive to the execution demand of the to-be-processed work order, screening out an optimal executor with the highest matching degree from the on-duty employees according to a preset intelligent dispatching strategy and the to-be-processed work order, and completing the automatic dispatching of the to-be-processed work order. According to the invention, the overall efficiency and accuracy of hotel customer demand work order processing can be significantly improved.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and in particular to a customer work order processing method, apparatus and equipment based on AI voice interaction. Background Technology

[0002] In the hotel guest service sector, the efficiency and accuracy of work order processing directly impact customer experience. Current technologies largely rely on manual answering, creation, and assignment of guest work orders, resulting in inefficiency and response delays. While some systems incorporate work order management modules, they lack intelligent AI voice interaction capabilities, requiring users to manually submit requests. This is not only cumbersome but also prone to errors in work order creation due to inaccurate intent recognition. Furthermore, the work order assignment process lacks intelligent adaptation logic for scheduling room service robots, and manual assignment struggles to accurately match employee skills with work order requirements, frequently leading to resource mismatch and uneven employee workload. This reduces service efficiency and easily causes customer dissatisfaction and employee fatigue.

[0003] Therefore, the existing customer work order processing methods suffer from low processing efficiency. Summary of the Invention

[0004] This invention provides a customer work order processing method, apparatus, and equipment based on AI voice interaction, aiming to solve the problem of low processing efficiency in existing customer work order processing methods.

[0005] In a first aspect, embodiments of the present invention provide a customer service order processing method based on AI voice interaction, the method comprising: The intercepted audio stream is fed into the ASR engine in real time, and the output is text data with timestamps. The text data is classified using an NLU model to identify the core user requests from a predefined hotel scenario intent library, and a work order to be processed is generated based on the core user requests. The priority level of the work order to be processed is determined according to the preset priority calculation formula and the work order information of the work order to be processed. The pending work orders are added to the global work order queue, and the pending work orders are sorted according to the priority level to obtain the corresponding sorting result. The dispatch process of the pending work orders is then started according to the sorting result. Once the dispatch process begins, if there are no available guest room service robots that can be scheduled and are suitable for the execution of the pending work order, the optimal executor with the highest matching degree will be selected from the on-duty employees according to the preset intelligent dispatch strategy and the pending work order, and the automatic dispatch of the pending work order will be completed.

[0006] Secondly, embodiments of the present invention also provide a customer work order processing device based on AI voice interaction, the device comprising: The output unit is used to send the intercepted voice stream into the ASR engine in real time and output text data with timestamps. The identification unit is used to classify the text data by means of an NLU model, identify the core user request from a predefined hotel scenario intent library, and generate a work order to be processed based on the core user request. The determining unit is used to determine the priority level of the work order to be processed based on a preset level calculation formula and the work order information of the work order to be processed. The starting unit is used to add the pending work order to the global work order queue, sort the pending work orders according to the priority level, obtain the corresponding sorting result, and start the dispatch process of the pending work order according to the sorting result. The dispatch unit is used to automatically dispatch the work order to the on-duty employees if there are no available guest room service robots that can be scheduled and are suitable for the execution of the work order to be processed after entering the dispatch process.

[0007] Thirdly, embodiments of the present invention also provide an electronic device, which is a server or a terminal. The electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. The characteristic is that when the processor corresponding to the server and the processor corresponding to the terminal execute the computer program simultaneously, the customer work order processing method based on AI voice interaction as described in any one of claims 1-7 is implemented.

[0008] Fourthly, embodiments of the present invention also provide a computer-readable storage medium storing a computer program, the computer program including program instructions that, when executed by a processor, can implement the method described in the first aspect.

[0009] This invention provides a method, apparatus, and device for processing customer service work orders based on AI voice interaction. The method includes: sending intercepted voice streams into an ASR engine in real time and outputting timestamped text data; classifying the text data using an NLU model to identify the core user request from a predefined hotel scenario intent library and generating a work order to be processed based on the core user request; determining the priority level of the work order to be processed based on a preset priority calculation formula and the work order information of the work order to be processed; adding the work order to be processed to a global work order queue, sorting the work orders to be processed according to the priority level, obtaining a corresponding sorting result, and initiating the dispatch process of the work order to be processed based on the sorting result; after entering the dispatch process, if there is no available room service robot that can be scheduled and is suitable for the execution requirements of the work order to be processed, then according to a preset intelligent dispatch strategy and the work order to be processed, selecting the optimal executor with the highest matching degree from the on-duty employees to complete the automatic dispatch of the work order to be processed. This invention enables intelligent creation of guest work orders through AI voice interaction (ASR+NLU), improving the accuracy of intent recognition and the efficiency of work order generation. By combining priority sorting and intelligent dispatching strategies, it achieves efficient collaborative scheduling between room service robots and on-duty staff, reducing work order response delays and optimizing resource allocation. This not only enhances the customer service experience but also balances staff workload, significantly improving the overall efficiency and accuracy of hotel guest work order processing. Attached Figure Description

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

[0011] Figure 1 A flowchart illustrating the customer work order processing method based on AI voice interaction provided in an embodiment of the present invention; Figure 2 A schematic block diagram of a customer work order processing device based on AI voice interaction provided in an embodiment of the present invention; Figure 3 A schematic block diagram of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0012] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0013] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0014] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0015] It should also be further understood that the term "and / or" as used in this specification and appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes such combinations. Embodiments of this invention provide a customer service order processing method, apparatus, and device based on AI voice interaction. For details on this AI voice interaction-based customer service order processing method, please refer to... Figure 1 , Figure 1 This is a flowchart illustrating a customer work order processing method based on AI voice interaction, provided in an embodiment of the present invention. The method is applied to the control terminal of a customer work order processing system to achieve intelligent processing of customer work orders.

[0016] Figure 1 This is a flowchart illustrating the customer work order processing method based on AI voice interaction provided in an embodiment of the present invention. Figure 1 As shown, the method includes the following steps S110-S150.

[0017] S110: The intercepted voice stream is sent to the ASR engine in real time, and the text data with timestamps is output.

[0018] In this embodiment, after a customer initiates a call via the guest room phone, the AI ​​voice gateway intercepts the call via SIP signaling, triggers the AI ​​robot to automatically answer within 1 second, and broadcasts a guiding message (such as "Hello, I am the front desk AI butler, please speak"); then, the user's voice stream is sent in real time to the privately deployed (hotel domain isolated) ASR engine, and outputs text data with timestamps.

[0019] S120. The text data is classified by NLU model to identify the user's core request from the predefined hotel scenario intent library, and a work order to be processed is generated based on the user's core request.

[0020] In this embodiment, the NLU (Natural Language Understanding) model is constructed using a BERT model fine-tuned based on the hotel scenario. The NLU model performs intent classification on the text data, identifies the user's core requests from a predefined hotel scenario intent library, such as "delivery", "repair request", "complaint", etc., and generates a work order to be processed based on the user's core requests. Furthermore, a unique work order ID is assigned to the work order to achieve full-process traceability.

[0021] Furthermore, the text data is classified using an NLU model to identify the user's core requests from a predefined hotel scenario intent library, and the corresponding core request confidence score is output synchronously. The core request confidence score is used to quantify the accuracy of the identification results. For example, when the core request confidence score is lower than 0.85, or when key fields (such as room number) are missing, a secondary confirmation dialogue with the AI ​​robot is directly triggered.

[0022] Furthermore, if the user's core request is "transfer to a human operator", then the user's call will be directly transferred to a human operator for processing.

[0023] In one embodiment, after step S120, the method further includes: extracting key entities related to the user's core request through preset slot extraction rules, generating a structured slot filling result and a corresponding first confidence score; performing fusion judgment on the text data and the speech stream based on a multimodal fusion network, and outputting an emotion label and a corresponding second confidence score.

[0024] In this embodiment, the slot extraction rules pre-define core slots for hotel service scenarios (such as room number, service type, item, time requirement, etc.). For the user's core request, the Named Entity Recognition (NER) technology combined with the Conditional Random Field (CRF) layer is used to fill the slots, extracting key entities related to the user's core intent such as "item", "quantity", and "location" and filling them into the corresponding slots, generating a structured slot filling result (such as "room number: 802, service type: delivery, item: mineral water") and the corresponding first confidence score.

[0025] Meanwhile, the text data and the speech stream are fused and judged based on the multimodal fusion network, and the emotion labels (such as "neutral", "angry", "anxious", "satisfied" etc.) and the second confidence score are output to provide a basis for subsequent work order priority determination.

[0026] In one embodiment, the step of fusing and determining the text data and the speech stream based on a multimodal fusion network to output an emotion label and a corresponding second confidence score includes: extracting vector representations of emotion-related words from the text data using a word embedding model to generate emotion word vectors; performing acoustic feature analysis on the speech stream to extract audio features that reflect emotional tendencies; and fusing and determining the emotion word vectors and the audio feature inputs based on a multimodal fusion network to output an emotion label and a corresponding second confidence score.

[0027] In this embodiment, a word embedding model is used to extract vector representations of emotion-related words (such as "anxious" and "very satisfied") from the text data to generate emotion word vectors. Acoustic feature analysis is performed on the speech stream to extract MFCC (Mel-Frequency Cepstral Coefficients) coefficients and pitch changes. Combined with prosodic features such as speech rate and volume, audio features strongly correlated with emotions are generated. A multimodal fusion network is constructed based on a pre-trained speech emotion recognition (SER) component (such as a HuBERT variant). The emotion word vectors and the audio features are input into the multimodal fusion network. After fusion and judgment, emotion labels such as "neutral", "angry", "anxious", and "satisfied" and their corresponding second confidence scores (based on softmax probability calculation) are output.

[0028] Furthermore, if the second confidence score is below 0.7, the emotion label is marked as "neutral" to avoid misjudgment.

[0029] In one embodiment, after fusing and determining the text data and the speech stream based on the multimodal fusion network and outputting the emotion label and the corresponding second confidence score, the method further includes: if the emotion label and the corresponding second confidence score meet the preset priority processing conditions, then transferring the user's call to a human operator.

[0030] In this embodiment, if the emotion tag and the corresponding second confidence score meet the preset priority processing conditions, the user's call will be transferred to a human agent for processing; for example, if the emotion tag is "anger" and the corresponding second confidence score is greater than 0.9, the user's call will be transferred to a human agent for processing.

[0031] S130. Determine the priority level of the work order to be processed according to the preset level calculation formula and the work order information of the work order to be processed.

[0032] In this embodiment, the priority level of the pending work order is determined according to a preset priority calculation formula and the work order information (including sentiment tags, timestamps, work order type, number of key entities, etc.). The priority calculation formula can be expressed as: Priority=Base_Type+Emotion_Bonus+Time_Factor+Entity_Complexity; The priority level is defined as follows: Base_Type is the base score (e.g., 2 points for warranty, 1 point for delivery), Emotion_Bonus is the emotion score (e.g., anger = 1 point, anxiety = 0.5 points, neutral = 0 points), Time_Factor is the time period score (nighttime = 1 point, daytime = 0 points), and Entity_Complexity is the entity complexity score (when the number of extracted entities exceeds 3, Entity_Complexity = 0.5). The final Priority is rounded down to the nearest integer, with a range of 1-3 (1 being the highest level). For example, if a guest complains at night that "the room light is broken and they are very angry," then Base_Type = 2 (repair request), Emotion_Bonus = 1, Time_Factor = 1, Entity_Complexity = 0, Priority = 4, and the highest level of 1 is used.

[0033] Furthermore, when the emotion label is "anger" or "anxiety" and the corresponding second confidence score is greater than 0.8, the highest level judgment will be triggered directly to ensure that urgent needs are responded to first.

[0034] Furthermore, the priority algorithm (level calculation formula) supports dynamic adjustment of core parameters. For example, when the average response time of nighttime work orders exceeds 10 minutes in the past week, the threshold of nighttime-related parameters will be automatically lowered (such as reducing the score bonus weight of nighttime periods) to adapt to the response pressure in actual operation.

[0035] In one embodiment, step S130 includes: obtaining the emotion tag, timestamp, work order type, and number of key entities of the work order to be processed based on the work order information; obtaining the emotion score, time period score, basic score, and entity complexity score corresponding to the emotion tag, the timestamp, the work order type, and the number of key entities, respectively, based on the preset correspondence between each key information item and the score; calculating the emotion score, the time period score, the basic score, and the entity complexity score according to the level calculation formula to obtain the total score; and quantifying and mapping the total score to obtain the priority level of the work order to be processed.

[0036] In this embodiment, the emotion tag, timestamp, work order type, and number of key entities of the work order to be processed are obtained based on the work order information; according to the preset correspondence between each key information item and the score, the emotion score, time period score, basic score, and entity complexity score corresponding to the emotion tag, the timestamp, the work order type, and the number of key entities are obtained respectively; the emotion score, the time period score, the basic score, and the entity complexity score are calculated according to the level calculation formula Priority=Base_Type+Emotion_Bonus+Time_Factor+Entity_Complexity. The total score, Priority, is calculated. Priority represents the priority level, Base_Type represents the base score (e.g., warranty service 2 points, delivery service 1 point), Emotion_Bonus represents the emotion score (e.g., anger = 1 point, anxiety = 0.5 points, neutral = 0 points), Time_Factor represents the time period score (nighttime = 1 point, daytime = 0 points), and Entity_Complexity represents the entity complexity score (when the number of extracted entities exceeds 3, Entity_Complexity = 0.5). The total score is then quantified and mapped to obtain the priority level of the work order to be processed.

[0037] S140. Add the pending work order to the global work order queue, sort the pending work orders according to the priority level, obtain the corresponding sorting result, and start the dispatch process of the pending work order according to the sorting result.

[0038] In this embodiment, the pending work orders are added to the global work order queue, and the pending work orders are sorted according to the priority level to obtain the corresponding sorting results. The assignment process of the pending work orders is then started according to the sorting results to improve the overall efficiency and accuracy of hotel guest work order processing. If there are work orders with the same priority, they are arranged in ascending order of work order creation time.

[0039] S150. After entering the dispatch process, if there is no room service robot available for scheduling and suitable for the execution requirements of the pending work order, the optimal executor with the highest matching degree is selected from the on-duty employees according to the preset intelligent dispatch strategy and the pending work order, and the automatic dispatch of the pending work order is completed.

[0040] In this embodiment, after entering the dispatch process, it is first determined whether there are any idle guest room service robots. The guest room service robots must have the execution capability to match the work order to be processed. If the guest room service robot is idle, the work order to be processed is sent to the guest room service robot. If the guest room service robot is not idle, according to the preset intelligent dispatch strategy and the work order to be processed, the optimal executor with the highest matching degree is selected from the on-duty employees to complete the automatic dispatch of the work order to be processed. Specifically, on-duty employees can be screened from dimensions such as skill matching degree, regional adaptability, load balancing, and historical efficiency.

[0041] In one embodiment, step S150 includes: selecting candidate employees with corresponding execution capabilities from on-duty employees according to the work order type of the work order to be processed; selecting the optimal executor from the candidate employees according to the work order information of the work order to be processed and the status information of each candidate employee, and completing the assignment of the work order to be processed.

[0042] In this embodiment, candidate employees with corresponding execution capabilities are selected from the on-duty employees based on the work order type of the work order to be processed; if there are no matching candidate employees, on-duty employees with similar capabilities are included in the candidate employee pool; based on the work order information of the work order to be processed (such as guest room area, priority level, etc.) and the status information of each candidate employee (such as responsible floor, current work order volume, historical service data, etc.), the optimal executor is selected from the candidate employees to complete the assignment of the work order to be processed; specifically, the selection can be based on dimensions such as distance, load, and efficiency.

[0043] Preferably, a heartbeat mechanism is used to update the status information of each employee at regular intervals to ensure data real-time performance.

[0044] In one embodiment, the step of selecting the optimal executor from the candidate employees based on the work order information of the work order to be processed and the status information of each candidate employee includes: calculating the distance factor score, load factor score and status factor score of each candidate employee based on the work order information and the status information; and selecting the optimal executor from the candidate employees based on the distance factor score, load factor score and status factor score of each candidate employee.

[0045] In this embodiment, based on work order information (such as room number) and the status information of candidate employees (such as responsible floor, current work order volume, working hours, etc.), the distance factor score, load factor score, and status factor score of each candidate employee are calculated respectively. Then, based on the distance factor score, load factor score, and status factor score of each candidate employee, the optimal executor is selected in sequence, as follows: 1. Distance factor score: determined based on the matching relationship between the candidate employee's responsible floor and the floor to be processed: 1 point when they are on the same floor, 0.5 points when they are on adjacent floors, and 0 points when they are not on adjacent floors; 2. Load factor score: Calculated based on "current number of work orders being processed / daily work order limit"; 3. Status Factor Score: Calculated based on the saturation calculation formula. The expression of the saturation calculation formula is: Saturation = (Number of completed work orders today / Daily maximum number of work orders) + (Total busy time today / 8 hours of work time).

[0046] During the screening process, the range of candidates is narrowed down step by step in the following order: First, select the subset of candidates with the highest distance factor score; if there are still multiple employees in this subset, then select the employee with the best load factor score (smallest value); if there are still multiple candidates, finally select the one with the highest state factor score as the optimal executor.

[0047] After a work order is assigned, if the person responsible for the work order fails to respond and process it within the preset time window, an outbound call will be made to the manager's preset contact information to prompt the manager to intervene in a timely manner. At the same time, the entire process of non-response will be recorded (including the person responsible, the duration of non-response, the time of escalation, and other key information), which will then be used as a basis for KPI assessment of employee work efficiency and response timeliness, urging employees to pay attention to work order response.

[0048] After completing the service, employees can click "Close Work Order" on their mobile devices to trigger the generation of a work order completion signal and synchronize it to the control terminal. The control terminal will then update the status of the corresponding work order and record the closed-loop processing data.

[0049] In summary, this invention enables intelligent creation of guest work orders through AI voice interaction (ASR+NLU), improving the accuracy of intent recognition and the efficiency of work order generation. By combining priority sorting and intelligent dispatching strategies, it achieves efficient collaborative scheduling between room service robots and on-duty staff, reducing work order response delays and optimizing resource allocation. This not only enhances the customer service experience but also balances staff workload, significantly improving the overall efficiency and accuracy of hotel guest work order processing.

[0050] Figure 2 This is a schematic block diagram of a customer work order processing device based on AI voice interaction, provided as an embodiment of the present invention. Figure 2 As shown, corresponding to the above-described customer work order processing method based on AI voice interaction, this invention also provides a customer work order processing device based on AI voice interaction. The device is configured in the control terminal of the customer work order processing system to realize an intelligent processing flow for customer work orders. For details, please refer to... Figure 2 The AI-based voice interaction-based customer work order processing device 700 includes: The output unit 701 is used to send the intercepted voice stream into the ASR engine in real time and output text data with timestamps. The identification unit 702 is used to classify the text data by means of an NLU model, identify the core user request from a predefined hotel scenario intent library, and generate a work order to be processed based on the core user request. The determining unit 703 is used to determine the priority level of the work order to be processed based on a preset level calculation formula and the work order information of the work order to be processed. The starting unit 704 is used to add the pending work order to the global work order queue, sort the pending work orders according to the priority level, obtain the corresponding sorting result, and start the dispatch process of the pending work order according to the sorting result. The dispatch unit 705 is used to automatically dispatch the work order to the on-duty employees if there is no available guest room service robot that can be scheduled and is suitable for the execution of the work order to be processed after entering the dispatch process.

[0051] In some embodiments, after performing the step of classifying the text data using an NLU model and identifying the user's core request from a predefined hotel scenario intent library, the recognition unit 702 is further configured to: Key entities related to the user's core request are extracted using preset slot extraction rules, generating structured slot filling results and corresponding first confidence scores; the text data and the speech stream are fused and judged based on a multimodal fusion network, and emotion tags and corresponding second confidence scores are output.

[0052] In some embodiments, when the recognition unit 702 performs the step of fusing and determining the text data and the speech stream based on a multimodal fusion network, and outputting an emotion label and a corresponding second confidence score, it is specifically used for: The emotional word vectors are generated by extracting vector representations of emotion-related words from the text data through a word embedding model; acoustic feature analysis is performed on the speech stream to extract audio features that reflect emotional tendencies; the emotional word vectors and audio feature inputs are fused and judged based on a multimodal fusion network to output emotion labels and corresponding second confidence scores.

[0053] In some embodiments, after performing the step of fusing and determining the text data and the speech stream based on a multimodal fusion network and outputting an emotion label and a corresponding second confidence score, the recognition unit 702 is further configured to: If the emotion tag and the corresponding second confidence score meet the preset priority processing conditions, the user's call will be transferred to a human operator.

[0054] In some embodiments, when the determining unit 703 performs the step of determining the priority level of the work order to be processed based on a preset level calculation formula and the work order information of the work order to be processed, it is specifically used for: Based on the work order information, obtain the emotion tag, timestamp, work order type, and number of key entities for the work order to be processed; based on the preset correspondence between each key information item and the score, obtain the emotion score, time period score, basic score, and entity complexity score corresponding to the emotion tag, the timestamp, the work order type, and the number of key entities, respectively; calculate the emotion score, the time period score, the basic score, and the entity complexity score according to the level calculation formula to obtain the total score; quantify and map the total score to obtain the priority level of the work order to be processed.

[0055] In some embodiments, when the dispatch unit 705 performs the step of automatically dispatching the work order to select the optimal executor with the highest matching degree from the on-duty employees according to a preset intelligent dispatch strategy and the work order to be processed, it is specifically used for: Based on the work order type of the pending work order, select candidate employees with corresponding execution capabilities from the on-duty employees; based on the work order information of the pending work order and the status information of each candidate employee, select the optimal executor from the candidate employees to complete the assignment of the pending work order.

[0056] In some embodiments, when the dispatch unit 705 performs the step of selecting the optimal executor from the candidate employees based on the work order information of the work order to be processed and the status information of each candidate employee, it is specifically used for: Based on the work order information and the status information, the distance factor score, load factor score and status factor score of each candidate employee are calculated; and the optimal executor is selected from the candidate employees based on the distance factor score, load factor score and status factor score of each candidate employee.

[0057] It should be noted that those skilled in the art can clearly understand that the specific implementation process of the above-mentioned AI voice interaction-based customer work order processing device and its various units can be found in the corresponding descriptions in the aforementioned method embodiments. For the sake of convenience and brevity, these details will not be repeated here.

[0058] The aforementioned customer work order processing device based on AI voice interaction can be implemented as a computer program, which can, for example... Figure 3 It runs on the electronic device shown.

[0059] Please see Figure 3 , Figure 3This is a schematic block diagram of an electronic device provided in an embodiment of the present invention. The electronic device 800 can be a terminal or a server. The terminal can be an electronic device with communication functions. The server can be a standalone server or a server cluster composed of multiple servers.

[0060] See Figure 3 The electronic device 800 includes a processor 802, a memory, and a network interface 805 connected via a system bus 801. The memory may include a non-volatile storage medium 803 and internal memory 804.

[0061] The non-volatile storage medium 803 may store an operating system 8031 ​​and a computer program 8032. The computer program 8032 includes program instructions that, when executed, cause the processor 802 to perform a customer work order processing method based on AI voice interaction.

[0062] The processor 802 provides computing and control capabilities to support the operation of the entire electronic device 800.

[0063] The internal memory 804 provides an environment for the execution of the computer program 8032 in the non-volatile storage medium 803. When the processor 802 corresponding to the server and the processor 802 corresponding to the terminal execute the computer program 8032 at the same time, a customer work order processing method based on AI voice interaction is implemented.

[0064] This network interface 805 is used for network communication with other devices. Those skilled in the art will understand that... Figure 3 The structure shown is merely a block diagram of a portion of the structure related to the present invention and does not constitute a limitation on the electronic device 800 to which the present invention is applied. The specific electronic device 800 may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0065] The processor 802 is used to run a computer program 8032 stored in the memory to perform the following steps: The intercepted voice stream is fed into the ASR engine in real time, outputting timestamped text data. The text data is then classified using an NLU model to identify the core user request from a predefined hotel scenario intent library, and a pending work order is generated based on this request. The priority level of the pending work order is determined according to a preset priority calculation formula and its information. The pending work order is added to a global work order queue and sorted according to its priority level to obtain a ranking result, which then initiates the work order dispatch process. If no room service robot is available for scheduling and suited to the work order's execution requirements, the optimal executor with the highest matching degree is selected from on-duty employees based on a preset intelligent dispatch strategy and the pending work order, thus completing the automatic dispatch of the pending work order.

[0066] In some embodiments, after the processor 802 performs intent classification on the text data using an NLU model and identifies the user's core request steps from a predefined hotel scenario intent library, it further performs the following steps: Key entities related to the user's core request are extracted using preset slot extraction rules, generating structured slot filling results and corresponding first confidence scores; the text data and the speech stream are fused and judged based on a multimodal fusion network, and emotion tags and corresponding second confidence scores are output.

[0067] In some embodiments, when the processor 802 performs the step of fusing and determining the text data and the speech stream based on a multimodal fusion network, and outputting an emotion label and the corresponding second confidence score, the following steps are specifically implemented: The emotional word vectors are generated by extracting vector representations of emotion-related words from the text data through a word embedding model; acoustic feature analysis is performed on the speech stream to extract audio features that reflect emotional tendencies; the emotional word vectors and audio feature inputs are fused and judged based on a multimodal fusion network to output emotion labels and corresponding second confidence scores.

[0068] In some embodiments, after implementing the steps of fusing and determining the text data and the speech stream based on a multimodal fusion network and outputting an emotion label and the corresponding second confidence score, the processor 802 further implements the following steps: If the emotion tag and the corresponding second confidence score meet the preset priority processing conditions, the user's call will be transferred to a human operator.

[0069] In some embodiments, when the processor 802 determines the priority level of the work order to be processed based on a preset priority calculation formula and the work order information of the work order to be processed, the processor 802 specifically implements the following steps: Based on the work order information, obtain the emotion tag, timestamp, work order type, and number of key entities for the work order to be processed; based on the preset correspondence between each key information item and the score, obtain the emotion score, time period score, basic score, and entity complexity score corresponding to the emotion tag, the timestamp, the work order type, and the number of key entities, respectively; calculate the emotion score, the time period score, the basic score, and the entity complexity score according to the level calculation formula to obtain the total score; quantify and map the total score to obtain the priority level of the work order to be processed.

[0070] In some embodiments, when the processor 802 performs the step of automatically assigning the work order to the task by selecting the optimal executor with the highest matching degree from the on-duty employees according to the preset intelligent assignment strategy and the work order to be processed, the specific steps are as follows: Based on the work order type of the pending work order, select candidate employees with corresponding execution capabilities from the on-duty employees; based on the work order information of the pending work order and the status information of each candidate employee, select the optimal executor from the candidate employees to complete the assignment of the pending work order.

[0071] In some embodiments, when the processor 802 selects the optimal executor from the candidate employees based on the work order information of the work order to be processed and the status information of each candidate employee, the processor 802 specifically implements the following steps: Based on the work order information and the status information, the distance factor score, load factor score and status factor score of each candidate employee are calculated; and the optimal executor is selected from the candidate employees based on the distance factor score, load factor score and status factor score of each candidate employee.

[0072] It should be understood that, in this embodiment of the invention, the processor 802 may be a Central Processing Unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0073] It will be understood by those skilled in the art that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program includes program instructions and can be stored in a storage medium, which is a computer-readable storage medium. The program instructions are executed by at least one processor in the computer system to implement the process steps of the embodiments of the above methods.

[0074] Therefore, the present invention also provides a storage medium. This storage medium can be a computer-readable storage medium. The storage medium stores a computer program, wherein the computer program includes program instructions. When executed by a processor, the program instructions cause the processor to perform the following steps: The intercepted voice stream is fed into the ASR engine in real time, outputting timestamped text data. The text data is then classified using an NLU model to identify the core user request from a predefined hotel scenario intent library, and a pending work order is generated based on this request. The priority level of the pending work order is determined according to a preset priority calculation formula and its information. The pending work order is added to a global work order queue and sorted according to its priority level to obtain a ranking result, which then initiates the work order dispatch process. If no room service robot is available for scheduling and suited to the work order's execution requirements, the optimal executor with the highest matching degree is selected from on-duty employees based on a preset intelligent dispatch strategy and the pending work order, thus completing the automatic dispatch of the pending work order.

[0075] In one embodiment, after the processor executes the program instructions to perform intent classification on the text data using an NLU model and identifies the user's core request steps from a predefined hotel scenario intent library, it further performs the following steps: Key entities related to the user's core request are extracted using preset slot extraction rules, generating structured slot filling results and corresponding first confidence scores; the text data and the speech stream are fused and judged based on a multimodal fusion network, and emotion tags and corresponding second confidence scores are output.

[0076] In one embodiment, when the processor executes the program instructions to perform the step of fusing and determining the text data and the speech stream based on a multimodal fusion network, and outputting an emotion label and the corresponding second confidence score, the specific steps are as follows: The emotional word vectors are generated by extracting vector representations of emotion-related words from the text data through a word embedding model; acoustic feature analysis is performed on the speech stream to extract audio features that reflect emotional tendencies; the emotional word vectors and audio feature inputs are fused and judged based on a multimodal fusion network to output emotion labels and corresponding second confidence scores.

[0077] In one embodiment, after executing the program instructions to perform the fusion determination of the text data and the speech stream based on the multimodal fusion network, and output the emotion label and the corresponding second confidence score, the processor further performs the following steps: If the emotion tag and the corresponding second confidence score meet the preset priority processing conditions, the user's call will be transferred to a human operator.

[0078] In one embodiment, when the processor executes the program instructions to determine the priority level of the work order to be processed based on a preset priority calculation formula and the work order information of the work order to be processed, the processor specifically implements the following steps: Based on the work order information, obtain the emotion tag, timestamp, work order type, and number of key entities for the work order to be processed; based on the preset correspondence between each key information item and the score, obtain the emotion score, time period score, basic score, and entity complexity score corresponding to the emotion tag, the timestamp, the work order type, and the number of key entities, respectively; calculate the emotion score, the time period score, the basic score, and the entity complexity score according to the level calculation formula to obtain the total score; quantify and map the total score to obtain the priority level of the work order to be processed.

[0079] In one embodiment, when the processor executes the program instructions to automatically assign the work order to the optimal executor with the highest matching degree from the on-duty employees based on a preset intelligent assignment strategy and the work order to be processed, the specific steps are as follows: Based on the work order type of the pending work order, select candidate employees with corresponding execution capabilities from the on-duty employees; based on the work order information of the pending work order and the status information of each candidate employee, select the optimal executor from the candidate employees to complete the assignment of the pending work order.

[0080] In one embodiment, when the processor executes the program instructions to select the optimal executor from the candidate employees based on the work order information of the work order to be processed and the status information of each candidate employee, the processor specifically implements the following steps: Based on the work order information and the status information, the distance factor score, load factor score and status factor score of each candidate employee are calculated; and the optimal executor is selected from the candidate employees based on the distance factor score, load factor score and status factor score of each candidate employee.

[0081] The storage medium can be any computer-readable storage medium capable of storing program code, such as a USB flash drive, portable hard drive, read-only memory (ROM), magnetic disk, or optical disk.

[0082] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0083] In the several embodiments provided by this invention, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For example, the division of each unit is merely a logical functional division, and there may be other division methods in actual implementation. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed.

[0084] The steps in the method of this invention can be adjusted, merged, or reduced in order according to actual needs. The units in the device of this invention can be merged, divided, or reduced according to actual needs. Furthermore, the functional units in the various embodiments of this invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0085] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause an electronic device (which may be a personal computer, a terminal, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention.

[0086] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A customer work order processing method based on AI voice interaction, characterized in that, The method includes: The intercepted audio stream is fed into the ASR engine in real time, and the output is text data with timestamps. The text data is classified using an NLU model to identify the core user requests from a predefined hotel scenario intent library, and a work order to be processed is generated based on the core user requests. The priority level of the work order to be processed is determined according to the preset priority calculation formula and the work order information of the work order to be processed. The pending work orders are added to the global work order queue, and the pending work orders are sorted according to the priority level to obtain the corresponding sorting result. The dispatch process of the pending work orders is then started according to the sorting result. Once the dispatch process begins, if there are no available guest room service robots that can be scheduled and are suitable for the execution of the pending work order, the optimal executor with the highest matching degree will be selected from the on-duty employees according to the preset intelligent dispatch strategy and the pending work order, and the automatic dispatch of the pending work order will be completed.

2. The customer work order processing method based on AI voice interaction according to claim 1, characterized in that, After classifying the text data using an NLU model and identifying the core user requests from a predefined hotel scenario intent library, the process also includes: Key entities related to the user's core request are extracted using preset slot extraction rules, and structured slot filling results and corresponding first confidence scores are generated. The text data and the speech stream are fused and determined based on a multimodal fusion network, and the emotion label and the corresponding second confidence score are output.

3. The customer work order processing method based on AI voice interaction according to claim 2, characterized in that, The process of fusing and determining the text data and the speech stream based on a multimodal fusion network, and outputting an emotion label and a corresponding second confidence score, includes: The emotional word vectors are generated by extracting vector representations of emotion-related words from the text data using a word embedding model. Acoustic feature analysis is performed on the speech stream to extract audio features that reflect emotional tendencies; The emotional word vectors and audio feature inputs are fused and determined based on a multimodal fusion network, and the emotional label and corresponding second confidence score are output.

4. The customer work order processing method based on AI voice interaction according to claim 2, characterized in that, After fusing and determining the text data and the speech stream based on a multimodal fusion network, and outputting the emotion label and the corresponding second confidence score, the method further includes: If the emotion tag and the corresponding second confidence score meet the preset priority processing conditions, the user's call will be transferred to a human operator.

5. The customer work order processing method based on AI voice interaction according to claim 1, characterized in that, The step of determining the priority level of the pending work order based on a preset priority calculation formula and the work order information of the pending work order includes: Based on the work order information, obtain the sentiment tag, timestamp, work order type, and number of key entities for the work order to be processed; Based on the preset correspondence between each key information item and the score, obtain the emotion score, time period score, basic score, and entity complexity score corresponding to the emotion tag, the timestamp, the work order type, and the number of key entities, respectively. The total score is obtained by calculating the emotion score, the time period score, the base score, and the entity complexity score according to the level calculation formula. The total score is quantified and mapped to obtain the priority level of the work orders to be processed.

6. The customer work order processing method based on AI voice interaction according to claim 1, characterized in that, The step of automatically assigning the work orders to the designated tasks by selecting the optimal executor with the highest matching degree from the on-duty employees based on a preset intelligent assignment strategy and the work orders to be processed includes: Based on the type of the work orders to be processed, select candidate employees with corresponding execution capabilities from the on-duty employees; Based on the work order information of the pending work orders and the status information of each candidate employee, the optimal executor is selected from the candidate employees to complete the assignment of the pending work orders.

7. The customer work order processing method based on AI voice interaction according to claim 6, characterized in that, The step of selecting the optimal executor from the candidate employees based on the work order information of the work order to be processed and the status information of each candidate employee includes: Based on the work order information and the status information, the distance factor score, load factor score and status factor score of each candidate employee are calculated. The optimal executor is selected from the candidate employees based on their distance factor score, load factor score, and state factor score.

8. A customer work order processing device based on AI voice interaction, characterized in that, The device includes: The output unit is used to send the intercepted voice stream into the ASR engine in real time and output text data with timestamps. The identification unit is used to classify the text data by means of an NLU model, identify the core user request from a predefined hotel scenario intent library, and generate a work order to be processed based on the core user request. The determining unit is used to determine the priority level of the work order to be processed based on a preset level calculation formula and the work order information of the work order to be processed. The starting unit is used to add the pending work order to the global work order queue, sort the pending work orders according to the priority level, obtain the corresponding sorting result, and start the dispatch process of the pending work order according to the sorting result. The dispatch unit is used to automatically dispatch the work order to the on-duty employees if there are no available guest room service robots that can be scheduled and are suitable for the execution of the work order to be processed after entering the dispatch process.

9. An electronic device, the electronic device being a server or terminal, the electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor corresponding to the server and the processor corresponding to the terminal execute the computer program simultaneously, the customer work order processing method based on AI voice interaction as described in any one of claims 1-7 is implemented.

10. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which includes program instructions that, when executed by a processor, cause the processor to perform the customer work order processing method based on AI voice interaction as described in any one of claims 1-7.

Images