Hotel stay service work order automatic distribution and closed-loop management system based on voice AI driving

Abstract

The application discloses a hotel stay service work order automatic distribution and closed-loop management system based on voice AI driving and belongs to the technical field of hotel management. The application comprises a voice interaction module, an AI analysis module, a work order management module, a terminal execution module and a data management module. The modules are clear in division of labor and data intercommunication, simultaneously realize interconnection and intercommunication with three-party systems, convert non-formatted voice information into standardized structured work order data, accurately identify industry professional vocabulary and service demand, extract work order core elements and combine responsible areas, skill labels, work load and other information of responsible terminals to perform dynamic score calculation and priority sorting, simultaneously set a backup pushing and load shunting mechanism, upgrade work order from traditional manual experience dispatching to data-driven intelligent precise dispatching, guarantee high adaptation of work order distribution and service terminals, and improve work order distribution efficiency and response timeliness.

Description

Technical Field

[0001] This invention relates to the field of hotel management technology, and in particular to a voice AI-driven automatic distribution and closed-loop management system for hotel in-stay service work orders. Background Technology

[0002] In the field of hotel stay service management, the traditional work order processing model relies heavily on manual operation. Guests' service requests must be communicated manually via telephone, front desk registration, etc., and staff then manually enter work order information and assign service tasks. This process is not only cumbersome and inefficient, but also prone to information discrepancies due to manual communication and data entry, leading to service errors. At the same time, the information exchange between different service links within the hotel is poor, and work order distribution relies solely on manual experience, failing to scientifically match work orders with staff based on workload, service skills, and assigned areas. This easily leads to work order backlog and resource imbalance, affecting service processing efficiency.

[0003] Furthermore, traditional work order management lacks a dynamic tracking and closed-loop confirmation mechanism throughout the entire process. Work order processing progress cannot be updated in real time, and standardized feedback and confirmation are difficult to achieve after service completion, making it impossible for managers to accurately control the entire service process. Moreover, most existing hotel intelligent systems operate independently and are disconnected from the work order management system. The service capabilities of intelligent devices such as intelligent guest room controls and delivery robots are not fully utilized, and the coordinated linkage between automated service execution and work order management cannot be achieved.

[0004] Meanwhile, the unformatted voice information generated during hotel services is difficult to effectively analyze and utilize, and cannot be transformed into structured data to support management decisions. This results in a low level of digitalization and intelligence in hotel service management, making it difficult to meet the current demand of guests for efficient and personalized services, and also failing to adapt to the development trend of refined hotel management. Summary of the Invention

[0005] The purpose of this invention is to provide an automated distribution and closed-loop management system for hotel in-stay service orders based on voice AI, in order to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an automatic distribution and closed-loop management system for hotel stay service work orders driven by voice AI, including a voice interaction module, an AI parsing module, a work order management module, a terminal execution module, and a data management module;

[0007] The voice interaction module is used to receive voice input information from hotel guests and employees and complete voice output feedback; the AI ​​parsing module is used to perform semantic understanding, intent recognition, and conversion of unformatted information to formatted information on the voice input information; the work order management module is used to automatically generate standardized work orders based on the AI ​​parsing results and complete work order distribution, progress tracking, and status updates according to preset rules; the terminal execution module is used to receive work order information and support voice work order node reporting from the execution terminal; the data management module is used to store, integrate, and analyze the data throughout the entire process.

[0008] Furthermore, the voice interaction module includes a customer voice interaction unit and an employee voice interaction unit;

[0009] The guest voice interaction unit is configured to adapt to the voice interaction scenarios of hotel smart phones, smart voice terminals and mini-program voice terminals, and is used to receive voice requests for inquiries, guest needs, complaints and maintenance.

[0010] The employee voice interaction unit is configured to be compatible with mobile work apps and smart voice terminals. It is used to receive work order execution status reported by voice and query work order information by voice. The voice interaction module continuously recognizes the intent based on the dialogue context during multiple rounds of intelligent human-computer dialogue.

[0011] Furthermore, the AI ​​parsing module incorporates an NLP algorithm model, including a speech-to-text submodule, a semantic understanding submodule, and an information formatting submodule;

[0012] The speech-to-text submodule is configured to filter invalid speech noise and transcribe speech information into speech text; the semantic understanding submodule is configured to perform intent classification, keyword extraction, and entity recognition on the transcribed speech text to determine the service demand type, service object, service location, and service requirements; the information formatting submodule is configured to convert the extracted service demand type, service object, service location, and service requirements into structured data and generate basic work order information.

[0013] Furthermore, the semantic understanding submodule performs intent classification, keyword extraction, and entity recognition on the transcribed speech text, specifically including:

[0014] The transcribed speech-text data is parsed and decomposed into a multi-layer semantic topic tree, with each layer containing several semantic clusters;

[0015] Semantic clustering analysis is performed on the sub-speech text data corresponding to each layer to obtain several semantic cluster sets; each semantic cluster set is matched and assigned to the corresponding semantic cluster.

[0016] The sentences contained in the semantic cluster set in each semantic cluster are segmented to obtain several feature extraction words in the sentences. The lexical semantic features of each feature extraction word and the syntactic position features of each feature extraction word in the sentence are determined.

[0017] Based on a self-developed thesaurus of synonyms for the hotel industry and the lexical semantic features of each feature extraction word, the first semantic similarity between feature extraction words is determined.

[0018] By combining the first semantic similarity and the syntactic position features of each feature extraction word in the sentence, a second fusion similarity between the feature extraction words is determined;

[0019] The feature extraction words with a second similarity higher than a preset similarity threshold are deduplicated and integrated to obtain a core feature word set. The feature extraction words with consecutive positions in the core feature word set are concatenated with the maximum length using the smallest semantic unit of speech and text to obtain fused feature words.

[0020] The fused feature words are cleaned and filtered based on the part-of-speech statistical characteristics of hotel service scenarios, and invalid function words are removed to obtain a set of service keywords;

[0021] Map the keywords in the service keyword set to hotel industry-specific word vectors;

[0022] Calculate the semantic distance between each dedicated word vector in any semantic cluster and the hotel service standard word vector corresponding to that cluster, and select the keyword corresponding to the dedicated word vector with the smallest semantic distance as the core target keyword in any semantic cluster;

[0023] Based on the core target keywords in any semantic cluster, determine the service core keywords contained in each layer of the semantic topic tree, and complete the keyword extraction of speech text.

[0024] Furthermore, the work order management module includes a work order generation submodule, an intelligent distribution submodule, a progress tracking submodule, and a closed-loop confirmation submodule;

[0025] The work order generation submodule is configured to automatically generate standardized work orders containing work order number, requirement type, responsible department, completion deadline, and service recipient information based on the basic work order information output by the AI ​​parsing module; the intelligent distribution submodule is configured to push work orders to the corresponding responsible terminals based on preset distribution rules according to the job responsibilities, employee workload, and service area of ​​each department in the hotel; the progress tracking submodule is configured to obtain the work order status reporting information from the terminal execution module in real time and update the work order status tags, which include pending, processing, and completed; the closed-loop confirmation submodule is configured to automatically send result feedback to the service initiator and obtain confirmation information after the work order is completed, thus completing the work order closure.

[0026] The intelligent distribution submodule is also used to automatically divert work orders to backup execution terminals and send work order diversion reminder information to hotel management personnel when the workload of the responsible department or employee's work order processing terminal exceeds a preset threshold.

[0027] Furthermore, matching weights are assigned to candidate responsible terminals based on work order processing priorities, and dynamic scores are calculated by combining the current processing load of each candidate responsible terminal, including:

[0028] The work order processing priority is converted using a standard score to obtain a priority score.

[0029] Priority dimension weights are set based on work order processing priority. The on-time completion rate of historical work orders with the same processing priority level as the candidate responsibility terminal is obtained. When the on-time completion rate is greater than the preset completion rate threshold, the adaptation weight of the candidate responsibility terminal is determined to be 1. Otherwise, the adaptation weight of the candidate responsibility terminal is determined based on the difference between the preset completion rate threshold and the on-time completion rate.

[0030] The product of the priority dimension weight and the adaptation weight is used as the matching weight for the candidate responsible terminal.

[0031] Based on the job type of the candidate responsibility terminal, a baseline weight is determined. Based on the number of pending work orders of the candidate responsibility terminal, the terminal load rate is determined. According to the rule that is negatively correlated with the load rate, the load dynamic coefficient is determined. Based on the product of the load dynamic coefficient and the baseline weight, the initial load weight of the candidate responsibility terminal is determined. Based on the real-time operation status of the hotel, the scenario coefficient is determined. The product of the scenario coefficient and the initial load weight is used as the load weight of the candidate responsibility terminal.

[0032] The load weight and matching weight are normalized to obtain the standard load weight and standard matching weight.

[0033] Based on the current processing load data of each candidate responsible terminal, the current processing load score is determined, and based on the historical operation data of the candidate responsible terminals, combined with the priority score and matching weight, the dynamic score of the candidate responsible terminals is calculated.

[0034] Furthermore, the intelligent distribution submodule pushes the work order to the corresponding responsible terminal, specifically including:

[0035] Extract the core matching elements contained in the work order. The core matching elements include at least the service request type, the physical area where the service occurs, the work order processing priority, and the required service skills.

[0036] The system retrieves a pre-set responsibility terminal matching library, which stores the responsible area, the types of requests that can be processed, the service skill tags of the terminal operators, the number of work orders currently pending processing, and the processing load of each responsibility terminal.

[0037] The core matching elements of the work order are initially screened and matched with the information in the responsibility terminal matching library to select candidate responsibility terminals that meet the matching of the responsible area, can handle the corresponding demand type, and have the required service skills.

[0038] The candidate responsible terminals are assigned matching weights based on the work order processing priority. At the same time, the current processing load of each candidate responsible terminal is combined to calculate the dynamic score. The lower the load, the higher the score.

[0039] Candidate responsibility terminals are sorted from highest to lowest according to their dynamic scores, and the terminal with the highest score is selected as the target responsibility terminal.

[0040] The system sends a work order push instruction to the target responsible terminal and marks the work order status as distributed. If the target responsible terminal does not receive the work order within a preset time, the system automatically pushes the work order to the next candidate responsible terminal in the ranking and sends a work order receipt reminder message to the administrator.

[0041] Furthermore, the terminal execution module includes a mobile work terminal for hotel staff and an intelligent voice terminal;

[0042] The mobile work terminal supports reporting work order status in both voice and text formats.

[0043] The intelligent voice terminal is a dedicated terminal for the hotel industry. It is configured to support voice reporting of work order node information, voice query of work order details, voice retrieval of service recipients and room status information, and can also activate the terminal camera through voice commands to record and take pictures of the service process. The recorded and photographed multimedia information can be synchronously associated with the corresponding work order and stored in the data management module.

[0044] Furthermore, it also includes third-party system integration modules;

[0045] The third-party system integration module is configured to interconnect with third-party systems, including a hotel management system (PMS), an intelligent guest control system, a delivery robot system, and an intelligent speaker system. It pushes guest-related work order information to the intelligent guest control system and the delivery robot system for automated service execution, and receives service execution feedback information from the third-party system and synchronizes it to the progress information of the corresponding work order.

[0046] Furthermore, based on the current processing load data of each candidate responsible terminal, a current processing load score is determined. Then, based on the historical operational data of the candidate responsible terminals, combined with priority scores and matching weights, a dynamic score for each candidate responsible terminal is calculated, including:

[0047] The maximum number of work orders that a candidate responsible terminal can handle and the number of work orders that are currently pending are obtained from the current processing load data. The average processing time of the candidate responsible terminal for all work orders and the average processing time for historical work orders that are the same as the current work order are also obtained.

[0048] Based on the maximum number of work orders that can be carried, the number of work orders currently pending, the average processing time for all work orders, and the average processing time for historical work orders that are the same as the current work order, the current processing load score is calculated according to the following formula.

[0049] The historical work order response rate, on-time completion rate and service satisfaction score of the candidate responsibility terminals are obtained from the historical operation data of the candidate responsibility terminals. Based on the historical operation data of all candidate responsibility terminals, the average response rate, average on-time completion rate and average satisfaction score of the hotels are obtained. Combined with the priority score and matching weight, the dynamic score of the candidate responsibility terminal is calculated according to the following formula.

[0050] Compared with the prior art, the beneficial effects of the present invention are:

[0051] 1. This invention establishes a modular architecture for voice interaction, AI parsing, work order management, terminal execution, and data management, and adds a third-party system integration module. Each module has a clear division of labor and data interoperability. At the same time, it realizes interconnection with third-party systems such as hotel PMS and intelligent guest room control, breaking down information silos in various aspects of hotel services. This allows the system to be quickly integrated into the hotel's existing intelligent system, realizing intelligent control of the entire process of work orders from initiation, parsing, generation to execution and closure. It also enables automated service execution with the help of third-party intelligent devices, expanding the service coverage and improving the overall intelligent service level of the hotel.

[0052] 2. This invention uses an AI parsing module to perform noise filtering, accurate transcription, and professional semantic analysis on voice information. Combined with a hotel industry-specific thesaurus and word vector analysis, it completes intent classification, keyword extraction, and entity recognition, transforming unformatted voice information into standardized structured work order data. Compared to general voice parsing solutions, this invention is deeply customized for hotel service scenarios, accurately identifying industry-specific terms and service requirements, avoiding recognition bias, and ensuring the accuracy and completeness of basic work order information. This lays a reliable data foundation for subsequent automated work order dispatch and service execution.

[0053] 3. This invention utilizes a multi-dimensional work order matching and distribution mechanism through an intelligent distribution submodule. It extracts core elements of work orders and combines them with information such as the responsible terminal's area of ​​responsibility, skill tags, and workload to dynamically calculate scores and prioritize them. Simultaneously, it sets up a backup push and load sharing mechanism, upgrading work order assignment from traditional manual experience-based dispatch to data-driven intelligent and precise dispatch. This ensures a high degree of compatibility between work order allocation and service terminals, achieving a balanced distribution of workload across terminals, avoiding processing delays caused by work order backlog, significantly improving work order distribution efficiency and response timeliness, and optimizing the allocation of internal hotel service resources. Attached Figure Description

[0054] Figure 1 This is a schematic diagram of the automatic distribution and closed-loop management process of service work orders according to the present invention. Detailed Implementation

[0055] 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 embodiments of the present invention, and not all embodiments. 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.

[0056] Please see Figure 1 The present invention provides the following technical solutions:

[0057] The voice AI-driven hotel in-stay service work order automatic distribution and closed-loop management system includes a voice interaction module, an AI parsing module, a work order management module, a terminal execution module, and a data management module.

[0058] The system includes a voice interaction module for receiving voice input from hotel guests and staff and providing voice output feedback; an AI parsing module for semantic understanding, intent recognition, and conversion of unformatted to formatted information from voice input; a work order management module for automatically generating standardized work orders based on AI parsing results and distributing, tracking progress, and updating status according to preset rules; a terminal execution module for receiving work order information and supporting voice-based work order node reporting from the execution terminal; and a data management module for storing, integrating, and analyzing data throughout the entire process.

[0059] In the above embodiments, by setting up a modular architecture of a voice interaction module, an AI parsing module, a work order management module, a terminal execution module, and a data management module, intelligent control of the entire process of hotel stay service work orders is realized, breaking down information silos in various aspects of hotel services. The voice interaction module enables two-way interaction of voice information across multiple terminals, the AI ​​parsing module completes the conversion of unformatted voice information into formatted data, the work order management module realizes the automated generation and circulation of work orders, the terminal execution module ensures efficient response at the service execution end, and the data management module realizes the accumulation of data throughout the entire process. With voice AI as the core driver, it replaces the cumbersome process of traditional manual input and manual order dispatch, improves the overall efficiency of work order processing, and provides a standardized and regulated management system for hotel stay services, effectively improving the quality of hotel services and guest experience.

[0060] The voice interaction module includes a customer voice interaction unit and an employee voice interaction unit;

[0061] The guest voice interaction unit is configured to adapt to the voice interaction scenarios of hotel smart phones, smart voice terminals and mini-program voice terminals, and is used to receive voice requests for inquiries, guest needs, complaints and maintenance.

[0062] The employee voice interaction unit is configured to be compatible with mobile work apps and smart voice terminals. It is used to receive work order execution status reports via voice and query work order information via voice. The voice interaction module continuously recognizes intent based on the dialogue context during multiple rounds of intelligent human-computer dialogue.

[0063] In the above embodiments, the voice interaction module is divided into guest and employee voice interaction units, realizing personalized interaction adaptation for different user terminals of hotel services. The guest unit covers multiple scenarios for receiving guest voice requests, allowing guests to conveniently initiate service requests through various commonly used terminals, reducing the operational threshold for guests. The employee unit is adapted to the work scenarios of hotel employees, supporting voice reporting and query functions, meeting the actual needs of employees' mobile work. At the same time, the module supports multi-turn intelligent human-computer dialogue and can recognize intent in combination with context, solving the problems of single dialogue and single intent recognition in voice interaction. It can accurately capture the user's real service needs, reduce service errors caused by intent recognition deviations, and make voice interaction not only fit the user's usage habits but also match the employee's workflow, realizing two-way efficient interaction between customer service and internal work, further improving the accuracy and timeliness of service connection.

[0064] The AI ​​parsing module has a built-in NLP algorithm model, including a speech-to-text submodule, a semantic understanding submodule, and an information formatting submodule;

[0065] The speech-to-text submodule is configured to filter out invalid speech noise and transcribe speech information into speech text; the semantic understanding submodule is configured to perform intent classification, keyword extraction, and entity recognition on the transcribed speech text to determine the service demand type, service object, service location, and service requirements; the information formatting submodule is configured to convert the extracted service demand type, service object, service location, and service requirements into structured data and generate basic work order information.

[0066] The semantic understanding submodule performs intent classification, keyword extraction, and entity recognition on the transcribed speech text, specifically including:

[0067] The transcribed speech-text data is parsed and decomposed into a multi-layer semantic topic tree, with each layer containing several semantic clusters;

[0068] Semantic clustering analysis is performed on the sub-speech text data corresponding to each layer to obtain several semantic cluster sets; each semantic cluster set is matched and assigned to the corresponding semantic cluster.

[0069] The sentences contained in the semantic cluster set in each semantic cluster are segmented to obtain several feature extraction words in the sentences. The lexical semantic features of each feature extraction word and the syntactic position features of each feature extraction word in the sentence are determined.

[0070] Based on a self-developed thesaurus of synonyms for the hotel industry and the lexical semantic features of each feature extraction word, the first semantic similarity between feature extraction words is determined.

[0071] By combining the first semantic similarity and the syntactic position features of each feature extraction word in the sentence, the fusion second similarity between feature extraction words is determined;

[0072] The feature words with a second similarity higher than the preset similarity threshold are deduplicated and integrated to obtain a core feature word set. The feature words with consecutive positions in the core feature word set are concatenated with the maximum length using the smallest semantic unit of speech and text to obtain fused feature words.

[0073] Based on the part-of-speech statistical characteristics of hotel service scenarios, the fused feature words are cleaned and filtered to remove invalid function words and obtain a set of service keywords;

[0074] Map the keywords in the service keyword set to hotel industry-specific word vectors;

[0075] Calculate the semantic distance between each dedicated word vector in any semantic cluster and the hotel service standard word vector corresponding to that cluster, and select the keyword corresponding to the dedicated word vector with the smallest semantic distance as the core target keyword in any semantic cluster;

[0076] Based on the core target keywords in any semantic cluster, determine the service core keywords contained in each layer of the semantic topic tree, and complete the keyword extraction of speech text.

[0077] In the above embodiments, the speech-to-text submodule filters noise to ensure transcription accuracy, the semantic understanding submodule focuses on hotel service scenarios to complete intent classification, keyword extraction and entity recognition, accurately locate core service elements, and the information formatting submodule converts the extracted elements into structured data, directly connecting to the work order generation process, realizing seamless connection between voice information and work order data. Compared with general speech parsing solutions, this module is customized for the scenario characteristics of hotel stay services, can accurately identify professional terms and demand types related to hotel services, avoid the recognition bias of general parsing, and ensure the accuracy and completeness of basic work order information.

[0078] The work order management module includes a work order generation submodule, an intelligent distribution submodule, a progress tracking submodule, and a closed-loop confirmation submodule;

[0079] The work order generation submodule is configured to automatically generate standardized work orders containing work order number, requirement type, responsible department, completion deadline, and service recipient information based on the basic work order information output by the AI ​​parsing module; the intelligent distribution submodule is configured to push work orders to the corresponding responsible terminals based on preset distribution rules according to the job responsibilities, employee workload, and service area of ​​each department in the hotel; the progress tracking submodule is configured to obtain the work order status report information from the terminal execution module in real time and update the work order status tags, which include pending, processing, and completed; the closed-loop confirmation submodule is configured to automatically send result feedback to the service initiator and obtain confirmation information after the work order is completed, thus completing the work order closure.

[0080] The intelligent distribution submodule is also used to automatically divert work orders to backup execution terminals when the workload of the responsible department or employee's terminal exceeds a preset threshold, and send work order diversion reminder information to hotel managers; at the same time, it supports managers to manually intervene in work order distribution and perform cross-department and cross-employee dispatch operations on the distributed work orders.

[0081] The intelligent distribution submodule pushes work orders to the corresponding responsible terminals, specifically including:

[0082] Extract the core matching elements contained in the work order. The core matching elements should include at least the service request type, the physical area where the service occurs, the work order processing priority, and the required service skills.

[0083] Retrieve the system's preset responsibility terminal matching library. The responsibility terminal matching library pre-stores the responsible area, the types of requests that can be processed, the service skill tags of the terminal operators, the number of work orders currently pending and the processing load of each responsibility terminal.

[0084] The core matching elements of the work order are initially screened and matched with the information in the responsibility terminal matching library to select candidate responsibility terminals that meet the matching of the responsible area, can handle the corresponding demand type, and have the required service skills.

[0085] The candidate responsible terminals are assigned matching weights based on the work order processing priority. At the same time, the current processing load of each candidate responsible terminal is combined to calculate the dynamic score. The lower the load, the higher the score.

[0086] Candidate responsibility terminals are sorted from highest to lowest according to their dynamic scores, and the terminal with the highest score is selected as the target responsibility terminal.

[0087] The system sends a work order push instruction to the target responsible terminal and marks the work order status as distributed. If the target responsible terminal does not receive the work order within a preset time, the system automatically pushes the work order to the next candidate responsible terminal in the ranking and sends a work order receipt reminder message to the administrator.

[0088] In the above embodiments, the work order generation submodule automatically generates work orders based on standardized data, ensuring the uniformity and standardization of work order formats and facilitating internal hotel management. The intelligent distribution submodule accurately dispatches work orders based on preset rules, and the progress tracking submodule updates the work order status in real time, allowing managers to monitor the work order processing progress at any time. The closed-loop confirmation submodule completes feedback and confirmation after the work order is completed, ensuring that service requests are fulfilled. At the same time, the intelligent distribution submodule has load monitoring and automatic diversion functions, which can dynamically adjust the work order allocation according to the actual workload of each terminal, avoiding processing delays caused by the accumulation of work orders on a single terminal. It can also push diversion reminders to managers in a timely manner, realizing intelligent and dynamic control of work order distribution and ensuring the overall efficiency and balance of work order processing.

[0089] In the above embodiments, the intelligent dispatch submodule achieves precise and scientific dispatching through a step-by-step work order and responsible terminal matching process. The extraction of core matching elements provides clear matching criteria for dispatching. The establishment of a responsible terminal matching database enables systematic management of information on various service terminals in the hotel. Initial screening and matching eliminates terminals that do not meet the conditions, reducing the workload of subsequent matching. Combining priority allocation weights and calculating dynamic scores based on load, dispatching not only considers service adaptability but also takes into account the actual working status of the terminal, avoiding blind dispatching. At the same time, a backup mechanism for work order push is set up. If the target terminal does not receive it in time, it is automatically pushed to the next candidate terminal, and a reminder is sent to the management personnel to ensure that the work order can be received and processed in a timely manner. This upgrades dispatching from traditional manual experience-based dispatching to data-driven intelligent dispatching, significantly improving the accuracy of dispatching and the response speed of work order processing.

[0090] The terminal execution module includes mobile work terminals and intelligent voice terminals for hotel staff;

[0091] The mobile work terminal supports reporting work order status in both voice and text formats;

[0092] The intelligent voice terminal is a dedicated terminal for the hotel industry. It is configured to support voice reporting of work order node information, voice query of work order details, voice retrieval of service recipients and room status information, and can also activate the terminal camera through voice commands to record and take pictures of the service process. The recorded and photographed multimedia information can be synchronously linked to the corresponding work order and stored in the data management module.

[0093] In the above embodiments, the terminal execution module is configured with a mobile work terminal and a hotel-specific intelligent voice terminal, realizing diversified and intelligent operation of the hotel service execution end. The mobile work terminal supports reporting work order status in both voice and text modes, meeting the operational needs of employees in different work scenarios and improving the flexibility of status reporting. It integrates multiple functions such as voice reporting, querying, information retrieval, and multimedia recording, which are tailored to the mobile work needs of hotel employees. The function of activating the camera with voice commands can realize the visual recording of the service process. Multimedia information is synchronously linked to the work order, realizing the traceability of the service process. At the same time, the information collected by the terminal can be synchronized to the data management module in real time, so that every node of the work order processing has data and information support, which facilitates the hotel management personnel to supervise the service execution process.

[0094] This system also includes a third-party system integration module;

[0095] The third-party system integration module is configured to interconnect with third-party systems, including a hotel management system (PMS), an intelligent guest room control system, a delivery robot system, and a smart speaker system. It can retrieve room status and guest check-in information from the hotel management system (PMS) to provide data support for work order generation and distribution; push guest-related work order information to the intelligent guest room control system and the delivery robot system for automated service execution; and receive service execution feedback information from the third-party systems and synchronize it to the progress information of the corresponding work orders.

[0096] In the above embodiments, a third-party system integration module is added to achieve interconnection and interoperability with various mainstream hotel systems. This breaks down the barriers between this system and the hotel's existing management system and smart device system, enabling cross-system data sharing. Room status and guest information retrieved from the PMS system provide accurate basic data for work order generation and distribution, allowing dispatching to accurately match the guest's room and related information. Simultaneously, guest-related work orders are pushed to systems such as intelligent guest room control and delivery robots, automating the execution of some services, replacing manual services, and improving the processing efficiency of simple guest requests. Execution feedback information from the third-party system can be synchronized to the work order, allowing work order progress tracking to cover automated service stages, achieving full-process progress monitoring of the work order.

[0097] In one embodiment, matching weights are assigned to candidate responsible terminals based on work order processing priorities, and dynamic scores are calculated by combining the current processing load of each candidate responsible terminal, including:

[0098] The work order processing priority is converted using a standard score to obtain a priority score.

[0099] Priority dimension weights are set based on work order processing priority. The on-time completion rate of historical work orders with the same processing priority level as the candidate responsibility terminal is obtained. When the on-time completion rate is greater than the preset completion rate threshold, the adaptation weight of the candidate responsibility terminal is determined to be 1. Otherwise, the adaptation weight of the candidate responsibility terminal is determined based on the difference between the preset completion rate threshold and the on-time completion rate.

[0100] The product of the priority dimension weight and the adaptation weight is used as the matching weight for the candidate responsible terminal.

[0101] Based on the job type of the candidate responsibility terminal, a baseline weight is determined. Based on the number of pending work orders of the candidate responsibility terminal, the terminal load rate is determined. According to the rule that is negatively correlated with the load rate, the load dynamic coefficient is determined. Based on the product of the load dynamic coefficient and the baseline weight, the initial load weight of the candidate responsibility terminal is determined. Based on the real-time operation status of the hotel, the scenario coefficient is determined. The product of the scenario coefficient and the initial load weight is used as the load weight of the candidate responsibility terminal.

[0102] The load weight and matching weight are normalized to obtain the standard load weight and standard matching weight.

[0103] Based on the current processing load data of each candidate responsible terminal, the current processing load score is determined, and based on the historical operation data of the candidate responsible terminals, combined with the priority score and matching weight, the dynamic score of the candidate responsible terminals is calculated.

[0104] In this embodiment, the larger the difference between the preset completion rate threshold and the on-time completion rate, the smaller the corresponding adaptation weight. And when the on-time completion rate is not greater than the preset completion rate threshold, the adaptation weight is always less than 1.

[0105] In this embodiment, the higher the work order processing priority, the higher the corresponding priority score.

[0106] In this embodiment, the higher the priority of the work order processing, the greater the corresponding priority dimension weight.

[0107] In this embodiment, the sum of the standard load weight and the standard matching weight is 1.

[0108] In this embodiment, the scenario coefficient is larger when the hotel's real-time operating status is high occupancy rate, and smaller when the hotel's real-time operating status is low occupancy rate.

[0109] In this embodiment, the baseline weights for the job types of candidate responsible terminals are determined as follows: Engineering and Maintenance > Guest Rooms > Delivery > Front Desk.

[0110] The beneficial effects of the above design scheme are as follows: By setting a dedicated priority dimension weight based on the work order processing priority, the core role of work order urgency in dispatching is highlighted, ensuring that high-priority work orders receive a higher weight in the score calculation, thus promoting rapid dispatching and response. The adaptation weight is determined by combining the historical on-time completion rate of candidate terminals handling work orders of the same priority level, incorporating the terminal's historical service capabilities into the weight allocation consideration, achieving an initial match between work order priority requirements and terminal processing capabilities. The preset completion rate threshold judgment rule provides a clear quantitative criterion for the adaptation weight. The system is accurate and highly executable. It provides a positive incentive of a matching weight of 1 to terminals with high completion rates, while reducing the weight of terminals with low completion rates by adjusting the weight based on the difference. This avoids assigning orders to terminals without the corresponding priority processing capabilities, reducing the probability of delays and errors in order processing from the source. By determining the load baseline weight according to the type of work, it aligns with the operational characteristics of different positions in the hotel. Considering the differences in load sensitivity and order processing characteristics among positions such as engineering maintenance, room service, and front desk consultation, the load weight allocation is adapted to the actual operational needs of each position. The dynamic load system is determined according to a rule negatively correlated with the load rate. The system accurately implements the core rule of the original patent: the lower the workload, the higher the score. This guides work orders to be allocated to low-load terminals, achieving load balancing across all hotel service terminals and avoiding processing delays caused by work order accumulation on a single terminal, thus optimizing internal service resource allocation. By combining the hotel's real-time operational status to determine scenario coefficients and adjust initial load weights, the system ensures scenario adaptability. The load weight ratio can be dynamically adjusted according to different operational states such as high occupancy rates, low load, and peak periods. For example, increasing the load weight to strengthen load balancing during high occupancy rates and decreasing the load weight to highlight priority during low load periods makes the load weight calculation more aligned with the hotel's real-time operational needs, improving the flexibility of the rules. Through normalization, the system unifies the dimensions of the two core dimensions' weights, allowing priority and processing load to play a reasonable role in the score calculation. By combining the current processing load score and historical operational data to calculate the dynamic score of candidate terminals, the system achieves comprehensiveness and accuracy in dynamic score calculation, making the dynamic score calculation more aligned with the actual operational needs of the hotel's in-stay services. Ultimately, this promotes the optimal allocation of hotel service resources and improves the overall efficiency and service quality of work order processing.

[0111] In one embodiment, a current processing load score is determined based on the current processing load data of each candidate responsible terminal, and a dynamic score is calculated based on the historical operating data of the candidate responsible terminals, combined with priority scores and matching weights, including:

[0112] The maximum number of work orders that a candidate responsible terminal can handle and the number of work orders that are currently pending are obtained from the current processing load data. The average processing time of the candidate responsible terminal for all work orders and the average processing time for historical work orders that are the same as the current work order are also obtained.

[0113] Based on the maximum number of work orders that can be carried, the number of work orders currently pending, the average processing time for all work orders, and the average processing time for historical work orders that are the same as the current work order, the current processing load score is calculated according to the following formula.

[0114]

[0115] in, This indicates the current processing load score of the candidate responsible terminal. This indicates the number of pending work orders currently being processed by the candidate responsible terminal. This indicates the maximum number of work orders that a candidate responsible terminal can carry. This indicates the average processing time of all work orders by the candidate responsible terminal. The average processing time of candidate responsibility terminals for historical work orders that are the same as the current work order;

[0116] The historical work order response rate, on-time completion rate and service satisfaction score of the candidate responsibility terminals are obtained from the historical operation data of the candidate responsibility terminals. Based on the historical operation data of all candidate responsibility terminals, the average response rate, average on-time completion rate and average satisfaction score of the hotels are obtained. Combined with the priority score and matching weight, the dynamic score of the candidate responsibility terminal is calculated according to the following formula.

[0117]

[0118] in, This represents the dynamic score of the candidate responsible terminal. This indicates the matching weight of the candidate responsible terminal. This indicates the priority score of the candidate responsible terminal. This indicates the load weight of the candidate responsible terminal. This indicates the historical work order response rate of the candidate responsible terminal. This indicates the hotel's average response rate. This indicates the on-time completion rate of work orders for candidate responsible terminals. This indicates the hotel's average on-time completion rate. This indicates the service satisfaction score of the candidate responsible terminal. This indicates the hotel's average satisfaction rating.

[0119] In this embodiment, the historical work order response rate refers to the response timeliness index of the candidate responsible terminal to dispatched work orders in the past, which refers to the proportion of the number of work orders received and responded to by the terminal within a preset time to the total number of dispatched work orders.

[0120] In this embodiment, the on-time completion rate of work orders refers to the proportion of work orders that the candidate responsible terminal has completed within the preset completion time limit to the total number of work orders processed, reflecting the timeliness of the terminal's work order processing.

[0121] The beneficial effects of the above design scheme are as follows: It extracts the maximum number of work orders currently being processed, the average processing time for all work orders, and the average processing time for historical work orders of the same type from the current processing load data. This provides an accurate, objective, and practical data foundation for the subsequent calculation of the current processing load score, ensuring the authenticity of the load score calculation. Through an overquantification formula, it achieves standardized score conversion of real-time load status, realizing automated and real-time calculation of the load score. This aligns with the automated needs of intelligent work order distribution in hotels, improving the efficiency of score calculation. By selecting three major indicators—historical work order response rate, work order on-time completion rate, and service satisfaction score—it calculates the score from the perspective of response time... The system comprehensively evaluates the historical operational capabilities of terminals across three dimensions: efficiency, completion timeliness, and service quality. This covers the core aspects of work order processing, avoiding the one-sidedness of single-indicator evaluations and truly reflecting the long-term service reliability of terminals. Through the weighted integration of three core dimensions—priority adaptability, real-time load status, and historical operational capabilities—it achieves the organic integration of multi-dimensional indicators. All parameters are quantified values, and the weights are preset fixed values, eliminating subjective human intervention. This makes the calculation of dynamic scores more scientific and provides a unique quantitative basis for ranking candidate terminals, avoiding the experience-based bias of manual work order dispatch. This effectively improves work order distribution efficiency and response timeliness, and optimizes the allocation of internal hotel service resources.

[0122] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A voice AI-driven automatic distribution and closed-loop management system for hotel in-stay service orders, characterized in that: It includes a voice interaction module, an AI analysis module, a work order management module, a terminal execution module, and a data management module; The voice interaction module is used to receive voice input information from hotel guests and employees and complete voice output feedback; the AI ​​parsing module is used to perform semantic understanding, intent recognition, and conversion of unformatted information to formatted information on the voice input information; the work order management module is used to automatically generate standardized work orders based on the AI ​​parsing results and complete work order distribution, progress tracking, and status updates according to preset rules; the terminal execution module is used to receive work order information and support voice work order node reporting from the execution terminal. The data management module is used to store, integrate, and analyze data throughout the entire process.

2. The voice AI-driven automatic distribution and closed-loop management system for hotel stay services as described in claim 1, characterized in that, The voice interaction module includes a customer voice interaction unit and an employee voice interaction unit; The guest voice interaction unit is configured to adapt to the voice interaction scenarios of hotel smart phones, smart voice terminals and mini-program voice terminals, and is used to receive voice requests for inquiries, guest needs, complaints and maintenance. The employee voice interaction unit is configured to be compatible with mobile work apps and smart voice terminals. It is used to receive work order execution status reported by voice and query work order information by voice. The voice interaction module continuously recognizes the intent based on the dialogue context during multiple rounds of intelligent human-computer dialogue.

3. The voice AI-driven automatic distribution and closed-loop management system for hotel stay services as described in claim 1, characterized in that, The AI ​​parsing module has a built-in NLP algorithm model, including a speech-to-text submodule, a semantic understanding submodule, and an information formatting submodule; The speech-to-text submodule is configured to filter out invalid speech noise and transcribe speech information into speech-text. The semantic understanding submodule is configured to perform intent classification, keyword extraction, and entity recognition on the transcribed speech text to determine the service demand type, service object, service location, and service requirements. The information formatting submodule is configured to convert the extracted service demand type, service object, service location, and service requirements into structured data and generate basic work order information.

4. The voice AI-driven automatic distribution and closed-loop management system for hotel stay services as described in claim 3, characterized in that, The semantic understanding submodule performs intent classification, keyword extraction, and entity recognition on the transcribed speech text, specifically including: The transcribed speech-text data is parsed and decomposed into a multi-layer semantic topic tree, with each layer containing several semantic clusters; Semantic clustering analysis is performed on the sub-speech text data corresponding to each layer to obtain several semantic cluster sets; each semantic cluster set is matched and assigned to the corresponding semantic cluster. The sentences contained in the semantic cluster set in each semantic cluster are segmented to obtain several feature extraction words in the sentences. The lexical semantic features of each feature extraction word and the syntactic position features of each feature extraction word in the sentence are determined. Based on a self-developed thesaurus of synonyms for the hotel industry and the lexical semantic features of each feature extraction word, the first semantic similarity between feature extraction words is determined. By combining the first semantic similarity and the syntactic position features of each feature extraction word in the sentence, a second fusion similarity between the feature extraction words is determined; The feature extraction words with a second similarity higher than a preset similarity threshold are deduplicated and integrated to obtain a core feature word set. The feature extraction words with consecutive positions in the core feature word set are concatenated with the maximum length using the smallest semantic unit of speech and text to obtain fused feature words. The fused feature words are cleaned and filtered based on the part-of-speech statistical characteristics of hotel service scenarios, and invalid function words are removed to obtain a set of service keywords; Map the keywords in the service keyword set to hotel industry-specific word vectors; Calculate the semantic distance between each dedicated word vector in any semantic cluster and the hotel service standard word vector corresponding to that cluster, and select the keyword corresponding to the dedicated word vector with the smallest semantic distance as the core target keyword in any semantic cluster; Based on the core target keywords in any semantic cluster, determine the service core keywords contained in each layer of the semantic topic tree, and complete the keyword extraction of speech text.

5. The voice AI-driven automatic distribution and closed-loop management system for hotel stay service orders as described in claim 1, characterized in that, The work order management module includes a work order generation submodule, an intelligent distribution submodule, a progress tracking submodule, and a closed-loop confirmation submodule. The work order generation submodule is configured to automatically generate standardized work orders containing work order number, requirement type, responsible department, completion deadline, and service recipient information based on the basic work order information output by the AI ​​parsing module. The intelligent distribution submodule is configured to push work orders to the corresponding responsible terminals based on the job responsibilities, employee workload, and service area of ​​each department in the hotel, according to preset distribution rules. The progress tracking submodule is configured to obtain the work order status reporting information from the terminal execution module in real time and update the status label of the work order, which includes pending, processing, and completed status labels; the closed-loop confirmation submodule is configured to automatically send result feedback to the service initiator and obtain confirmation information after the work order is completed, and complete the work order closure. The intelligent distribution submodule is also used to automatically divert work orders to backup execution terminals and send work order diversion reminder information to hotel management personnel when the workload of the responsible department or employee's work order processing terminal exceeds a preset threshold.

6. The voice AI-driven automatic distribution and closed-loop management system for hotel stay services as described in claim 5, characterized in that, The intelligent distribution submodule pushes work orders to the corresponding responsible terminals, specifically including: Extract the core matching elements contained in the work order. The core matching elements include at least the service request type, the physical area where the service occurs, the work order processing priority, and the required service skills. The system retrieves a pre-set responsibility terminal matching library, which stores the responsible area, the types of requests that can be processed, the service skill tags of the terminal operators, the number of work orders currently pending processing, and the processing load of each responsibility terminal. The core matching elements of the work order are initially screened and matched with the information in the responsibility terminal matching library to select candidate responsibility terminals that meet the matching of the responsible area, can handle the corresponding demand type, and have the required service skills. The candidate responsible terminals are assigned matching weights based on the work order processing priority. At the same time, the current processing load of each candidate responsible terminal is combined to calculate the dynamic score. The lower the load, the higher the score. Candidate responsibility terminals are sorted from highest to lowest according to their dynamic scores, and the terminal with the highest score is selected as the target responsibility terminal. The system sends a work order push instruction to the target responsible terminal and marks the work order status as distributed. If the target responsible terminal does not receive the work order within a preset time, the system automatically pushes the work order to the next candidate responsible terminal in the ranking and sends a work order receipt reminder message to the administrator.

7. The voice AI-driven automatic distribution and closed-loop management system for hotel stay service orders as described in claim 6, characterized in that, Based on the work order processing priority, matching weights are assigned to candidate responsible terminals. Simultaneously, dynamic scores are calculated considering the current processing load of each candidate responsible terminal, including: The work order processing priority is converted using a standard score to obtain a priority score. Priority dimension weights are set based on work order processing priority. The on-time completion rate of historical work orders with the same processing priority level as the candidate responsibility terminal is obtained. When the on-time completion rate is greater than the preset completion rate threshold, the adaptation weight of the candidate responsibility terminal is determined to be 1. Otherwise, the adaptation weight of the candidate responsibility terminal is determined based on the difference between the preset completion rate threshold and the on-time completion rate. The product of the priority dimension weight and the adaptation weight is used as the matching weight for the candidate responsible terminal. Based on the job type of the candidate responsibility terminal, a baseline weight is determined. Based on the number of pending work orders of the candidate responsibility terminal, the terminal load rate is determined. According to the rule that is negatively correlated with the load rate, the load dynamic coefficient is determined. Based on the product of the load dynamic coefficient and the baseline weight, the initial load weight of the candidate responsibility terminal is determined. Based on the real-time operation status of the hotel, the scenario coefficient is determined. The product of the scenario coefficient and the initial load weight is used as the load weight of the candidate responsibility terminal. The load weight and matching weight are normalized to obtain the standard load weight and standard matching weight. Based on the current processing load data of each candidate responsible terminal, the current processing load score is determined, and based on the historical operation data of the candidate responsible terminals, combined with the priority score and matching weight, the dynamic score of the candidate responsible terminals is calculated.

8. The voice AI-driven automatic distribution and closed-loop management system for hotel stay service orders as described in claim 1, characterized in that, The terminal execution module includes a mobile work terminal for hotel staff and an intelligent voice terminal. The mobile work terminal supports reporting work order status in both voice and text formats. The intelligent voice terminal is a dedicated terminal for the hotel industry. It is configured to support voice reporting of work order node information, voice query of work order details, voice retrieval of service recipients and room status information, and can also activate the terminal camera through voice commands to record and take pictures of the service process. The recorded and photographed multimedia information can be synchronously associated with the corresponding work order and stored in the data management module.

9. The voice AI-driven automatic distribution and closed-loop management system for hotel stay service orders as described in claim 1, characterized in that, It also includes third-party system integration modules; The third-party system integration module is configured to interconnect with third-party systems, including a hotel management system (PMS), an intelligent guest control system, a delivery robot system, and an intelligent speaker system. It pushes guest-related work order information to the intelligent guest control system and the delivery robot system for automated service execution, and receives service execution feedback information from the third-party system and synchronizes it to the progress information of the corresponding work order.

10. The voice AI-driven automatic distribution and closed-loop management system for hotel stay services as described in claim 7, characterized in that, Based on the current processing load data of each candidate responsible terminal, a current processing load score is determined. Then, based on the historical operational data of the candidate responsible terminals, combined with priority scores and matching weights, a dynamic score for each candidate responsible terminal is calculated, including: The maximum number of work orders that a candidate responsible terminal can handle and the number of work orders that are currently pending are obtained from the current processing load data. The average processing time of the candidate responsible terminal for all work orders and the average processing time for historical work orders that are the same as the current work order are also obtained. Based on the maximum number of work orders that can be carried, the number of work orders currently pending, the average processing time for all work orders, and the average processing time for historical work orders that are the same as the current work order, the current processing load score is calculated according to the following formula. The historical work order response rate, on-time completion rate and service satisfaction score of the candidate responsibility terminals are obtained from the historical operation data of the candidate responsibility terminals. Based on the historical operation data of all candidate responsibility terminals, the average response rate, average on-time completion rate and average satisfaction score of the hotels are obtained. Combined with the priority score and matching weight, the dynamic score of the candidate responsibility terminal is calculated according to the following formula.

Images