Competition system for hotel service and working method thereof

Abstract

The invention belongs to the technical field of hotel service training, and discloses a competition system for hotel service and a working method thereof, and the competition system comprises a user terminal which is used for receiving and displaying a task instruction and collecting service operation data of competitors; the task generation module is in communication connection with the user terminal and comprises a standardized process database; the dynamic task generation unit is configured to generate a simulation task containing time constraint and resource limitation; the scoring module comprises a motion capture unit, a voice analysis unit and a virtual customer feedback unit; the data processing module is used for calculating a comprehensive score and generating a real-time ranking; the visual terminal is used for dynamically displaying a three-dimensional virtual scene and match data; the deep learning optimization module is used for analyzing historical data and generating personalized training suggestions; the problem that existing hotel service training lacks real-time performance, quantification performance and interactivity is solved.

Description

Technical Field

[0001] This solution belongs to the technical field of hotel service training, and specifically relates to a competition system for hotel service and its working method. Background Art

[0002] With the continuous development of the hotel industry, the requirements for the professional qualities of service staff are getting higher and higher, and traditional training methods can no longer meet the needs of modern hotel service training. Hotel service competitions are tools to improve hotel service quality and train service staff. By simulating real hotel scenarios, it allows contestants to complete various service tasks in a virtual environment, thereby improving service skills and emergency handling capabilities.

[0003] Referring to the literature with the existing publication (announcement) number CN119417540A, a method for evaluating and assessing intelligent hotel room service based on AI is disclosed, including the following steps: S1: Data collection: Collect feedback data from customers on room service by designing questionnaires, deploying feedback systems, and integrating application programming interfaces API. This method for evaluating and assessing intelligent hotel room service based on AI enables hotels to deeply understand the personalized needs and preferences of each customer through a detailed scoring system and multi-dimensional data collection.

[0004] The above evaluation method lacks real-time performance and cannot monitor the operation details of service staff in real time, resulting in delayed feedback. Secondly, it lacks quantification and cannot quantitatively evaluate the posture standardization, communication efficiency, etc. of service staff. In addition, the existing technology lacks interactivity and cannot enhance the participation and learning effect of service staff through virtual scenario simulation and real-time scoring. Finally, it lacks personalization and cannot generate personalized training suggestions based on the weak links of service staff. Summary of the Invention

[0005] The purpose of this solution is to provide a competition system for hotel service to solve the problems of lack of real-time performance, quantification, and interactivity in existing hotel service training.

[0006] To achieve the above purpose, this solution provides a competition system for hotel service, including:

[0007] A user terminal, configured to receive and display task instructions and collect service operation data of contestants;

[0008] A task generation module, communicatively connected to the user terminal, including: a standardized process database storing service operation specification data based on the ISO9001 standard; a dynamic task generation unit configured to generate simulation tasks including time constraints and resource limitations;

[0009] A scoring module, including an action capture unit, a voice analysis unit, and a virtual customer feedback unit;

[0010] A data processing module, used to calculate a comprehensive score and generate a real-time ranking;

[0011] A visualization terminal, used to dynamically display a three-dimensional virtual scene and competition data;

[0012] A deep learning optimization module, used to analyze historical data and generate personalized training suggestions;

[0013] The user terminal, task generation module, scoring module, data processing module, visualization terminal and deep learning optimization module achieve data interaction through a cloud server.

[0014] The principle and effect of this solution are as follows: The system receives and displays task instructions in real time through the user terminal, and collects service operation data of the contestants, and conducts real-time monitoring and data collection during the task execution process. The task generation module combines the ISO9001 standard and a dynamic task generation algorithm to create a simulated task with time constraints and resource limitations, making the training closer to the actual work scenario. The scoring module integrates motion capture, speech analysis and virtual customer feedback technologies to comprehensively evaluate the service operations of the contestants, covering multiple dimensions such as the standardization of postures, communication efficiency and customer satisfaction. The data processing module can calculate a comprehensive score and generate a real-time ranking, providing immediate feedback to the contestants to help them promptly understand their performance and improvement directions. The visualization terminal uses a three-dimensional rendering engine to dynamically display the virtual hotel scene and competition data, enabling the contestants to be more actively involved in the training process. The deep learning optimization module analyzes historical data to generate personalized training suggestions for service personnel, providing specific improvement measures for their weak links to achieve personalized training. This solution improves the professional qualities and emergency handling capabilities of hotel service personnel through real-time monitoring, quantitative evaluation and personalized training, and improves the efficiency and effect of training.

[0015] Furthermore, the dynamic task generation unit includes:

[0016] A time pressure calculation sub-unit, which generates a task countdown threshold:

[0017] T limit = μ T ·(1 + k·rand())

[0018] where μ T is the industry standard task duration, and k is a floating coefficient and satisfies 0.1 ≤ k ≤ 0.3;

[0019] A resource allocation sub-unit, which sets the upper limit of each group of resources:

[0020]

[0021] where ri is the reference value of the i-th type of resource, and N p is the number of people in the group, and η is the resource compression factor and satisfies 0.7 ≤ η ≤ 0.9.

[0022] The principle and effect of this solution are as follows: The time pressure calculation subunit generates a simulated task with time constraints through the industry-standard task duration and the floating coefficient, enabling the contestants to complete the task within the specified time and improving their time management ability. The resource allocation subunit then sets the upper limit of resources for each group according to the resource reference value, the number of people in the group, and the resource compression factor, reasonably allocates the limited resources, avoids resource waste, and at the same time increases the complexity and authenticity of the task.

[0023] Furthermore, the scoring module includes:

[0024] The motion capture unit collects data through nine-axis inertial sensors deployed on the limbs and torso of the contestants and calculates the posture norm score:

[0025]

[0026] where θ j is the deviation value of the j-th joint angle;

[0027] The speech analysis unit uses speech recognition technology based on natural language processing to calculate the communication score:

[0028]

[0029] where N key is the number of keyword matches.

[0030] The principle and effect of this solution are as follows: The motion capture unit uses inertial sensors to collect the motion data of the limbs and torso of the contestants, fuses the sensor signals through algorithms such as Kalman filtering, calculates the deviation value of the joint angle, and thus quantitatively evaluates the posture norm of the service personnel. The speech analysis unit uses natural language processing technology to extract the features of the speech signal (such as the number of keyword matches, speech quality, etc.), and evaluates the speech quality through a deep learning model to calculate the communication efficiency. This evaluation mechanism not only improves the interactivity and participation of the training, but also provides instant feedback to the service personnel, helping them quickly identify and improve the weak links, thus significantly improving the efficiency and effect of hotel service training.

[0031] Furthermore, the virtual customer feedback unit generates the customer satisfaction score:

[0032]

[0033] where R k is the completion degree of the k-th service indicator, and W kis a preset weight and satisfies ∑W k = 1, T resp is the response time, E score is the empathy index and satisfies 0 ≤ E score ≤ 5.

[0034] The principle and effect of this solution are as follows: The virtual customer feedback unit constructs a comprehensive customer satisfaction scoring system based on key parameters such as the completion degree of multiple service indicators, response time, and empathy index. It can not only reflect the performance of service personnel in various service indicators in real time but also measure the understanding and attention of service personnel to customer needs through the empathy index, thereby providing accurate feedback to service personnel to help them identify and improve weak links in the service.

[0035] Furthermore, the data processing module includes:

[0036] Comprehensive score calculation engine, which executes:

[0037]

[0038] where w i is the task difficulty weight and satisfies 1.0 ≤ w i ≤ 2.5;

[0039] Real-time ranking algorithm, which generates:

[0040]

[0041] where is the score growth rate, I nov is the innovation score.

[0042] The principle and effect of this solution are as follows: The comprehensive score calculation engine weights and sums up the performance of contestants in each task according to the task difficulty weight, making the scoring scientific and fair. The weight range is set between 1.0 and 2.5, considering the importance and difficulty differences of different tasks. The real-time ranking algorithm dynamically generates ranking results by introducing key indicators such as the score growth rate and innovation score, reflecting the current performance of the contestants.

[0043] Furthermore, the visualization terminal displays the virtual hotel scene through a 3D rendering engine, with a positioning error ≤ 0.5 meters, and generates a heat map and a six-dimensional radar chart data dashboard.

[0044] The principle and effect of this solution are as follows: The 3D rendering engine uses physically based rendering technology to convert 2D models into realistic 3D images by simulating the propagation and reflection of light in 3D space, capturing every detail, texture and lighting effect. Through multi-sampling anti-aliasing technology, the rendering engine can eliminate the jagged edges of geometric objects, making the edges of the scene clear and smooth, improving the realism of the scene and enhancing the immersion and participation of the contestants.

[0045] Furthermore, the deep learning optimization module includes:

[0046] The input layer receives 52 feature parameters, including task duration, error code, and customer feedback text;

[0047] ResNet-50 network architecture, outputting skill improvement suggestion vectors;

[0048] Training strategy, using learning rate 10 -4 , batch size 32, cross entropy loss function.

[0049] The principle and effect of this solution are as follows: the input layer of the module receives 52 feature parameters, including task duration, error code and customer feedback text, to provide input data for the model. The ResNet-50 network architecture can extract features and avoid the gradient vanishing problem through its residual learning mechanism and jump connection, thereby learning more effective feature representations. The training strategy uses learning rate, batch size 32 and cross entropy loss function to ensure stable and efficient model training. This deep learning optimization module can not only analyze historical game data in real time, but also generate targeted skill improvement suggestion vectors to help service personnel quickly identify and improve weak links, thereby significantly improving the efficiency and effectiveness of hotel service training.

[0050] Furthermore, it also includes illegal operation penalty items, which include graded penalty points for violation of safety regulations and privacy leakage penalty. The privacy leakage penalty formula is:

[0051] P privacy =5·(1+log2(1+N info ))

[0052] Among them, N info The number of leaked customer information items.

[0053] The principle and effect of this scheme are as follows: the graded deduction mechanism for violations of safety regulations deducts points for not wearing a work badge, operating equipment incorrectly, etc., which strengthens the safety awareness and operation standardization of service personnel and ensures the professionalism of the service process. The privacy leakage penalty formula quantifies the degree of customer information leakage and imposes severe penalties on leakage behaviors, thus protecting the privacy security of customers and enhancing the confidentiality awareness of service personnel.

[0054] A competition method for hotel services, including applying a competition system for hotel services according to any one of claims 1-8, characterized by comprising the following steps:

[0055] Step S10: Divide the contestants into N groups, and assign the roles of service staff, front desk receptionist, and management personnel to each group;

[0056] Step S20: Generate tasks; in the qualification stage, release 20 basic service tasks; in the promotion stage, randomly insert sudden tasks 3 times per hour;

[0057] Step S30: Collect action data through sensors and calculate scores;

[0058] Step S40: Update the leaderboard every 24 hours and output suggestions for skill optimization.

[0059] Further, in the step S20, the generation of sudden tasks includes:

[0060] Generate an event sequence based on the Markov chain model, and the dimension of the event transition probability matrix is 15×15;

[0061] Dynamically adjust resource constraint conditions:

[0062]

[0063] Among them, S max Is the highest score in the current stage. Brief Description of the Drawings

[0064] Figure 1 It is a schematic flow chart of the competition method for hotel services of the present invention. Detailed Embodiments

[0065] The following will clearly and completely describe the concept and technical effects generated by the present invention in combination with embodiments to fully understand the purpose, features, and effects of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all embodiments. Other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative efforts shall fall within the scope of protection of the present invention:

[0066] Embodiment:

[0067] A competition system for hotel services adopts the following configuration parameters:

[0068] User terminal: Adopt a Huawei MatePad Pro tablet computer, equipped with a 10.8-inch touch display screen (resolution 2560×1600) and a voice input device, running the Android 11 operating system, and used for task reception and data upload.

[0069] Task generation module: Deployed on an Alibaba Cloud ECS server (model: ecs.g6e.xlarge), it includes: 1. Standardized process database: Stores service operation specification data based on the ISO 9001:2015 standard. The database uses MySQL 8.0 and stores 1000 hotel service cases, covering scenarios such as room service, front desk reception, and emergency handling. 2. Dynamic task generation unit: Using the Python 3.8 programming language, it generates simulation tasks based on the Markov decision process. The task types include basic services, crisis handling, and innovative services.

[0070] The scoring module includes: 1. Action capture unit: Using a nine-axis inertial sensor (model: Bosch BNO055), deployed on the limbs and torso of the contestants, with a sampling frequency of 100Hz, used to collect service posture data. 2. Voice analysis unit: Using the Wave2Vec 2.0 model, running on an NVIDIA Tesla T4 GPU server, used to identify the compliance of service language. 3. Virtual customer feedback unit: Integrating a generative adversarial network (GAN), using the PyTorch 1.9 framework, used to simulate customer satisfaction.

[0071] Data processing module: Deployed on an Alibaba Cloud ECS server, using the TensorFlow 2.4 framework, calculates the comprehensive score and generates a real-time ranking.

[0072] Visualization terminal: Constructs a virtual hotel environment using the UE5 engine, runs on a PC, and is configured as follows: GPU uses NVIDIA RTX 4080 (12GB video memory); CPU uses Intel i9-11900K (3.5GHz); the monitor uses a Dell UltraSharp 32-inch 4K monitor (3840×2160).

[0073] Deep learning optimization module: Based on the ResNet-50 model, using the Keras framework, the training strategy includes a learning rate of 10 -4 、batch size of 32, and cross-entropy loss function.

[0074] Competition process:

[0075] Step S10, group the contestants and assign roles:

[0076] Divide 24 contestants into 6 groups, with 4 people in each group, and assign the roles of service staff, front desk receptionist, and manager.

[0077] Group A: Service staff (Zhang San, Li Si), Front desk receptionist (Wang Wu), Manager (Zhao Liu).

[0078] Step S20, Dynamic Task Generation and Allocation:

[0079] Qualifying Stage (72 hours): Release 20 basic service tasks, including room tidying (benchmark time consumption: 20 minutes) and routine complaint handling (benchmark time consumption: 15 minutes).

[0080] Advancing Stage: Randomly insert emergency tasks 3 times per hour, such as fire emergency response (resource constraint factor: 0.7) and sudden illness handling (time pressure factor: 1.3).

[0081] Final Stage: Conduct personalized customer care plan design (innovation scoring weight: 30%).

[0082] Step S30, Multi-dimensional Real-time Scoring:

[0083] Calculation of Task Countdown Threshold:

[0084] T limit = μ T ·(1 + k·rand())

[0085] Assume the industry standard task time consumption μ T = 20 minutes, floating coefficient k = 0.2, random number rand() = 0.5, then:

[0086] T limit = 20·(1 + 0.2·0.5) = 22 min

[0087] Calculation of Resource Upper Limit:

[0088]

[0089] Assume r1 = 5 (room cleaning tools), r2 = 3 (emergency equipment), r3 = 2 (customer information), number of people in the group N p = 4, resource compression factor η = 0.8, then:

[0090] R max = (5 + 3 + 2)·4·0.8 = 32

[0091] Calculation of Posture Norm Score:

[0092]

[0093] Assume that within n = 10 time points, the average value of joint angle deviation is 30°, then:

[0094]

[0095] Calculation of Communication Efficiency Score:

[0096]

[0097] Assume the number of keyword hits is N key = 8 (keywords include "Welcome", "Please wait a moment", "Thank you for your patience", etc.), the total number of words is N total = 10, and the cosine similarity of the voice feature vector is 0.9, then:

[0098]

[0099] Calculation of customer satisfaction:

[0100]

[0101] Assume R1 = 0.9 (problem-solving degree), R2 = 0.8 (service attitude), R3 = 0.7 (response speed), R4 = 0.6 (environmental cleanliness), R5 = 0.8 (extra service), and the weight W k = [0.2, 0.2, 0.2, 0.2, 0.2], the response time T resp = 30 seconds, and the empathy index E score = 3, then:

[0102]

[0103] Calculation of single-task score:

[0104]

[0105] Assume the industry benchmark time-consuming is T base = 20 minutes, and the actual time-consuming is T actual = 18 minutes, and the postural standardization score is Q std = 0.8333, and the customer satisfaction C sat = 0.034, then:

[0106]

[0107] Step S40, comprehensive ranking and feedback optimization:

[0108] Calculation of comprehensive score:

[0109]

[0110] Assume that 3 tasks are completed, and the scores are 0.7 (room tidying), 0.85 (complaint handling), and 0.88 (emergency response) respectively, and the weights are 1.0 (basic task), 1.5 (sudden task), and 2.0 (innovative task) respectively, and the cooperation bonus is B m = 0.1, and there is no deduction, then:

[0111] S total=0.7·1.0+0.85·1.5+0.88·2.0+0.1=3.835

[0112] Final ranking calculation:

[0113]

[0114] Assuming the comprehensive score S total =3.835, score growth rate Innovation Points I nov =8 (innovative services include personalized welcome gifts, prediction of special customer needs, etc.), then:

[0115] Rank=0.6·3.835+0.3·0.5+0.1·8≈3.251

[0116] Results:

[0117] According to the above calculations, the final ranking of Group A was 3.251, successfully completing the competition and achieving excellent results.

[0118] Deep learning optimization module application:

[0119] The deep learning optimization module is based on the ResNet-50 model, analyzes historical game data, and generates personalized training plans. After Group A completed the qualifying round, the module analyzed its motion capture data and voice interaction records and found that the service staff had low posture standardization scores in the "customer greeting" link. The training suggestions generated by the module include: Intensive training for the "customer greeting" action, with a recommended training time of 30 minutes per day. Optimize service terms and increase the frequency of use of keywords such as "welcome" and "please take care". Improve emergency response speed, and it is recommended to conduct special training to simulate fire scenes. By implementing these suggestions, Group A's scores on relevant indicators in the qualifying round increased by 15%.

[0120] Illegal operation handling:

[0121] During the competition, Group B was judged as violating the safety regulations level 1 because they did not wear their work badges as required. According to the formula:

[0122] P privacy =5·(1+log2(1+N info ))

[0123] Where N info =1 (disclosure of customer name), deduction points are:

[0124] P privacy =5·(1+log2(2))=5·1.5=7.5

[0125] This deduction directly affects the final ranking of Group B.

[0126] Verification of System Operation Stability:

[0127] To verify the stability of the system, we conducted a stress test on the system during the qualifying stage. During the test, the system simultaneously processed data of 24 groups of contestants, including motion capture, voice analysis, and real-time scoring. The results showed that the average system response time was 2.3 seconds, and the data upload success rate reached 99.7%, proving that the system has good concurrent processing capabilities and stability.

[0128] User Feedback and System Improvement:

[0129] After the competition, the contestants evaluated the system. The feedback showed that 92% of the contestants thought the task assignment and scoring mechanism of the system were fair and transparent, and 88% of the contestants said they optimized their service skills through the system's feedback. Based on the feedback, we plan to add more real-time feedback functions in the subsequent version, such as voice prompts for improvement directions, action demonstration videos, etc., to further enhance the training effect of the system.

[0130] The above are only embodiments of the present invention, and common knowledge such as specific structures and characteristics known in the solution is not described in detail here. It should be noted that for those skilled in the art, without departing from the structure of the present invention, several modifications and improvements can still be made, and these should also be regarded as the protection scope of the present invention, which will not affect the implementation effect of the present invention and the practicality of the patent. The protection scope required by this application should be based on the content of its claims, and the specific implementation manners described in the specification can be used to explain the content of the claims.

Claims

1. A competition system for hotel services, characterized in that, include: User terminal, used to receive and display task instructions and collect service operation data of contestants; The task generation module is connected to the user terminal in communication, and includes: a standardized process database storing service operation specification data based on ISO9001 standard; a dynamic task generation unit configured to generate simulation tasks including time constraints and resource constraints; Scoring module, including motion capture unit, voice analysis unit and virtual customer feedback unit; Data processing module, used to calculate comprehensive scores and generate real-time rankings; Visualization terminal, used to dynamically display three-dimensional virtual scenes and competition data; A deep learning optimization module that analyzes historical data and generates personalized training recommendations; The user terminal, task generation module, scoring module, data processing module, visualization terminal and deep learning optimization module realize data interaction through the cloud server.

2. The competition system for hotel services according to claim 1, characterized in that: The dynamic task generation unit comprises: The time pressure calculation subunit generates the task countdown threshold: T limit = μ T ·(1 + k·rand()) Among them, μ T is the task time-consuming of the industry standard, and k is the floating coefficient and satisfies 0.1 ≤ k ≤ 0.3; Resource allocation subunit, set the upper limit of each resource group: Among them, r i is the reference value of the i-th type of resource, N p is the number of people in the group, and η is the resource compression factor and satisfies 0.7 ≤ η ≤ 0.

9.

3. The competition system for hotel services according to claim 1, wherein: The scoring module includes: The motion capture unit collects data through nine-axis inertial sensors deployed on the limbs and torso of the contestants and calculates the posture standardization score: where θ j is the angular deviation value of the j-th joint; The speech analysis unit uses speech recognition technology based on natural language processing to calculate the communication score: Among them, N key is the number of keyword matches.

4. The competition system for hotel services according to claim 1, characterized in that: The virtual customer feedback unit generates a customer satisfaction score: Among them, R k is the completion degree of the k-th service indicator, W k is the preset weight and satisfies ∑W k = 1, T resp is the response time, E score is the empathy index and satisfies 0 ≤ E score ≤ 5.

5. The competition system for hotel services according to claim 1, characterized in that: The data processing module comprises: The comprehensive score calculation engine performs: Among them, w i is the task difficulty weight and satisfies 1.0 ≤ w i ≤ 2.5; Real-time ranking algorithm, generating: Among them, is the score growth rate, and I nov is the innovation score.

6. The competition system for hotel services according to claim 1, characterized in that: The visualization terminal displays the virtual hotel scene through a three-dimensional rendering engine, with a positioning error of ≤0.5 meters, and generates a heat map and a six-dimensional radar map data dashboard.

7. The competition system for hotel services according to claim 1, characterized in that: The deep learning optimization module includes: The input layer receives 52 feature parameters, including task duration, error code, and customer feedback text; ResNet-50 network architecture, outputting skill improvement suggestion vectors; Training strategy, with a learning rate of 10 -4 , a batch size of 32, and a cross-entropy loss function.

8. The competition system for hotel services according to claim 1, characterized in that: It also includes illegal operation penalty items, which include graded penalty points for violation of safety regulations and privacy leakage penalty. The privacy leakage penalty formula is: P privacy = 5·(1 + log2(1 + N info )) where N info is the number of customer information items leaked.

9. A competition method for hotel services, including applying the competition system for hotel services according to any one of claims 1-8, characterized in that, The steps include: Step S10: Divide the contestants into N groups, and assign service personnel, front desk receptionists and management personnel roles to each group; Step S20: Generate tasks; During the qualifying phase, 20 basic service tasks were released; During the promotion round, emergency tasks will be randomly inserted 3 times per hour; Step S30: collecting motion data through sensors and calculating scores; Step S40: Update the ranking list every 24 hours and output skill optimization suggestions.

10. The competition method for hotel services according to claim 9, characterized in that: In step S20, the burst task generation includes: The event sequence is generated based on the Markov chain model, and the event transition probability matrix dimension is 15×15; Dynamically adjust resource constraints: Among them, S max is the highest score in the current stage.

Images