A hotel operation management system and method based on big data

Abstract

The present application discloses a hotel operation management system and method based on big data, which belongs to the technical field of hotel operation management. It constructs a customer portrait recognition model by adopting an artificial intelligence model, and trains the customer portrait recognition model by adopting the customer portrait label corresponding to the customer feature big data after preprocessing, so as to obtain the trained customer portrait recognition model, and then identifies the target customer portrait of the customer to be checked in through the customer portrait recognition model, so as to realize personalized service recommendation according to the target customer portrait, which can effectively improve the competitiveness of the hotel, and at the same time, can predict the consumption of hotel supplies according to the personalized recommendation service, so as to realize the dynamic configuration of hotel resources and improve the customer experience.

Description

Technical Field

[0001] The present application belongs to the technical field of hotel operation management, and specifically relates to a hotel operation management system and method based on big data. Background Art

[0002] With the development of economy, the prosperity of tourism industry and the increase of floating population, various business places providing temporary accommodation such as hotels, serviced apartments, family hotels and short-term rental houses are also increasing, and the competition among them is also intensifying. How to effectively expand its business space, reduce operating costs and gain advantages under new competitive conditions is an important issue faced by every hotel or inn. In addition, with the advancement of science and technology and the development of society, people's social relations, ideas and lifestyles have also changed greatly compared with the past. Network and communication terminals have become an indispensable part of people's lives with their convenience and speed. How hotels and serviced apartments can adapt to this change to meet the needs of modern customers, provide customers with more humane services and ultimately bring more customers has also become an urgent problem for enterprises to solve. In the prior art, hotels generally adopt manual management, and do not provide personalized services for customers, which cannot effectively meet customer needs and reduce the competitiveness of hotels. Summary of the invention

[0003] The present application provides a hotel operation management system and method based on big data, which is used to solve the problem that the existing technology relies on manual management, resulting in the inability to provide better services and reducing the competitiveness of the hotel.

[0004] On the one hand, the present application provides a hotel operation management system based on big data, including: a historical data acquisition module, an artificial intelligence model training module, a customer portrait recognition module, a personalized service acquisition module and a comprehensive resource management module;

[0005] The historical data acquisition module is used to acquire customer feature big data and customer portrait labels corresponding to the manually marked customer feature big data, and preprocess the customer feature big data and the customer portrait labels corresponding to the customer feature big data to obtain the customer portrait labels corresponding to the preprocessed customer feature big data;

[0006] The artificial intelligence model training module is used to construct a customer portrait recognition model using an artificial intelligence model, and train the customer portrait recognition model using the customer portrait labels corresponding to the pre-processed customer feature big data to obtain a trained customer portrait recognition model;

[0007] The customer portrait recognition module is used to collect target customer feature big data corresponding to the customer to be checked in in real time, and use the trained customer portrait recognition model to recognize the target customer feature big data to obtain the target customer portrait corresponding to the customer to be checked in;

[0008] The personalized service acquisition module is used to search and match the target customer portrait corresponding to the customer to be accommodated in the relationship library between the preset customer portraits and personalized recommendation services based on the target customer portrait corresponding to the customer to be accommodated, and determine at least one target personalized recommendation service; wherein each customer portrait in the relationship library has an association relationship with at least one personalized recommendation service;

[0009] The comprehensive resource management module is used to recommend at least one target personalized recommendation service to customers waiting to check in, and to merge at least one target personalized recommendation service with historical personalized recommendation service data, to predict hotel supplies consumption, and to transmit the hotel supplies consumption to a warehouse manager to dynamically configure hotel resources.

[0010] In a possible implementation, obtaining customer feature big data and customer portrait labels corresponding to the manually labeled customer feature big data includes:

[0011] Obtain the gender, age, region, historical ratings of accommodation, housing type, whether coupons were used during accommodation, and booking channels of historical guests, and obtain big data on customer characteristics;

[0012] The customer feature big data is displayed to management personnel to obtain customer portrait labels corresponding to the customer feature big data marked by the management personnel.

[0013] In a possible implementation, preprocessing the customer feature big data and the customer portrait label corresponding to the customer feature big data to obtain the customer portrait label corresponding to the preprocessed customer feature big data includes:

[0014] Numerical processing is performed on the customer feature big data and the customer portrait label corresponding to the customer feature big data to convert the text data into digital data, thereby obtaining the customer feature big data after numerical processing and the customer portrait label corresponding to the customer feature big data;

[0015] Perform outlier processing on the customer feature big data after numerical processing and the customer portrait label corresponding to the customer feature big data to obtain the customer feature big data after outlier processing and the customer portrait label corresponding to the customer feature big data;

[0016] The customer feature big data after outlier processing and the customer portrait labels corresponding to the customer feature big data are normalized to obtain the customer portrait labels corresponding to the customer feature big data after preprocessing.

[0017] In a possible implementation, an artificial intelligence model is used to construct a customer portrait recognition model, and the customer portrait recognition model is trained using customer portrait labels corresponding to the preprocessed customer feature big data to obtain a trained customer portrait recognition model, including:

[0018] An artificial intelligence model is used to build a customer portrait recognition model, and the hyperparameters of the customer portrait recognition model are used as parameters to be trained and the initial training population;

[0019] Based on the customer portrait label corresponding to the pre-processed customer feature big data, the customer portrait recognition model is used to obtain the fitness corresponding to each individual, and the individual with the largest fitness is selected as the optimal individual;

[0020] According to the optimal individual, a social information exchange strategy with adaptive cosine weights is used to perform adaptive learning on each individual in the training population to obtain an individual after adaptive learning;

[0021] For individuals after adaptive learning, the population boundary reference search strategy is used to perform an initial search in the local area to obtain individuals after the initial search in the local area;

[0022] For individuals after the initial search in the local area, a spiral progressive search strategy is used to perform local area expansion search on the individuals to obtain individuals after the local area expansion search;

[0023] For individuals after the local area expansion search, the global jump greedy search strategy is used to perform global area jump search on the individuals to obtain the individuals after the global area jump search;

[0024] Determine whether the number of training times has reached the maximum number of training times. If so, re-determine the optimal individual based on the individual after the global area jump search, and apply the hyperparameters contained in the optimal individual to the customer portrait recognition model to obtain the trained customer portrait recognition model. Otherwise, return to the step of determining the optimal individual.

[0025] In a possible implementation, according to the optimal individual, a social information exchange strategy with adaptive cosine weights is used to perform adaptive learning on each individual in the training population to obtain an individual after adaptive learning, including:

[0026] According to the current number of training times, the adaptive inertia weight is obtained as: ;in, represents the adaptive inertia weight, represents pi, t represents the current number of training times, and T represents the preset maximum number of training times;

[0027] Adaptive inertia weight is used to i By weighting each individual, the first weighted term is: ;in, Indicates t During the training process i Individuals, i =1,2,…,I, where I represents the total number of individuals in the training population;

[0028] Randomly generate the first learning factor, the second learning factor and the third learning factor between (0,1);

[0029] According to the first learning factor, let i Individual By learning the optimal individual, the first learning item is: ;in, represents the first constant factor, represents the first learning factor, represents the optimal individual;

[0030] According to the second learning factor, let i Individual The best individual in the last training process is learned, and the second learning item is obtained as follows: ;in, represents the second constant factor, represents the second learning factor, represents the best individual in the t-1th training process, that is, the best individual in the previous training process;

[0031] According to the first weighted item, the first learning item, the second learning item and the third learning factor, the individual after adaptive learning is obtained as follows: ;in, represents the third learning factor, Represents the individual after adaptive learning .

[0032] In a possible implementation, for individuals after adaptive learning, a population boundary reference search strategy is used to perform an initial local area search on the individuals to obtain individuals after the initial local area search, including:

[0033] According to the fitness of each individual, the contribution of each individual is determined as: ;in, Indicatest During the training m The fitness of an individual after adaptive learning, m =1,2,…,I, I represents the total number of individuals in the training population, Indicates m The corresponding contribution of each individual after adaptive learning;

[0034] According to the sharing degree of each individual, the center of gravity of the training population is obtained as: ;in, represents the centroid position of the training population, Indicates m An individual after adaptive learning;

[0035] According to the centroid position of the training population and the worst individual, the first reference search item is obtained as: ;in, represents the first learning rate, and is a random number uniformly distributed between [0,1]. Indicates t The worst individual in the training process;

[0036] According to the center of gravity position of the training population and the optimal individual, the second reference search item is obtained as: ;in, represents the second learning rate, and is a random number uniformly distributed between [0,1]. Indicates t The best individual in the training process;

[0037] According to the first reference search item and the second reference search item, an initial local area search is performed on the individual, and the individual after the initial local area search is obtained as follows: ;in, represents a random adjustment factor between (0,1), Indicates m Individuals after initial search in local areas .

[0038] In a possible implementation, for individuals after the initial search in the local area, a spiral progressive search strategy is used to perform a local area expansion search on the individuals to obtain individuals after the local area expansion search, including:

[0039] For individuals after the initial search in the local area, re-determine the fitness of each individual;

[0040] For any individual after the initial search in the local area, randomly match it with an individual with greater fitness to obtain the individual to be learned corresponding to each individual; among which, the individual to be learned corresponding to the optimal individual is itself;

[0041] According to the individual to be learned corresponding to each individual, the individual is searched in the local area expansion, and the individual after the local area expansion search is obtained as follows: ;in, Indicates t During the training process n individuals after the initial search in the local area, Represents the individual after the local area expansion search , n =1,2,…,I, I represents the total number of individuals in the training population, represents the third learning rate, represents pi, represents the spiral search factor between (-1,1), represents a natural constant, b represents the spiral progressive search shape factor, Represents an individual The corresponding individual to be learned, pm (nn) represents the individual to be learned The corresponding fitness ranking, pm(n) represents the individual The corresponding fitness ranking.

[0042] In a possible implementation, for individuals after the local area expansion search, a global jump greedy search strategy is used to perform a global area jump search on the individuals to obtain individuals after the global area jump search, including:

[0043] For any individual after the local area expansion search, randomly match the first differential evolution individual and the second differential evolution individual for it;

[0044] According to the current number of training times, the adaptive differential evolution coefficient is obtained as: ;in, represents the adaptive differential evolution coefficient, represents a natural constant, represents the first random number between (0,1), T represents the maximum number of training times, represents pi;

[0045] According to the first differential evolution individuals, the second differential evolution individuals and the adaptive differential evolution coefficient, a global regional jump search is performed on the individuals, and the individuals after the global regional jump search are obtained as follows: ;in, Indicates jIndividuals after local area expansion search, j =1,2,…,I, I represents the total number of individuals in the training population, Represents an individual The corresponding first differential evolution individual, Represents an individual The corresponding second differential evolution individual, Represents the individual after the global area jump search , represents the second random number between (0,1), represents the third random number between (0,1), represents the fourth random number between (0,1), Represents the fifth random number between (0,1);

[0046] Determine whether the fitness of the individual after the global area jump search increases. If so, accept the global area jump search, otherwise reject the global area jump search.

[0047] In a possible implementation, at least one target personalized recommendation service is recommended to a guest to be accommodated, and at least one target personalized recommendation service is integrated with historical personalized recommendation service data to predict hotel supplies consumption, and the hotel supplies consumption is transmitted to a warehouse manager, including:

[0048] Recommend at least one target personalized recommendation service to the customer waiting to check in;

[0049] Based on historical personalized recommendation service data, determine the target probability of target customer profiles selecting personalized recommendation services;

[0050] Determine whether the target probability is greater than the set threshold. If so, determine the hotel supplies consumption corresponding to each target personalized recommendation service, and transmit the hotel supplies consumption corresponding to each target personalized recommendation service to the warehouse manager. Otherwise, set the hotel supplies consumption to zero and end the process.

[0051] In a second aspect, the present application provides a hotel operation management method based on big data, comprising:

[0052] Obtaining customer feature big data and customer portrait labels corresponding to the manually marked customer feature big data, and preprocessing the customer feature big data and the customer portrait labels corresponding to the customer feature big data to obtain customer portrait labels corresponding to the preprocessed customer feature big data;

[0053] An artificial intelligence model is used to build a customer portrait recognition model, and the customer portrait label corresponding to the pre-processed customer feature big data is used to train the customer portrait recognition model to obtain a trained customer portrait recognition model;

[0054] Collect target customer feature big data corresponding to the customers to be accommodated in real time, and use the trained customer portrait recognition model to recognize the target customer feature big data to obtain the target customer portrait corresponding to the customers to be accommodated;

[0055] Based on the target customer portrait corresponding to the customer to be accommodated, a search and match is performed in a pre-set relationship library between customer portraits and personalized recommendation services to determine at least one target personalized recommendation service; wherein each customer portrait in the relationship library has an association relationship with at least one personalized recommendation service;

[0056] At least one target personalized recommendation service is recommended to the customer waiting to check in, and at least one target personalized recommendation service is integrated with historical personalized recommendation service data to predict hotel supplies consumption, and the hotel supplies consumption is transmitted to the warehouse manager to dynamically configure hotel resources.

[0057] The present application provides a hotel operation management system and method based on big data. The customer portrait recognition model is constructed by adopting an artificial intelligence model, and the customer portrait recognition model is trained by adopting the customer portrait label corresponding to the customer feature big data after preprocessing to obtain the trained customer portrait recognition model. The target customer portrait of the customer to be checked in is then identified by the customer portrait recognition model, so as to realize personalized service recommendation according to the target customer portrait, which can effectively enhance the competitiveness of the hotel. At the same time, the consumption of hotel supplies can be predicted according to the personalized recommendation service, so as to realize the dynamic configuration of hotel resources and enhance the customer experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0059] Figure 1 A schematic diagram of the structure of a hotel operation management system based on big data provided in an embodiment of the present application.

[0060] Figure 2 A flowchart of a hotel operation management method based on big data provided in an embodiment of the present application.

[0061] The above drawings have shown clear embodiments of the present application, which will be described in more detail later. These drawings and text descriptions are not intended to limit the scope of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0062] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims.

[0063] The embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0064] like Figure 1 As shown, the embodiment of the present application provides a hotel operation management system based on big data, including: a historical data acquisition module 101, an artificial intelligence model training module 102, a customer portrait recognition module 103, a personalized service acquisition module 104 and a resource comprehensive management module 105;

[0065] The historical data acquisition module 101 is used to acquire customer feature big data and customer portrait tags corresponding to the customer feature big data manually marked, and pre-process the customer feature big data and the customer portrait tags corresponding to the customer feature big data to obtain the customer portrait tags corresponding to the pre-processed customer feature big data;

[0066] Customer feature big data can be data from the hotel or the entire hotel booking platform, but hotel data is often insufficient. Therefore, data from the hotel booking platform can be collected to achieve a more accurate customer portrait. Customer portrait labels can be manually marked by management personnel and only need to be marked once during the initial training process, so that customer portrait labels can be directly identified in the future, reducing the requirements for hotel operators and improving service accuracy.

[0067] The customer feature big data and the customer portrait labels corresponding to the customer feature big data are preprocessed, which may be missing value processing, normalization processing, etc., to make the data more regular.

[0068] The artificial intelligence model training module 102 is used to construct a customer portrait recognition model using an artificial intelligence model, and train the customer portrait recognition model using the customer portrait labels corresponding to the pre-processed customer feature big data to obtain a trained customer portrait recognition model;

[0069] Using an artificial intelligence model to construct a customer portrait recognition model may include: using a convolutional neural network model to construct a customer portrait recognition model, but it is worth noting that the convolutional neural network is only a preferred method of the embodiment of the present application, and other neural network models may also be used to construct a customer portrait recognition model.

[0070] After building the customer portrait model, it is also necessary to optimize the hyperparameters of the customer portrait recognition model before the given task can be achieved. Therefore, the customer portrait recognition model can be trained using the customer portrait labels corresponding to the preprocessed customer feature big data.

[0071] The customer portrait recognition module 103 is used to collect target customer feature big data corresponding to the customer to be accommodated in real time, and use the trained customer portrait recognition model to recognize the target customer feature big data to obtain the target customer portrait corresponding to the customer to be accommodated;

[0072] After collecting the target customer feature big data corresponding to the customers waiting to check in in real time, the target customer feature big data should also be pre-processed to ensure the consistency of the data structure, so as to achieve accurate target customer portrait recognition.

[0073] The personalized service acquisition module 104 is used to search and match the target customer profile corresponding to the customer to be accommodated in the preset relationship library between the customer profile and the personalized recommendation service based on the target customer profile corresponding to the customer to be accommodated, and determine at least one target personalized recommendation service; wherein each customer profile in the relationship library has an association relationship with at least one personalized recommendation service;

[0074] The relationship library is a library composed of pre-set relationships between personalized recommendation services and customer profiles. After matching the target personalized recommendation service, it can arrange and recommend the target personalized recommendation service in descending order of the number of selections based on the number of times historical customers with the same target customer profile have selected the personalized recommendation service.

[0075] The comprehensive resource management module 105 is used to recommend at least one target personalized recommendation service to the guests waiting to check in, and to merge at least one target personalized recommendation service with historical personalized recommendation service data, to predict the consumption of hotel supplies, and to transmit the consumption of hotel supplies to the warehouse manager to dynamically configure hotel resources.

[0076] The present application provides a hotel operation management system based on big data. The customer portrait recognition model is constructed by adopting an artificial intelligence model, and the customer portrait recognition model is trained by adopting the customer portrait label corresponding to the customer feature big data after preprocessing to obtain the trained customer portrait recognition model. The target customer portrait of the customer to be checked in is then identified by the customer portrait recognition model, so as to realize personalized service recommendation according to the target customer portrait, which can effectively enhance the competitiveness of the hotel. At the same time, the consumption of hotel supplies can be predicted according to the personalized recommendation service, so as to realize the dynamic configuration of hotel resources and enhance the customer experience.

[0077] In a possible implementation, obtaining customer feature big data and customer portrait labels corresponding to the manually labeled customer feature big data includes:

[0078] Obtain the gender, age, region, historical ratings of accommodation, housing type, whether coupons were used during accommodation, and booking channels of historical guests, and obtain customer feature big data; for example, a piece of information of historical guests includes gender, age, region, historical ratings of accommodation, housing type, whether coupons were used during accommodation, and booking channels. Multiple pieces of information can form a data matrix, i.e. customer feature big data. Enter each piece of data, although there are duplicate parts, but it does not affect the final data recognition effect, and ensures the regularity of the data.

[0079] The customer feature big data is displayed to management personnel to obtain customer portrait labels corresponding to the customer feature big data marked by the management personnel.

[0080] It is worth noting that when obtaining customer feature big data, the most recent N pieces of data can be selected to ensure the uniformity of the data. When the data is less than N pieces, the data of such users can be filled with zeros or removed. This embodiment preferably removes the data corresponding to users with less than N pieces of historical data, and the above customer feature big data can also add other features to make user portrait recognition more accurate.

[0081] In a possible implementation, preprocessing the customer feature big data and the customer portrait label corresponding to the customer feature big data to obtain the customer portrait label corresponding to the preprocessed customer feature big data includes:

[0082] Numerical processing is performed on the customer feature big data and the customer portrait label corresponding to the customer feature big data to convert the text data into digital data, thereby obtaining the customer feature big data after numerical processing and the customer portrait label corresponding to the customer feature big data;

[0083] Perform outlier processing on the customer feature big data after numerical processing and the customer portrait label corresponding to the customer feature big data to obtain the customer feature big data after outlier processing and the customer portrait label corresponding to the customer feature big data;

[0084] For example, when garbled characters or missing values ​​appear, the abnormal data can be removed to ensure the accuracy of the data.

[0085] The customer feature big data after outlier processing and the customer portrait labels corresponding to the customer feature big data are normalized to obtain the customer portrait labels corresponding to the customer feature big data after preprocessing.

[0086] Since some data may be large, normalization is performed on the data to reduce data complexity and improve data processing efficiency.

[0087] In a possible implementation, an artificial intelligence model is used to construct a customer portrait recognition model, and the customer portrait recognition model is trained using customer portrait labels corresponding to the preprocessed customer feature big data to obtain a trained customer portrait recognition model, including:

[0088] An artificial intelligence model is used to build a customer portrait recognition model, and the hyperparameters of the customer portrait recognition model are used as parameters to be trained and the initial training population;

[0089] The hyperparameters (such as weights) of the customer profile recognition model generally have upper and lower limits. Hyperparameters can be randomly generated between the upper and lower limits and encoded into a vector to obtain an individual. After repeated multiple times, multiple different individuals can be obtained, and these individuals can be used to form a training population.

[0090] Based on the customer portrait label corresponding to the pre-processed customer feature big data, the customer portrait recognition model is used to obtain the fitness corresponding to each individual, and the individual with the largest fitness is selected as the optimal individual;

[0091] You can first use the cross entropy loss function or the root mean square loss function to obtain the loss function value corresponding to each individual, and then after taking the negative of the loss function value, you can get the fitness corresponding to each individual. The individual with the largest fitness is the optimal individual.

[0092] According to the optimal individual, a social information exchange strategy with adaptive cosine weights is used to perform adaptive learning on each individual in the training population to obtain an individual after adaptive learning;

[0093] For individuals after adaptive learning, the population boundary reference search strategy is used to perform an initial search in the local area to obtain individuals after the initial search in the local area;

[0094] For individuals after the initial search in the local area, a spiral progressive search strategy is used to perform local area expansion search on the individuals to obtain individuals after the local area expansion search;

[0095] For individuals after the local area expansion search, the global jump greedy search strategy is used to perform global area jump search on the individuals to obtain the individuals after the global area jump search;

[0096] Determine whether the number of training times has reached the maximum number of training times. If so, re-determine the optimal individual based on the individual after the global area jump search, and apply the hyperparameters contained in the optimal individual to the customer portrait recognition model to obtain the trained customer portrait recognition model. Otherwise, return to the step of determining the optimal individual.

[0097] In the process of hyperparameter optimization in the prior art, there are problems such as falling into local optimality, slow convergence speed and poor convergence accuracy. Therefore, an embodiment of the present application provides a training method for a customer portrait recognition model to improve the training effect of the customer portrait recognition model, thereby improving the accuracy of customer portrait recognition.

[0098] In a possible implementation, according to the optimal individual, a social information exchange strategy with adaptive cosine weights is used to perform adaptive learning on each individual in the training population to obtain an individual after adaptive learning, including:

[0099] According to the current number of training times, the adaptive inertia weight is obtained as: ;in, represents the adaptive inertia weight, represents pi, t represents the current number of training times, and T represents the preset maximum number of training times;

[0100] Adaptive inertia weight is used to i By weighting each individual, the first weighted term is: ;in, Indicates t During the training process i Individuals, i =1,2,…,I, where I represents the total number of individuals in the training population;

[0101] Randomly generate the first learning factor, the second learning factor and the third learning factor between (0,1);

[0102] According to the first learning factor, let iIndividual By learning the optimal individual, the first learning item is: ;in, represents the first constant factor, represents the first learning factor, represents the optimal individual;

[0103] According to the second learning factor, let i Individual The best individual in the last training process is learned, and the second learning item is obtained as follows: ;in, represents the second constant factor, represents the second learning factor, represents the best individual in the t-1th training process, that is, the best individual in the previous training process;

[0104] According to the first weighted item, the first learning item, the second learning item and the third learning factor, the individual after adaptive learning is obtained as follows: ;in, represents the third learning factor, Represents the individual after adaptive learning .

[0105] The social information exchange strategy with adaptive cosine weights provided in the embodiment of the present application can make the algorithm have a larger weight in the early stage of iteration, so that the discoverer can search widely in the search space, which is beneficial to improving the global search performance of the algorithm. Then, by searching the optimal position in different training processes, the search in the optimal direction can be effectively realized. In addition, a first learning factor, a second learning factor and a third learning factor are set, which have a certain volatility and can avoid falling into the local optimum too early to a certain extent.

[0106] In a possible implementation, for individuals after adaptive learning, a population boundary reference search strategy is used to perform an initial local area search on the individuals to obtain individuals after the initial local area search, including:

[0107] According to the fitness of each individual, the contribution of each individual is determined as: ;in, Indicates t During the training process m The fitness of an individual after adaptive learning, m =1,2,…,I, I represents the total number of individuals in the training population, Indicates m The corresponding contribution of each individual after adaptive learning;

[0108] According to the sharing degree of each individual, the center of gravity of the training population is obtained as: ;in, represents the centroid position of the training population, Indicates m An individual after adaptive learning;

[0109] According to the centroid position of the training population and the worst individual, the first reference search item is obtained as: ;in, represents the first learning rate, and is a random number uniformly distributed between [0,1]. Indicates t The worst individual in the training process;

[0110] According to the center of gravity position of the training population and the optimal individual, the second reference search item is obtained as: ;in, represents the second learning rate, and is a random number uniformly distributed between [0,1]. Indicates t The best individual in the training process;

[0111] According to the first reference search item and the second reference search item, an initial local area search is performed on the individual, and the individual after the initial local area search is obtained as follows: ;in, represents a random adjustment factor between (0,1), Indicates m Individuals after initial search in local areas .

[0112] The group boundary reference search strategy provided in the embodiment of the present application can search the local area based on the worst position and the optimal position, so that more local unfamiliar areas can be searched, which can improve the algorithm's ability to escape from the local optimal solution to a certain extent.

[0113] In a possible implementation, for individuals after the initial search in the local area, a spiral progressive search strategy is used to perform a local area expansion search on the individuals to obtain individuals after the local area expansion search, including:

[0114] For individuals after the initial search in the local area, re-determine the fitness of each individual;

[0115] For any individual after the initial search in the local area, randomly match it with an individual with greater fitness to obtain the individual to be learned corresponding to each individual; among which, the individual to be learned corresponding to the optimal individual is itself;

[0116] According to the individual to be learned corresponding to each individual, the individual is searched in the local area expansion, and the individual after the local area expansion search is obtained as follows: ;in, Indicates t During the training process n individuals after the initial search in the local area, Represents the individual after the local area expansion search , n =1,2,…,I, I represents the total number of individuals in the training population, represents the third learning rate, represents pi, represents the spiral search factor between (-1,1), represents a natural constant, b represents the spiral progressive search shape factor, Represents an individual The corresponding individual to be learned, pm (nn) represents the individual to be learned The corresponding fitness ranking, pm(n) represents the individual The corresponding fitness ranking.

[0117] The spiral progressive search strategy provided in the embodiment of the present application enables the entire training population to search in a spiral shape towards different and better areas, thereby maintaining the diversity of the entire training population and improving the convergence accuracy of the algorithm.

[0118] In a possible implementation, for individuals after the local area expansion search, a global jump greedy search strategy is used to perform a global area jump search on the individuals to obtain individuals after the global area jump search, including:

[0119] For any individual after the local area expansion search, randomly match the first differential evolution individual and the second differential evolution individual for it;

[0120] According to the current number of training times, the adaptive differential evolution coefficient is obtained as: ;in, represents the adaptive differential evolution coefficient, represents a natural constant, represents the first random number between (0,1), T represents the maximum number of training times, represents pi;

[0121] According to the first differential evolution individuals, the second differential evolution individuals and the adaptive differential evolution coefficient, a global regional jump search is performed on the individuals, and the individuals after the global regional jump search are obtained as follows: ;in, Indicatesj Individuals after local area expansion search, j =1,2,…,I, I represents the total number of individuals in the training population, Represents an individual The corresponding first differential evolution individual, Represents an individual The corresponding second differential evolution individual, Represents the individual after the global area jump search , represents the second random number between (0,1), represents the third random number between (0,1), represents the fourth random number between (0,1), Represents the fifth random number between (0,1);

[0122] Determine whether the fitness of the individual after the global area jump search increases. If so, accept the global area jump search, otherwise reject the global area jump search.

[0123] The global jump greedy search provided in the embodiment of the present application adopts an adaptive differential evolution strategy, which can enable the algorithm to have a larger search range in the early stage, greatly improving the global search capability, and gradually reducing the differential evolution capability in the later stage of the algorithm to improve the convergence accuracy of the algorithm. It is worth noting that in order to avoid excessive greed, the greedy strategy can also be replaced with an annealing simulation algorithm.

[0124] Optionally, after each search, each individual may be subjected to out-of-bounds processing to ensure parameter validity.

[0125] In a possible implementation, at least one target personalized recommendation service is recommended to a guest to be accommodated, and at least one target personalized recommendation service is integrated with historical personalized recommendation service data to predict hotel supplies consumption, and the hotel supplies consumption is transmitted to a warehouse manager, including:

[0126] Recommend at least one target personalized recommendation service to the customer waiting to check in;

[0127] Based on historical personalized recommendation service data, determine the target probability of target customer profiles selecting personalized recommendation services;

[0128] Determine whether the target probability is greater than the set threshold. If so, determine the hotel supplies consumption corresponding to each target personalized recommendation service, and transmit the hotel supplies consumption corresponding to each target personalized recommendation service to the warehouse manager. Otherwise, set the hotel supplies consumption to zero and end the process. The hotel supplies consumption corresponding to each target personalized recommendation service is fixed, which can be used to determine the hotel supplies consumption.

[0129] Optionally, the hotel supplies consumption corresponding to the target personalized recommendation service with the most historical selection times may be transmitted to the warehouse manager, or each hotel supplies consumption may be transmitted to the warehouse manager, so that the warehouse manager can dynamically configure hotel resources.

[0130] like Figure 2 As shown, the embodiment of the present application provides a hotel operation management method based on big data, including:

[0131] S201, obtaining customer feature big data and customer portrait labels corresponding to the manually marked customer feature big data, and preprocessing the customer feature big data and the customer portrait labels corresponding to the customer feature big data to obtain customer portrait labels corresponding to the preprocessed customer feature big data;

[0132] S202, using an artificial intelligence model to build a customer portrait recognition model, and using the customer portrait label corresponding to the pre-processed customer feature big data to train the customer portrait recognition model to obtain a trained customer portrait recognition model;

[0133] S203, collecting target customer feature big data corresponding to the customer to be checked in in real time, and using the trained customer portrait recognition model to recognize the target customer feature big data, to obtain the target customer portrait corresponding to the customer to be checked in;

[0134] S204, based on the target customer portrait corresponding to the customer to be accommodated, searching and matching in a preset relationship library between customer portraits and personalized recommendation services, and determining at least one target personalized recommendation service; wherein each customer portrait in the relationship library has an association relationship with at least one personalized recommendation service;

[0135] S205, recommending at least one target personalized recommendation service to the customer waiting to check in, and integrating at least one target personalized recommendation service with historical personalized recommendation service data, predicting hotel supplies consumption, and transmitting the hotel supplies consumption to the warehouse manager to dynamically configure hotel resources.

[0136] The hotel operation management method based on big data provided in the embodiment of the present application can be applied to the above-mentioned system embodiment. Its principles and beneficial effects are similar and will not be repeated here.

[0137] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0138] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0139] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0140] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0141] A person of ordinary skill in the art can understand that all or part of the steps in realizing the above-mentioned facts and methods can be completed by instructing the relevant hardware through a program, and the program involved or the program described can be stored in a computer-readable storage medium. When the program is executed, it includes the following steps: At this time, the corresponding method steps are derived, and the storage medium can be ROM / RAM, a disk, an optical disk, etc.

[0142] The specific implementation methods described above further illustrate the purpose, technical solutions and beneficial effects of the present application in detail. It should be understood that the above description is only the specific implementation method of the present application and is not intended to limit the scope of protection of the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the scope of protection of the present application.

Claims

1. A hotel operation management system based on big data, characterized in that: include: Historical data acquisition module, artificial intelligence model training module, customer portrait recognition module, personalized service acquisition module and comprehensive resource management module; The historical data acquisition module is used to acquire customer feature big data and customer portrait labels corresponding to the manually marked customer feature big data, and preprocess the customer feature big data and the customer portrait labels corresponding to the customer feature big data to obtain the customer portrait labels corresponding to the preprocessed customer feature big data; The artificial intelligence model training module is used to construct a customer portrait recognition model using an artificial intelligence model, and train the customer portrait recognition model using the customer portrait labels corresponding to the pre-processed customer feature big data to obtain a trained customer portrait recognition model; The customer portrait recognition module is used to collect target customer feature big data corresponding to the customer to be checked in in real time, and use the trained customer portrait recognition model to recognize the target customer feature big data to obtain the target customer portrait corresponding to the customer to be checked in; The personalized service acquisition module is used to search and match the target customer portrait corresponding to the customer to be accommodated in the relationship library between the preset customer portraits and personalized recommendation services based on the target customer portrait corresponding to the customer to be accommodated, and determine at least one target personalized recommendation service; wherein each customer portrait in the relationship library has an association relationship with at least one personalized recommendation service; The resource comprehensive management module is used to recommend at least one target personalized recommendation service to the customer waiting to check in, and merge at least one target personalized recommendation service with historical personalized recommendation service data to predict the consumption of hotel supplies, and transmit the consumption of hotel supplies to the warehouse manager to dynamically configure hotel resources; The customer portrait recognition model is constructed by using an artificial intelligence model, and the customer portrait recognition model is trained by using the customer portrait label corresponding to the pre-processed customer feature big data to obtain the trained customer portrait recognition model, including: Use the artificial intelligence model to build a customer portrait recognition model, and use the hyperparameters of the customer portrait recognition model as the parameters to be trained to initialize the training population; Based on the customer portrait label corresponding to the preprocessed customer feature big data, the fitness corresponding to each individual is obtained, and the individual with the largest fitness is selected as the optimal individual; According to the optimal individual, a social information exchange strategy with adaptive cosine weights is used to perform adaptive learning on each individual in the training population to obtain an individual after adaptive learning; For individuals after adaptive learning, the population boundary reference search strategy is used to perform an initial search in the local area to obtain individuals after the initial search in the local area; For individuals after the initial search in the local area, a spiral progressive search strategy is used to perform local area expansion search on the individuals to obtain individuals after the local area expansion search; For individuals after local area expansion search, a global jump greedy search strategy is used to perform global area jump search on the individuals to obtain individuals after global area jump search; Determine whether the number of training times has reached the maximum number of training times. If so, re-determine the optimal individual based on the individual after the global area jump search, and apply the hyperparameters contained in the optimal individual to the customer portrait recognition model to obtain the trained customer portrait recognition model. Otherwise, return to the step of determining the optimal individual.

2. The hotel operation management system based on big data according to claim 1 is characterized in that: Obtain customer feature big data and customer portrait labels corresponding to the manually labeled customer feature big data, including: Obtain the gender, age, region, historical ratings of accommodation, housing type, whether coupons were used during accommodation, and booking channels of historical guests, and obtain big data on customer characteristics; The customer feature big data is displayed to management personnel to obtain customer portrait labels corresponding to the customer feature big data marked by the management personnel.

3. The hotel operation management system based on big data according to claim 2 is characterized in that: Preprocessing the customer feature big data and the customer portrait label corresponding to the customer feature big data to obtain the customer portrait label corresponding to the preprocessed customer feature big data includes: Numerical processing is performed on the customer feature big data and the customer portrait label corresponding to the customer feature big data to convert the text data into digital data, thereby obtaining the customer feature big data after numerical processing and the customer portrait label corresponding to the customer feature big data; Perform outlier processing on the customer feature big data after numerical processing and the customer portrait label corresponding to the customer feature big data to obtain the customer feature big data after outlier processing and the customer portrait label corresponding to the customer feature big data; The customer feature big data after outlier processing and the customer portrait labels corresponding to the customer feature big data are normalized to obtain the customer portrait labels corresponding to the customer feature big data after preprocessing.

4. The hotel operation management system based on big data according to claim 1 is characterized in that: According to the optimal individual, a social information exchange strategy with adaptive cosine weights is used to perform adaptive learning on each individual in the training population to obtain an individual after adaptive learning, including: According to the current number of training times, the adaptive inertia weight is obtained as: ;in, represents the adaptive inertia weight, represents pi, t represents the current number of training times, and T represents the preset maximum number of training times; Adaptive inertia weight is used to i By weighting each individual, the first weighted term is: ;in, Indicates t During the training i Individuals, i =1,2,…,I, where I represents the total number of individuals in the training population; Randomly generate the first learning factor, the second learning factor and the third learning factor between (0,1); According to the first learning factor, let i Individual By learning the optimal individual, the first learning item is: ;in, represents the first constant factor, represents the first learning factor, represents the optimal individual; According to the second learning factor, let i Individual The best individual in the last training process is learned, and the second learning item is obtained as follows: ;in, represents the second constant factor, represents the second learning factor, represents the best individual in the t-1th training process, that is, the best individual in the previous training process; According to the first weighted item, the first learning item, the second learning item and the third learning factor, the individual after adaptive learning is obtained as follows: ;in, represents the third learning factor, Represents the individual after adaptive learning .

5. The hotel operation management system based on big data according to claim 4 is characterized in that: For individuals after adaptive learning, the population boundary reference search strategy is used to perform an initial search in the local area to obtain individuals after the initial search in the local area, including: According to the fitness of each individual, the contribution of each individual is determined as: ;in, Indicates t During the training m The fitness of an individual after adaptive learning, m =1,2,…,I, I represents the total number of individuals in the training population, Indicates m The corresponding contribution of each individual after adaptive learning; According to the contribution of each individual, the center of gravity of the training population is obtained as follows: ;in, represents the centroid position of the training population, Indicates m An individual after adaptive learning; According to the centroid position of the training population and the worst individual, the first reference search item is obtained as: ;in, represents the first learning rate, and is a random number uniformly distributed between [0,1]. Indicates t The worst individual in the training process; According to the center of gravity position of the training population and the optimal individual, the second reference search item is obtained as: ;in, represents the second learning rate, and is a random number uniformly distributed between [0,1]. Indicates t The best individual in the training process; According to the first reference search item and the second reference search item, an initial local area search is performed on the individual, and the individual after the initial local area search is obtained as follows: ;in, represents a random adjustment factor between (0,1), Indicates m Individuals after initial search in local areas .

6. The hotel operation management system based on big data according to claim 5 is characterized in that: For individuals after the initial search in the local area, a spiral progressive search strategy is used to perform local area expansion search on the individuals, and the individuals after the local area expansion search are obtained, including: For individuals after the initial search in the local area, re-determine the fitness of each individual; For any individual after the initial search in the local area, randomly match it with an individual with greater fitness to obtain the individual to be learned corresponding to each individual; among which, the individual to be learned corresponding to the optimal individual is itself; According to the individual to be learned corresponding to each individual, the individual is searched in the local area expansion, and the individual after the local area expansion search is obtained as follows: ;in, Indicates t During the training n individuals after the initial search in the local area, Represents the individual after the local area expansion search , n =1,2,…,I, I represents the total number of individuals in the training population, represents the third learning rate, represents pi, represents the spiral search factor between (-1,1), represents a natural constant, b represents the spiral progressive search shape factor, Represents an individual The corresponding individual to be learned, pm (nn) represents the individual to be learned The corresponding fitness ranking, pm(n) represents the individual The corresponding fitness ranking.

7. The hotel operation management system based on big data according to claim 6 is characterized in that: For individuals after the local area expansion search, the global jump greedy search strategy is used to perform a global area jump search on the individuals, and the individuals after the global area jump search are obtained, including: For any individual after the local area expansion search, randomly match the first differential evolution individual and the second differential evolution individual for it; According to the current number of training times, the adaptive differential evolution coefficient is obtained as: ;in, represents the adaptive differential evolution coefficient, represents a natural constant, represents the first random number between (0,1), T represents the maximum number of training times, represents pi; According to the first differential evolution individuals, the second differential evolution individuals and the adaptive differential evolution coefficient, a global regional jump search is performed on the individuals, and the individuals after the global regional jump search are obtained as follows: ;in, Indicates j Individuals after local area expansion search, j =1,2,…,I, I represents the total number of individuals in the training population, Represents an individual The corresponding first differential evolution individual, Represents an individual The corresponding second differential evolution individual, Represents the individual after global area jump search , represents the second random number between (0,1), represents the third random number between (0,1), represents the fourth random number between (0,1), Represents the fifth random number between (0,1); Determine whether the fitness of the individual after the global area jump search increases. If so, accept the global area jump search, otherwise reject the global area jump search.

8. The hotel operation management system based on big data according to claim 7 is characterized in that: Recommending at least one target personalized recommendation service to a customer waiting to check in, fusing at least one target personalized recommendation service with historical personalized recommendation service data, predicting hotel supplies consumption, and transmitting the hotel supplies consumption to a warehouse manager, including: Recommend at least one target personalized recommendation service to the customer waiting to check in; Based on historical personalized recommendation service data, determine the target probability of target customer profiles selecting personalized recommendation services; Determine whether the target probability is greater than the set threshold. If so, determine the hotel supplies consumption corresponding to each target personalized recommendation service, and transmit the hotel supplies consumption corresponding to each target personalized recommendation service to the warehouse manager. Otherwise, set the hotel supplies consumption to zero and end the process.

9. A hotel operation management method based on big data, which is applied to the hotel operation management system described in any one of claims 1 to 8, characterized in that: include: Obtaining customer feature big data and customer portrait labels corresponding to the manually marked customer feature big data, and preprocessing the customer feature big data and the customer portrait labels corresponding to the customer feature big data to obtain customer portrait labels corresponding to the preprocessed customer feature big data; An artificial intelligence model is used to build a customer portrait recognition model, and the customer portrait label corresponding to the pre-processed customer feature big data is used to train the customer portrait recognition model to obtain a trained customer portrait recognition model; Collect target customer feature big data corresponding to the customers to be accommodated in real time, and use the trained customer portrait recognition model to recognize the target customer feature big data to obtain the target customer portrait corresponding to the customers to be accommodated; Based on the target customer portrait corresponding to the customer to be accommodated, a search and match is performed in a pre-set relationship library between customer portraits and personalized recommendation services to determine at least one target personalized recommendation service; wherein each customer portrait in the relationship library has an association relationship with at least one personalized recommendation service; At least one target personalized recommendation service is recommended to the customer waiting to check in, and at least one target personalized recommendation service is integrated with historical personalized recommendation service data to predict hotel supplies consumption, and the hotel supplies consumption is transmitted to the warehouse manager to dynamically configure hotel resources.

Images