Hotel design method and system based on user portrait vector construction

Abstract

The application provides a hotel design method and system based on a user portrait vector, the method comprising: obtaining historical data of a hotel check-in user, and constructing a user portrait vector based on the historical data; mapping the user portrait vector to a continuous first style vector; clustering the continuous first style vector to determine a core guest group style vector of the hotel; converting the core guest group style vector into guest room design parameters of different guest groups through a generative adversarial network; performing desensitization processing on the collected data based on basic data and network behavior data of a new check-in user, obtaining a deep style vector of the new check-in user based on the desensitized data, generating personalized environment adjustment parameters based on the distance between the deep style vector and the check-in guest room design parameters, and sending the personalized adjustment parameters to intelligent execution hardware of the check-in guest room. The application designs a hotel based on the characteristics of historical users, adjusts the guest room environment parameters in real time when a new user checks in, and improves the living experience of the user.

Description

Technical Field

[0001] This application relates to the field of data processing technology, specifically a hotel design method and system based on user profile vector construction. Background Technology

[0002] Hotel design encompasses the overall planning and implementation of space and experience for the hotel industry, helping owners define their hotel's positioning and achieve efficient and sustainable operation.

[0003] Traditional hotel design relies on market analysis, competitive research, or brand positioning. However, traditional hotel design is relatively coarse-grained, only meeting the needs of average users and not fully aligning with their actual requirements. Therefore, overcoming these technical problems and shortcomings becomes a key issue that needs to be addressed. Summary of the Invention

[0004] To overcome the aforementioned problems in the existing technology, this application provides a hotel design method and system based on user profile vector construction, which adopts the following technical solution:

[0005] Firstly, this application provides a hotel design method based on user profile vector construction, including:

[0006] Obtain historical data of hotel guests and construct user profile vectors based on the historical data;

[0007] Map user profile vectors to continuous first style vectors;

[0008] Cluster the continuous first style vectors to determine the style vectors of the hotel's core customer group;

[0009] The style vectors of the core customer group are transformed into room design parameters for different customer group styles through a generative adversarial network.

[0010] Based on the basic data and network behavior data of new guests, a deep style vector of the new guests is obtained. Based on the distance between the deep style vector and the design parameters of the guest room, a guest room is recommended and personalized environmental adjustment parameters are generated. The personalized adjustment parameters are then sent to the smart execution hardware in the guest room.

[0011] Furthermore, the historical data of hotel guests includes: basic data, online behavior data, and sensor data.

[0012] Furthermore, historical data of hotel guests is obtained, and user profile vectors are constructed based on this historical data, including:

[0013] The basic data, network behavior data, and sensor data are encoded separately to obtain the basic data encoding, network behavior data encoding, and sensor data encoding.

[0014] The basic data encoding, network behavior data encoding, and sensor encoded data are dimensionally aligned through a linear layer to obtain an alignment vector.

[0015] The alignment vector is input into a deep self-attention network to obtain attention scores between different modalities, and corresponding weights are assigned to each modality. Based on the attention weights, the deep self-attention network performs a weighted summation of all single-modality vectors, and obtains the user profile vector through the fully connected layers of the deep self-attention network.

[0016] Furthermore, the basic data is encoded, including: entity embedding of discrete features of the basic data, normalization of continuous features, and inputting the entity embedding and normalization results into the fully connected layer to obtain the basic data encoding.

[0017] Furthermore, the network behavior data is encoded, including: using a transformer encoder to capture the latent style preference vector in the network behavior data, thereby achieving data encoding of the network behavior.

[0018] Furthermore, the sensor data is encoded, including fusing user environment adjustment records and emotion feedback data from the sensor data to obtain encoded sensor data.

[0019] Furthermore, mapping the user profile vectors to continuous first style vectors includes:

[0020] A fully connected network consisting of an input layer, hidden layers, hidden layers, and an output layer is constructed as the mapping function. During training, the mapping function performs forward propagation on each user profile vector in the triplet to obtain the corresponding style embedding point in the triplet. The loss function uses these three corresponding style embedding points as input to calculate the loss. The distance between the three embedding points is obtained. If it does not meet the preset formula, a positive loss value is generated. Based on the loss value, the gradient of the loss function with respect to all parameters in the mapping function is calculated through backpropagation. The parameters of the mapping function are updated through the gradient, and the mapping function is updated until the geometric distance in the embedding space and the true style semantics meet the preset threshold. The trained mapping function maps the user profile vector to the first style vector, so that the first style vector has style semantics in the embedding space.

[0021] Furthermore, clustering the continuous first style vectors to determine the style vectors of the hotel's core customer group includes: based on the hotel's core positioning, pre-setting the number of clusters, and using a clustering algorithm to assign the first style vectors to the pre-set number of style clusters;

[0022] Obtain the geometric center point of each style cluster. Based on the multimodal data within the geometric center point cluster, obtain the style keywords of the customer groups in that cluster. Assign an interpretable style label to the geometric center point cluster based on the style keywords.

[0023] Furthermore, the style vectors of the core customer group are transformed into room design parameters for different customer group styles through a generative adversarial network, including:

[0024] Based on the hotel's core positioning, a preset number of clusters is used, and the first style vector is assigned to the preset number of style clusters through a clustering algorithm.

[0025] Obtain the geometric center point of each style cluster. Based on the multimodal data within the geometric center point cluster, obtain the style keywords of the customer groups in that cluster. Assign an interpretable style label to the geometric center point cluster based on the style keywords.

[0026] Secondly, this application also provides a hotel design system based on user profile vectors, including:

[0027] The user profile vector construction module is used to obtain historical data of hotel guests and construct user profile vectors based on the historical data.

[0028] The first style vector mapping module is used to map user profile vectors to continuous first style vectors;

[0029] The customer style vector determination module is used to cluster continuous first style vectors to determine the hotel's core customer style vectors.

[0030] The guest room design parameter conversion module is used to convert the style vector of the core customer group into guest room design parameters for different customer groups through a generative adversarial network.

[0031] The personalized parameter adjustment module is used to obtain the deep style vector of new guests based on their basic data and network behavior data. Based on the distance between the deep style vector and the design parameters of the guest room, it recommends guest rooms and generates personalized environmental adjustment parameters, which are then sent to the smart execution hardware in the guest room.

[0032] Thirdly, this application provides an electronic device, comprising:

[0033] One or more processors; a memory; and one or more computer programs, wherein the one or more computer programs are stored in the memory, and the one or more computer programs include instructions that, when executed by the device, cause the device to perform the method as described in the first aspect.

[0034] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when run on a computer, causes the computer to perform the method described in the first aspect.

[0035] Fifthly, this application provides a computer program that, when executed by a computer, performs the method described in the first aspect.

[0036] In one possible design, the program in the fifth aspect can be stored wholly or partially on a storage medium packaged with the processor, or it can be stored wholly or partially on a memory not packaged with the processor.

[0037] This application has the following beneficial effects:

[0038] 1. This application obtains historical data of hotel guests and constructs user profile vectors based on the historical data, wherein the historical data includes basic user data, network behavior data, and sensor data. This enables personalized design of hotels, eliminating the need to rely on market analysis, competitive research, or brand positioning during the hotel design process, thus avoiding the problem of coarse-grained hotel design.

[0039] 2. By mapping the user profile vector to a continuous first style vector, this application can transform the user's style preferences into quantifiable geometric information, thereby enabling style clustering of the hotel's historical users and further guiding the hotel's design work.

[0040] 3. This application determines the style vector of the hotel's core customer group by clustering the continuous first style vector. By extracting style keywords of different customer groups, it avoids subjectively setting style tags and has certain time coverage and customer attraction capabilities.

[0041] 4. This application transforms the core customer group style vector into room design parameters for different customer group styles through a generative adversarial network, converts the hard decoration parameters into corresponding executable instructions, and uses the soft decoration parameters as default basic settings to generate executable hotel design data.

[0042] 5. This application obtains the deep style vector of new occupants based on their basic data and network behavior data. Based on the distance between the deep style vector and the design parameters of the guest room, it recommends guest rooms and generates personalized environmental adjustment parameters. The personalized adjustment parameters are sent to the intelligent execution hardware of the guest room, which can dynamically adjust the basic default settings according to the initial information of the new occupants to meet their living experience. Attached Figure Description

[0043] Figure 1 This is a flowchart illustrating the hotel design method based on user profile vector construction according to an embodiment of this application.

[0044] Figure 2 This is a flowchart illustrating the user profile vector construction process of the hotel design method based on user profile vector construction, as described in this application embodiment.

[0045] Figure 3 This is a flowchart illustrating the first style vector acquisition process of the hotel design method based on user profile vector construction, as described in this application embodiment.

[0046] Figure 4 This is a flowchart illustrating the core customer group style vector acquisition process of the hotel design method based on user profile vector construction in this application embodiment.

[0047] Figure 5 This is a system flowchart of an embodiment of this application.

[0048] Figure 6 This is a schematic diagram of a computer device according to an embodiment of this application. Detailed Implementation

[0049] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the terminology used herein in the specification of the application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application; the terms "comprising" and "having," and any variations thereof, in the specification, claims, and foregoing drawings of this application, are intended to cover non-exclusive inclusion. The terms "first," "second," etc., in the specification, claims, or foregoing drawings of this application are used to distinguish different objects, not to describe a particular order.

[0050] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0051] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0052] Please refer to Figure 1 The present application provides a hotel design method based on user profile vector construction, the details of which are as follows:

[0053] Step S1: Obtain historical data of hotel guests and construct a user profile vector based on the historical data, wherein the historical data includes the user's basic data, network behavior data, and sensor data.

[0054] It should be noted that the online behavior data in this application includes users' browsing content on the Internet, booking preference tags, filtering data, and comparison records.

[0055] It should be noted that this application, by collecting historical data from hotel guests, utilizes user emotional feedback data and historical interaction data during the user profile vector construction process to more accurately obtain users' style preference characteristics. It also obtains users' potential tendencies through online behavior data, with booking preference tags and filtering data from this data serving as guidance for room functionality design. In subsequent hotel room booking recommendations, initial room configurations tailored to user characteristics can be generated based on the user's online behavior data.

[0056] In this embodiment, historical data of hotel guests is obtained, and user profile vectors are constructed based on the historical data. Please refer to [reference needed]. Figure 2 The specific content includes:

[0057] Step 11: Entity embedding is performed on the discrete features of the basic data, and continuous features are normalized. The entity embedding and normalization results are input into the fully connected layer to obtain the basic data encoding.

[0058] It should be noted that discrete features can be users' gender, occupation, etc.; continuous features can be users' age and consumption level, etc.

[0059] This application achieves encoding of basic data by capturing the potential relationships between users' basic data.

[0060] It should be noted that the basic data in this application embodiment can provide stable style prior knowledge even when user network behavior data and sensor data are missing.

[0061] The entity embedding technique in this application is a technique that maps discrete categorical variables to a low-dimensional vector space. Through entity embedding, even if the user occupation information obtained later is uncommon, it can still make associations in the embedding space based on similar occupations. In the subsequent feature fusion stage, similar styles are fused to improve the generalization ability for sparse features.

[0062] Step 12: The user's network behavior data is encoded by capturing the latent style preference vector in the network behavior data through the transformer encoder.

[0063] It should be noted that, in this embodiment of the application, a transformer encoder is used to extract key intent weights from user browsing behavior and keyword search sequences in the user's network behavior, thereby obtaining the user's potential style preferences.

[0064] Since user network behavior data may span a long period of time and generate multiple click behaviors, the self-attention mechanism of the transformer encoder can capture long-distance dependencies in network behavior. At the same time, the embodiments of this application can assign higher weights to core intent behaviors through the attention weights of the transformer, thereby improving the quality and accuracy of data fusion during the data fusion process.

[0065] In this embodiment of the application, the latent style preference vector in the user's network behavior data is captured by a transformer encoder to achieve data encoding of network behavior, including:

[0066] Based on the original network behavior sequence, each network behavior is transformed into a feature vector. The position code associated with each sequence position is calculated, and the position code is added to the corresponding feature vector to obtain the input feature vector.

[0067] The transformer encoder projects the input feature vector into a query, key, and value matrix; the similarity scores of the query and key are calculated, and the results are normalized using softmax to obtain the attention weight matrix. The weight matrix is ​​then multiplied by the value matrix to obtain a new sequence representation.

[0068] The new sequence representation is residually connected to the input feature vector and subjected to layer normalization. A feedforward neural network is then used for non-linear activation and feature mapping to obtain an enhanced feature representation. The final layer of the encoder performs global average pooling on the enhanced feature representation, aggregating it into a fixed-dimensional vector. This fixed-dimensional vector is used as the data encoding for network behavior. This application avoids losing crucial information during the calculation of the attention weight matrix by using a new sequence representation with residual connections and layer normalization.

[0069] It should be noted that this application uses a transformer encoder to encode the user's network behavior data, transforming the user's network behavior into mathematical vectors, thereby providing raw training data for hotel design.

[0070] It should be noted that, in the process of capturing latent style preference vectors through the self-attention mechanism of the transformer encoder, this application assigns corresponding weights to each user behavior to form a comprehensive style vector.

[0071] Step 13: Fuse the user environment adjustment records and emotion feedback data from the sensor data to obtain sensor coded data.

[0072] In this embodiment of the application, user environment adjustment records and emotion feedback data from sensor data are fused to obtain sensor coded data, including:

[0073] The user's environmental conditioning record is input into the Long Short-Term Memory (LSTM) network. The user's environmental conditioning record is encoded through a multi-layer LSTM network. The hidden state vector of the last time step in the LSTM network is obtained as the first feature.

[0074] The emotion feedback data is concatenated and input into the BERT encoder to obtain the final hidden state vector corresponding to the CLS marker position in the BERT output, which is used as the second feature.

[0075] The first and second features are mapped to a unified fusion dimension through independent linear layers. The mapped first and second features are then concatenated to obtain the sensor fusion vector, which is used as the sensor encoding data.

[0076] In this embodiment of the application, the number of emotion feedbacks includes at least one of text data and voice data.

[0077] It should be noted that this application can obtain users' environmental adjustment records through user data on environmental adjustments, such as lighting dimming records, air conditioner temperature adjustment records, and curtain opening and closing times. When obtaining users' emotional feedback data, this application can be based on users' comments or voice information.

[0078] Step 14: Align the basic data encoding, network behavior data encoding, and sensor encoded data through a linear layer to obtain the alignment vector.

[0079] In this embodiment of the application, the basic data encoding, network behavior data encoding, and sensor encoded data are dimensionally aligned through a linear layer to obtain an alignment vector, including:

[0080] Based on a preset target dimension, a learnable linear layer is defined for each modality code. Each encoded vector is input into the corresponding linear layer to obtain a new aligned vector.

[0081] Step 15: Input the alignment vector into the deep self-attention network to obtain the attention scores between different modalities, and assign corresponding weights to each modality. The deep self-attention network performs a weighted summation of all single-modality vectors based on the attention weights, and obtains the user profile vector through the fully connected layer of the deep self-attention network.

[0082] It should be noted that this application can achieve personalized hotel design by obtaining users' basic data, network behavior data, and sensor data. In the process of hotel design, it no longer relies on market analysis, competitive research, or brand positioning, thus avoiding the problem of coarse-grained hotel design.

[0083] Step S2: Map the user profile vector to a continuous first style vector.

[0084] In this embodiment, the user profile vector is mapped to a continuous first style vector. Please refer to [link / reference]. Figure 3 ,include:

[0085] Step 21: Construct a fully connected network consisting of an input layer, hidden layers, hidden layers, and an output layer, which will serve as the mapping function.

[0086] Step 22: During training, the mapping function performs forward propagation on each user profile vector in the triplet to obtain the corresponding style embedding point in the triplet. The loss function uses these three corresponding style embedding points as input to calculate the loss.

[0087] Step 23: Obtain the distance between the three embedding points. If the distance does not meet the preset formula, generate a positive loss value. Based on the loss value, calculate the gradient of the loss function with respect to all parameters in the mapping function through backpropagation. Update the parameters of the mapping function through the gradient and update the mapping function until the geometric distance in the embedding space and the true style semantics meet the preset threshold.

[0088] Step 24: The user profile vector is mapped to the first style vector through the trained mapping function, so that the first style vector has style semantics in the embedding space.

[0089] It should be noted that the triple includes anchor samples, positive samples, and negative samples. Anchor samples are basic descriptions of user style features during training, positive samples are those with style preferences consistent with anchor samples, and negative samples are those with style preferences conflicting with anchor samples.

[0090] It should be noted that the first style vector in this application represents the geometric coordinates obtained by mapping user data in a continuous embedding space. Each value in the first style vector represents the intensity of a possible style feature. The first style vector can transform the user's style preferences into quantifiable geometric information, which can then be used to cluster the hotel's historical users by style, further guiding the hotel's design work.

[0091] It should be noted that this application converts user data into geometric information, thereby enabling the quantitative processing of hotel style during the hotel design process.

[0092] It should be noted that this application uses historical data of hotel guests as a supervisory signal to extract the first style vector. This historical data serves as a knowledge base for style parameters in hotel design, enabling the rapid generation of hotel design parameters that meet the user's needs. This provides a more accurate data source for hotel design and offers stronger data support for the design process.

[0093] Step S3: Cluster the continuous first style vectors to determine the style vectors of the hotel's core customer group.

[0094] In this embodiment, the continuous first style vectors are clustered to determine the style vectors of the hotel's core customer group. Please refer to [link / reference]. Figure 4 ,include:

[0095] Step 31: Based on the hotel's core positioning, preset the number of clusters, and use a clustering algorithm to assign the first style vector to the preset number of style clusters.

[0096] Step 32: Obtain the geometric center point of each style cluster. Based on the multimodal data within the geometric center point cluster, obtain the style keywords of the customer groups in that cluster. Assign an interpretable style tag to the geometric center point cluster based on the style keywords. The interpretable style tag can be minimalist tech style or new Chinese minimalist style, etc.

[0097] In this embodiment, the geometric center point is the core customer group style vector. Here, the geometric center is the core customer group of each cluster, which is a group of users with similar style preferences.

[0098] It should be noted that this application can extract style keywords for different customer groups by clustering the first style vector, avoiding the subjective setting of style tags, and does not have a certain time coverage capability or customer attraction capability.

[0099] Step S3 of this application can transform the user's style into data that guides the design of hotel style types, thereby further realizing precise market positioning and style design for hotel design.

[0100] Step S4: The style vector of the core customer group is transformed into room design parameters for different customer group styles through a generative adversarial network.

[0101] In this embodiment of the application, the core customer group style vector is transformed into room design parameters for different customer group styles through a generative adversarial network, including:

[0102] Use the style vectors of the core customer group as input conditions for the generator.

[0103] The input conditions are used as constraints to generate a room configuration vector representing the style of the corresponding customer group.

[0104] The hard furnishing parameters in the configuration vector of each guest room are semantically transformed using a Gaussian regression process to determine the basic room type with a preset clustering style.

[0105] It should be noted that semantic transformation through Gaussian regression converts the numerical values ​​in the guest room configuration vector into instructions that engineers can execute, such as the color, material, and texture of the furniture used.

[0106] Use the soft furnishings section in the guest room configuration vector as the default basic settings for the corresponding room type.

[0107] It should be noted that this application transforms the style vector of the core customer group into room design parameters by using generative adversarial networks, which can shorten the hotel design cycle and ensure that the hotel design is centered on the needs of the customers, rather than depending on the personal aesthetics of the designers.

[0108] In this embodiment of the application, the guest room design parameters include hard furnishings and soft furnishings.

[0109] Step S5: Based on the basic data and network behavior data of the new resident, obtain the deep style vector of the new resident. Based on the distance between the deep style vector and the design parameters of the guest room, recommend guest rooms and generate personalized environmental adjustment parameters. Send the personalized adjustment parameters to the smart execution hardware of the guest room.

[0110] In this embodiment of the application, based on the basic data and network behavior data of new resident users, a deep style vector of the new resident user is obtained, and a recommended room is provided, including:

[0111] The discrete features in the basic data of the newly registered users are embedded into entities, and the continuous features are normalized. The results of entity embedding and normalization are input into a fully connected layer to obtain the second basic data encoding.

[0112] The user's network behavior data is encoded by a transformer encoder to capture the latent style preference vector in the network behavior data, thereby obtaining the second network behavior data encoding.

[0113] The second basic data encoding and the second network behavior data encoding are dimensionally aligned through a linear layer to obtain the second alignment vector.

[0114] The second alignment vector is input into a deep self-attention network to obtain attention scores between different modalities, and corresponding weights are assigned to each modality. The deep self-attention network performs a weighted summation of all single-modal vectors based on the attention weights, and obtains the deep style vector of the newly registered user through the fully connected layer of the deep self-attention network.

[0115] In this embodiment, based on the distance between the deep style vector and the design parameters of the guest room, a guest room is recommended, and personalized environmental adjustment parameters are generated. These personalized adjustment parameters are then sent to the intelligent execution hardware of the guest room, including:

[0116] The deep style vector is used as the query vector and input into the guest room design parameters. The Euclidean distance between the deep style vector and all guest room design parameters is calculated.

[0117] The room design parameters with the shortest distance are selected as the most suitable room style, and the corresponding room type is matched based on the most suitable room style.

[0118] The default basic settings matching the corresponding room type are used as personalized environmental adjustment parameters and sent to the smart execution hardware in the guest room via preset API commands.

[0119] It should be noted that this application can match corresponding rooms to users based on their basic information even when their online behavior is missing. This application can predict users' potential style preferences based on the online behavior data of newly checked users. This online behavior data can be obtained through pre-set official channels, such as the hotel's official website or other app platforms.

[0120] Example 2 provides a hotel design system based on user profile vector construction, which includes a user profile vector construction module 501, a first style vector mapping module 502, a customer group style vector determination module 503, a room design parameter conversion module 504, and a personalized parameter adjustment module 505. Wherein:

[0121] User profile vector construction module 501 is used to obtain historical data of hotel guests and construct user profile vectors based on the historical data.

[0122] The first style vector mapping module 502 is used to map user profile vectors to continuous first style vectors.

[0123] The customer style vector determination module 503 is used to cluster the continuous first style vectors to determine the core customer style vector of the hotel.

[0124] The guest room design parameter conversion module 504 is used to convert the core customer group style vector into guest room design parameters for different customer groups through a generative adversarial network.

[0125] The personalized parameter adjustment module 505 is used to obtain the deep style vector of the new resident based on the basic data and network behavior data of the new resident, recommend the resident room based on the distance between the deep style vector and the design parameters of the resident room, generate personalized environmental adjustment parameters, and send the personalized adjustment parameters to the smart execution hardware of the resident room.

[0126] This application acquires historical data of hotel guests and constructs user profile vectors based on this data. These user profile vectors are then mapped to continuous first-style vectors. Clustering of these continuous first-style vectors determines the style vectors of the hotel's core customer groups. These core customer group style vectors are then transformed into room design parameters for different customer group styles using a generative adversarial network. Based on the basic data and network behavior data of new guests, a deep style vector for each new guest is obtained. Based on the distance between the deep style vector and the room design parameters, room recommendations are made, and personalized environmental adjustment parameters are generated. These personalized adjustment parameters are then sent to the intelligent execution hardware in the guest rooms. This application designs hotels based on the characteristics of historical users and adjusts room environmental parameters in real time when new guests check in, thereby improving the guest's stay experience.

[0127] Example 3: This application provides an electronic device, including: one or more processors; a memory; and one or more computer programs, wherein the one or more computer programs are stored in the memory, and the one or more computer programs include instructions that, when executed by the device, cause the device to perform the method as described in Example 1.

[0128] Example 4: This application provides a computer-readable storage medium storing a computer program that, when run on a computer, causes the computer to perform the method described in Example 1.

[0129] Example 5: This application provides a computer program that, when executed by a computer, performs the method described in Example 1.

[0130] To address the aforementioned technical problems, embodiments of this application also provide a computer device. Please refer to [link / reference needed]. Figure 6 , Figure 6 This is a basic structural block diagram of the computer device in this embodiment.

[0131] The computer device 6 includes a memory 6a, a processor 6b, and a network interface 6c that are interconnected via a system bus. It should be noted that only the computer device 6 with components 6a-6c is shown in the figure; however, it should be understood that it is not required to implement all the shown components, and more or fewer components can be implemented alternatively. Those skilled in the art will understand that the computer device described here is a device capable of automatically performing numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes, but is not limited to, microprocessors, application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), digital signal processors (DSPs), embedded devices, etc.

[0132] The computer device can be a desktop computer, laptop, handheld computer, or cloud server, etc. The computer device can interact with the user via a keyboard, mouse, remote control, touchpad, or voice control.

[0133] The memory 6a includes at least one type of readable storage medium, including flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 6a may be an internal storage unit of the computer device 6, such as the hard disk or memory of the computer device 6. In other embodiments, the memory 6a may also be an external storage device of the computer device 6, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the computer device 6. Of course, the memory 6a may also include both the internal storage unit and its external storage device of the computer device 6. In this embodiment, the memory 6a is typically used to store the operating system and various application software installed on the computer device 6, such as the program code of a hotel design method based on user profiles. In addition, the memory 6a can also be used to temporarily store various types of data that have been output or will be output.

[0134] In some embodiments, the processor 6b may be a central processing unit (CPU), controller, microcontroller, microprocessor, or other data processing chip. The processor 6b is typically used to control the overall operation of the computer device 6. In this embodiment, the processor 6b is used to run program code stored in the memory 6a or process data, for example, to run the program code for the hotel design method based on user profiles.

[0135] The network interface 6c may include a wireless network interface or a wired network interface, which is typically used to establish communication connections between the computer device 6 and other electronic devices.

[0136] This application also provides another embodiment, namely, a non-volatile computer-readable storage medium storing a program of a hotel design method based on user profiles, wherein the hotel design based on user profiles can be executed by at least one processor to perform the steps of the hotel design method based on user profiles as described above.

[0137] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0138] Obviously, the embodiments described above are only some embodiments of this application, not all embodiments. The accompanying drawings show preferred embodiments of this application, but do not limit the patent scope of this application. This application can be implemented in many different forms; rather, the purpose of providing these embodiments is to provide a more thorough and comprehensive understanding of the disclosure of this application. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing specific embodiments, or make equivalent substitutions for some of the technical features. Any equivalent structures made using the content of this application's specification and drawings, directly or indirectly applied to other related technical fields, are similarly within the scope of patent protection of this application.

Claims

1. A hotel design method based on a user portrait vector construction, characterized in that, The method comprises the following steps: obtaining historical data of a hotel check-in user, and constructing a user portrait vector based on the historical data; the historical data of the hotel check-in user comprises basic data, network behavior data, and sensor data; mapping the user portrait vector to a continuous first style vector; performing clustering on the continuous first style vector to determine a core guest group style vector of the hotel, including: based on the core positioning of the hotel, presetting the number of clusters, and dividing the first style vector into the preset number of style clusters through a clustering algorithm; obtaining a geometric center point of each style cluster, obtaining style keywords of the guest group in the cluster based on the multi-modal data in the geometric center point cluster, and assigning an interpretable style label to the geometric center point cluster based on the style keywords; transforming the core guest group style vector into guest room design parameters of different guest group styles through a generative adversarial network; based on the basic data and network behavior data of a new check-in user, obtaining a deep style vector of the new check-in user, recommending a check-in room based on the distance between the deep style vector and the check-in room design parameters, and generating personalized environment adjustment parameters, and sending the personalized adjustment parameters to the intelligent execution hardware of the check-in room. 2.The hotel design method based on the user portrait vector construction according to claim 1, characterized in that, obtaining historical data of a hotel check-in user, and constructing a user portrait vector based on the historical data, comprising: respectively encoding the basic data, network behavior data, and sensor data to obtain basic data encoding, network behavior data encoding, and sensor data encoding; aligning the dimensions of the basic data encoding, network behavior data encoding, and sensor data encoding through a linear layer to obtain an alignment vector; inputting the alignment vector into a deep self-attention network to obtain attention scores between different modalities and assign corresponding weights to each modality; the deep self-attention network performs weighted summation on all single-modal vectors according to the attention weights, and obtains the user portrait vector through the fully connected layer of the deep self-attention network. 3.The hotel design method based on the user portrait vector construction according to claim 2, characterized in that, The encoding process of the basic data comprises: entity embedding of the discrete features of the basic data, normalization processing of the continuous features, inputting the entity embedding and the normalization processing result into a fully connected layer, and obtaining the basic data encoding. 4.The hotel design method based on the user portrait vector construction according to claim 2, characterized in that, The encoding process of the network behavior data comprises: capturing the latent style preference vector in the network behavior data through a transformer encoder to realize data encoding of the network behavior. 5.The hotel design method based on user profiling vector construction according to claim 2, characterized in that, The encoding process of the sensor data comprises: fusing the user environment adjustment records and emotional feedback data in the sensor data to obtain the sensor data encoding. 6.The hotel design method based on user profiling vector construction according to claim 1, wherein, The method comprises the following steps: constructing a fully connected network composed of an input layer, a hidden layer, a hidden layer, and an output layer as a mapping function; mapping the user portrait vector to the first style vector through the trained mapping function, so that the first style vector has style semantics in the embedding space. 7.The hotel design method based on user profiling vector construction according to claim 1, wherein, The method comprises the following steps: taking the core guest group style vector as the input condition of the generator; taking the input condition as a constraint to generate a guest room configuration vector representing the corresponding guest group style; The hard outfit parameter part in the configuration vector of each set of guest rooms is subjected to semantic conversion through a Gaussian regression process to determine a preset number of clustered basic room types; The soft outfit part in the guest room configuration vector is taken as the default basic setting of the corresponding room type.

8. A hotel design system based on user portrait vector construction, used to implement the hotel design method based on user portrait vector construction of claims 1-7, characterized in that, The method comprises the following steps: a user portrait vector construction module is configured to obtain historical data of users staying in the hotel and construct a user portrait vector based on the historical data; a first style vector mapping module is configured to map the user portrait vector to a continuous first style vector; a guest group style vector determination module is configured to cluster the continuous first style vector to determine a core guest group style vector of the hotel; a guest room design parameter conversion module is configured to convert the core guest group style vector into guest room design parameters of different guest groups through a generative adversarial network; a personalized parameter adjustment module is configured to obtain a deep style vector of a new staying user based on basic data and network behavior data of the new staying user, recommend a staying room based on the distance between the deep style vector and the staying room design parameters, and generate personalized environment adjustment parameters, and send the personalized adjustment parameters to the intelligent execution hardware of the staying room.

Images