Intelligent application management method and system for hotel

Abstract

The invention discloses an intelligent application management method and system for a hotel, and relates to the technical field of hotel management. An intelligent application management system for a hotel comprises a hotel user behavior reasoning module and an intelligent application management module. According to the method, the equipment application data of the user in the guest room and the behavior moving line data of the user in the hotel public space are collected, and the historical preference information of the user is fused, so that user behavior global modeling and state perception are realized, and the coverage range and the recognition precision of the system on the real behavior of the user are improved; space-time alignment and map construction are carried out on different behavior data of the user inside and outside the hotel space, deep semantic understanding is carried out on complex behavior states shown by the user at a specific moment, an accurate behavior intention label is deduced, and the intelligent and generalization ability is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of hotel management, and in particular to an intelligent application management method and system for hotels. Background Art

[0002] With the rapid development of artificial intelligence, the Internet of Things, and big data analysis technologies, smart hotel management systems are gradually being widely used in the hotel industry. Existing smart hotel management systems mainly achieve centralized control and automatic adjustment of guest room environments, lighting, air conditioning, televisions, and other equipment by deploying sensing devices, smart terminals, and back-end control platforms, thereby improving guest experience and operational efficiency. However, most existing technologies rely on fixed rules or preset scenario modes for equipment scheduling, lacking dynamic perception and intelligent response to users' personalized behavior, making it difficult to meet guests' diverse and real-time service needs. For example, existing systems usually only control air conditioning switches and lighting modes based on time periods, preset instructions, or room card status, failing to fully integrate user activity data inside and outside the hotel to make targeted service adjustments. In addition, current systems are mostly limited to device usage data within the room in terms of behavior perception, lacking effective tracking and correlation analysis of users' behavior activities in the hotel's public areas, and unable to achieve a comprehensive perception of the user's overall behavior path.

[0003] Therefore, there is a need for an intelligent hotel application management method that is based on the fusion of multi-source user behavior data and can realize the coordination of user behavior reasoning and equipment intelligent scheduling, so as to improve the adaptability of hotel equipment applications and the refinement of user services. Summary of the Invention

[0004] The present invention aims to provide an intelligent application management method and system for hotels, which are used to improve the adaptability of hotel equipment applications and the refinement of user services.

[0005] A smart application management method for a hotel, comprising the following steps: For at least one user staying in a guest room at a target hotel, real-time user device application data is collected from the user's room. User device operation data includes environmental perception data and device application switch status data. The corresponding publicly available user behavior route data and hotel preference data are obtained. The user behavior route data includes n-th behavior data, where n = 1, 2, ..., N, where n represents the n-th behavior event recorded by the user in public places in the hotel. Perform activity state inference on the current corresponding nth behavior data to obtain the user's external activity prediction data; obtain the room leaving timestamp T of the nth behavior data n ; Get the time stamp of leaving the room T n-1 Time stamp of departure T nThe user device operation data between them is combined with the inference network to analyze user behavior and obtain the user's internal activity prediction data; Align the user's internal activity prediction data and the user's external activity prediction data in time and space, and identify the room departure timestamp T n Corresponding behavioral intention labels; based on the identified behavioral intention labels, match the real-time device application update index of the guest rooms; based on the real-time device application update index, intelligently adjust the hotel device applications in the guest rooms and the hotel device applications outside the guest rooms to achieve intelligent management of the device applications in the target hotel.

[0006] As a preferred technical solution of the present invention, the specific steps of performing activity state inference on the currently corresponding n-th behavior data include: Extracting continuous spatiotemporal features based on the nth behavior data to obtain a user spatiotemporal feature set; wherein the user spatiotemporal feature set includes the behavior path, duration, and corresponding location coordinates of public places in the hotel; Obtain the 1st to n-1th behavior data to construct the behavior sequence context features; take the user's spatiotemporal feature set, behavior sequence context features and hotel preference data as input, input them into the pre-trained behavior sequence learning model for analysis, and obtain the user's external activity prediction data.

[0007] As a preferred technical solution of the present invention, obtain the room leaving timestamp T n-1 Time stamp of departure T n The specific steps for combining the user device operation data between the two devices with the inference network to perform user behavior analysis include: Identify user intentions for environmental changes based on environmental perception data; identify user intentions for actions based on device application switch status data; construct scenario features based on user intentions for environmental changes and actions to obtain an internal behavior feature vector sequence; Obtain historical stay data of users staying in guest rooms at the target hotel; establish comprehensive stay context features based on historical stay data and hotel preference data; The comprehensive living situation characteristics and internal behavior feature vector sequence are input into the inference network to analyze the room leaving timestamp T n-1 Time stamp of departure T n The device usage pattern, environmental change trend and human activity pattern within the system are combined to obtain the user's internal activity prediction data.

[0008] As a preferred technical solution of the present invention, the time stamp of the room leaving is identified. n The specific steps for corresponding behavioral intention labels include: Align the user's internal activity prediction data and the user's external activity prediction data in time and space to obtain the user's comprehensive context behavior vector; match the user's behavior intention based on the comprehensive context behavior vector to obtain the behavior intention label; The specific steps of spatiotemporal alignment are: extracting all behavioral events from the user's internal activity prediction data and the user's external activity prediction data; mapping the user's internal activity prediction data and the user's external activity prediction data to the user's behavior prediction coordinate system, and annotating the spatial position coordinates and absolute timestamps of the behavioral events according to the granularity of the behavioral events; Define dynamic time windows based on the granularity of behavioral events; establish several event correspondences for user internal activity prediction data and user external activity prediction data based on the dynamic time windows; set a behavioral response tolerance interval for each event correspondence; A spatiotemporally aligned graph structure is constructed with each behavioral event as a node and the event correspondence as an edge. The aligned graph structure is vectorized using a behavioral semantic embedding model to obtain a comprehensive user contextual behavior vector.

[0009] As a preferred technical solution of the present invention, the specific steps of matching the real-time device application update index of the guest room based on the identified behavioral intention label include: Constructing a device application control factor library containing multiple hotel device applications; the device application control factor library contains several types of device application control factors; Based on the target device application set associated with the behavior intention label matching, the corresponding device application control factor is matched in the device application control factor library for each target device application in the target device application set as the behavior association weight coefficient; the device application operating parameters of the current guest room are obtained; and the real-time device application update index is calculated by combining the device application operating parameters and the behavior association weight coefficient.

[0010] As a preferred technical solution of the present invention, the inference network uses the Bayesian inference network as the basic network structure to perform user behavior analysis.

[0011] An intelligent application management system for a hotel, comprising: The hotel user behavior inference module includes a data perception unit and an activity prediction unit; the data perception unit is used to collect user device application data of at least one user staying in a guest room in a target hotel in real time; the user device operation data includes environmental perception data and device application switch status data; the corresponding public user behavior route data and hotel preference data are obtained; the user behavior route data includes n-th behavior data, n=1, 2, ..., N, n represents the n-th behavior event recorded by the user in public places in the hotel; the activity prediction unit is used to perform activity state inference on the current corresponding n-th behavior data to obtain user external activity prediction data; and the departure timestamp T of the n-th behavior data is obtained. n ; Get the time stamp of leaving the room T n-1 Time stamp of departure T n The user device operation data between them is combined with the inference network to analyze user behavior and obtain the user's internal activity prediction data; The intelligent application management module includes a chip control unit; the chip control unit is used to align the user's internal activity prediction data and the user's external activity prediction data in time and space, and identify the room leaving timestamp T n Corresponding behavioral intention labels; based on the identified behavioral intention labels, match the real-time device application update index of the guest rooms; based on the real-time device application update index, intelligently adjust the hotel device applications in the guest rooms and the hotel device applications outside the guest rooms to achieve intelligent management of the device applications in the target hotel.

[0012] The present invention has the following advantages: 1. The present invention collects user device application data in guest rooms and user behavior flow data in hotel public spaces, and integrates user historical preference information to achieve global modeling and state perception of user behavior. The integrated perception of cross-spatial, multi-dimensional data breaks through the limitations of traditional hotel intelligent systems that are limited to single rooms or preset scene rules, and improves the system's coverage and recognition accuracy of users' real behaviors. By aligning and mapping different user behavior data inside and outside the hotel space in time and space, and generating user comprehensive contextual behavior vectors with the help of a behavioral semantic embedding model, the system can deeply understand the complex behavior states exhibited by users at specific moments, and infer accurate behavioral intention labels based on this. Compared with methods based on single actions or static rules, this system is more intelligent and generalizable.

[0013] 2. The present invention matches the target device set related to the intention through the identified behavioral intention label, and calculates the real-time update index of each device in combination with the device application control factor library and the user's current device operating status, thereby realizing quantitative management of device response priority. This mechanism supports refined sorting and hierarchical response between devices, ensuring that high-energy consumption devices, user-preferred devices or environmentally sensitive devices can be regulated first, thereby improving resource utilization efficiency and service personalization. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 This is a structural diagram of an intelligent application management system for a hotel adopted in an embodiment of the present invention. DETAILED DESCRIPTION

[0015] In order to enable those skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0016] Example 1, a smart application management method for a hotel, comprising the following steps: Customer traffic flow prediction; context-aware services; device data and operating status data; lost item management; air monitoring (micro-scene control); obstacle and barrier-free areas; determining whether cleaning is necessary (for customers who stay for at least one day); and considering the device self-cleaning cycle using multiple sensing methods (without privacy concerns). For at least one user staying in a guest room at a target hotel, real-time user device application data is collected from the user's room. User device operation data includes environmental perception data and device application switch status data. The corresponding publicly available user behavior route data and hotel preference data are obtained. The user behavior route data includes n-th behavior data, where n = 1, 2, ..., N, where n represents the n-th behavior event recorded by the user in public places in the hotel. It should be noted that user device application data refers to the recordable and analyzable data set generated by users interacting with various smart devices in hotel rooms. This data primarily includes two categories: first, environmental perception data, which refers to physical environmental information automatically collected by sensors deployed in the guest rooms, including temperature, humidity, light intensity, noise decibel level, air quality index (such as PM2.5 and carbon dioxide concentration), and presence detection information (such as whether someone is in the room). Second, device application switch status data, which refers to user control operations on guest room devices via remote controls, panels, voice commands, or mobile applications, such as the air conditioner's on / off status, set temperature, and operating mode; the lighting on / off, brightness level, and color temperature selection; the curtain opening / closing degree; the TV's on / off, channel, and volume; and the current usage status of other devices such as air purifiers, aromatherapy diffusers, and music players. This data is collected in real time via standardized communication protocols and forms a data record stream with device identification, timestamp, operation type, and parameter values ​​as basic fields. It reflects the user's specific activities and usage preferences in the room and serves as an important basis for behavior recognition and scene-based intelligent control. At the same time, in the target hotel, all data is transmitted by integrated chips to achieve device interconnection.

[0017] Public user behavior data refers to the activity trajectories and behavior records of users in the hotel's public areas. This data mainly comes from the perception systems deployed in the hotel, such as video analysis systems, facial recognition cameras, infrared motion detectors, Bluetooth positioning devices, wireless network probes or card access control systems. These systems can identify the appearance time, stop location, movement path and behavior events of users in different open areas of the hotel (such as the lobby, elevator hall, gym, restaurant, conference room, etc.), and record them as a continuous behavior sequence; each behavior data includes the occurrence time of the behavior event, event type (for example, "entering the elevator", "entering the restaurant", "leaving the gym"), location code (floor, area number), duration and possible additional information (such as user identity confirmation method, behavior frequency, etc.). This data can be combined into a complete behavior trajectory according to user identity and time sequence, which can be used to determine the user's current state and infer the purpose of their behavior or the guest room scene they are about to enter.

[0018] The nth behavioral data is used to represent the nth time-ordered behavioral event of a user identified in the public areas of the hotel. The role of n is to establish a clear time series index relationship for user behavioral events, which is used for behavioral path analysis, context modeling, and behavioral intention reasoning. A behavioral event refers to an independent behavioral unit generated by a user in the public areas of the hotel that has observable attributes, identifiable boundaries, and is closely related to spatial location and time. A behavioral event can be entering or leaving an area, performing a certain functional activity, or interacting with a certain system. Common behavioral event types include but are not limited to: entering the elevator, arriving at the restaurant and staying, entering the gym through the access control, staying in the public sofa area, queuing for check-in in the lobby, and swiping a card to leave the room.

[0019] Hotel preference data refers to a collection of information related to a specific user's behavior patterns, usage habits, and personalized settings within the hotel. It comes from historical records formed during the user's previous stays, proactively set service options, preference templates, or habitual behavior data obtained by the system through learning and summary. The content of preference data includes but is not limited to: the type of guest room equipment that the user prefers to use (such as the habit of using smart speakers or smart aromatherapy), the commonly used temperature and humidity ranges in the room, lighting brightness and color temperature settings, whether a quiet environment is preferred, whether the user is accustomed to opening curtains or playing background music at specific times in the morning and evening, etc. These preference data are usually stored with the user as the primary key. Each preference item contains fields such as behavior tags, device attributes, priority levels, and trigger conditions. In the subsequent system decision-making process, it serves as context input to participate in device intelligent control and service personalized configuration, thereby improving the consistency and comfort of the guest experience.

[0020] Perform activity state inference on the current corresponding nth behavior data to obtain the user's external activity prediction data; obtain the room leaving timestamp T of the nth behavior data n ; Get the time stamp of leaving the room T n-1 Time stamp of departure T n The user device operation data between them is combined with the inference network to analyze user behavior and obtain the user's internal activity prediction data; The specific steps for inferring the activity state for the current corresponding n-th behavior data include: Extracting continuous spatiotemporal features based on the nth behavior data to obtain a user spatiotemporal feature set; wherein the user spatiotemporal feature set includes the behavior path, duration, and corresponding location coordinates of public places in the hotel; The user spatiotemporal feature set is derived by performing structured parsing of the current n-th behavior data and extracting continuous location-based behavior events within adjacent time periods. This extraction process first identifies the user's activity path within the hotel's public areas, identifying the spatial nodes, dwell time, movement speed, and route pattern from the previous event location to the current event location. Semantic labels are then assigned based on the type of behavioral event (e.g., dining, fitness, or meeting). The spatial coordinates of the event, the functional label of the corresponding hotel area (e.g., public rest area or service channel), the duration of the event, the time of occurrence, and the triggering method (e.g., whether it was accompanied by access control, location tracking, or code scanning) are then extracted. This information is then combined into a vector-based spatiotemporal feature set for the current moment. The data structure typically includes a timestamp sequence, a spatial location index, a behavior category code, a duration value, and path topology information. This data set reflects the user's behavior status and spatial location relationship within the hotel space at that moment, providing rich spatial context and short-term dynamic features for subsequent model analysis. This is particularly helpful in determining whether the current behavior represents a continuous path, a habitual behavior, or an unusual dwelling.

[0021] Obtain the 1st to n-1th behavior data to construct behavior sequence context features; take the user's spatiotemporal feature set, behavior sequence context features, and hotel preference data as input, input them into the pre-trained behavior sequence learning model for analysis, and obtain the user's external activity prediction data; The behavioral sequence context features of the 1st to n-1th behavioral data are constructed by serial encoding, time alignment and semantic abstraction of historical behavioral events; the first n-1 behavioral events of the user in the hotel are extracted in chronological order, ensuring that each event contains complete fields such as timestamp, location coordinates, behavior category label, and length of stay; these events are embedded in the time series in sequence, and the unequal interval events are aligned through the time normalization method; the system converts each event into a fixed-length behavior vector through a behavioral embedding algorithm (such as location embedding, event type encoding, behavioral semantic graph embedding, etc.), and then uses a sliding window mechanism or recursive structure (such as recurrent neural network) to establish evolutionary associations and contextual dependencies between events, ultimately forming a behavioral sequence context feature that reflects the user's short-term and medium-term behavioral tendencies.

[0022] User external activity prediction data is derived from the system's analysis of user behavioral event data, spatiotemporal characteristics, and historical behavioral context in public hotel areas, inferring a user's current or upcoming behavior status, behavioral purpose, or activity intention. It provides a dynamic, high-level semantic interpretation and prediction of user behavior outside of the guest room, represented by behavioral labels (e.g., dining out, short-term absences, attending meetings, working out) and corresponding confidence levels or probability distributions. The pre-trained behavioral sequence learning model is a time series modeling network based on deep learning, and is often trained using long short-term memory networks, gated recurrent units, transformer structures, or graph neural networks. The training process mainly relies on a large amount of historical user behavior route data of the hotel. Through supervised learning, historical behavioral event sequences and user preference data are input, and the output label is the user's actual external activity intention or behavioral status (such as short-term room vacancy, going to the gym, dining out, accompanying visitors, etc.). During the construction of the training dataset, each behavioral sequence needs to be manually or semi-automatically labeled with its final behavioral intention, and the corresponding time period, spatial path, and preference label are constructed. During training, the model learns how to predict the current behavioral state based on the characteristics of the previous behavior through loss function optimization, and generalizes different behavioral patterns. The trained model can capture the semantic evolution of behavior, spatial preference migration, and the influence of preference habits, thereby having the ability to predict and summarize when facing new users or new sequences.

[0023] Get the check-out timestamp T n-1 Time stamp of departure T n The specific steps for combining the user device operation data between the two devices with the inference network to perform user behavior analysis include: Identify user intentions for environmental changes based on environmental perception data; identify user intentions for actions based on device application switch status data; construct scenario features based on user intentions for environmental changes and actions to obtain an internal behavior feature vector sequence; User environmental change intention refers to the implicit demand or potential intention of the user to adjust the room environment, inferred by the system through analyzing the changing trends of the environmental perception data in the guest room. For example, when the system detects that the room temperature continues to rise and the humidity drops, and the user does not directly operate the air conditioner, the system may infer that the user has an environmental change intention to adjust the temperature. This intention is not expressed by the user through explicit control, but is the user's implicit demand reflected by the passive fluctuation of environmental variables. The possible target it points to is often identified by analyzing the deviation between the change rate, fluctuation amplitude, and duration of environmental data and the established comfort range, such as the need to cool down, dim the lights, or desire ventilation. This type of intention reflects the user's perception and expected adjustment of the environmental state and is an important basis for understanding the behavioral context.

[0024] User action intention refers to the explicit behavioral purpose expressed by the system through analysis of the user's active control behavior of guest room equipment. For example, when a user turns off the TV, air conditioner, and curtains in a short period of time, while simultaneously triggering the door lock, the system can identify the user's intention to leave the room or prepare to go out. Unlike environmental change intention, human action intention reflects the user's direct intention expressed through actual operation of the equipment. These operations include switch control, mode switching, parameter adjustment, etc., and are often based on equipment switch status data, combined with the timing of the behavioral operation, the linkage sequence between devices, and consistency with past behavior to make intention judgments. Identifying such intentions helps to clarify the actual goals of the user's operating behavior and improve the system's ability to respond immediately to changes in user status.

[0025] Situational feature construction involves fusing the user's device usage behavior and environmental change data during the pre-departure period to generate a set of high-level behavioral features that reflect the user's actual behavioral context. Specifically, the system extracts key feature variables from environmental change intentions and human action intentions, such as temperature change rate, illumination change amplitude, device operation frequency, device combination usage pattern, and operation sequence stability. These variables are organized into temporal feature sequences through time window sliding, event density analysis, and semantic label encoding. The system uses temporal modeling structures (such as recursive networks or temporal encoders) to integrate them into a structured sequence of internal behavioral feature vectors. Each vector in this sequence represents the behavioral state within a certain time slice, and its content includes information such as user operation intensity, response frequency, and environmental deviation. It is used to represent the continuous behavioral trajectory within the entire departure window. This feature sequence serves as the basic data for modeling the user's internal room behavior and is an important input for the inference model for pattern recognition and intent inference.

[0026] Obtain historical stay data of users staying in guest rooms at the target hotel; establish comprehensive stay context features based on historical stay data and hotel preference data; Comprehensive in-hotel contextual features refer to the global semantic representation of the user's personalized behavioral background and environmental adaptation characteristics, which is constructed by the system based on the user's historical check-in behavior data and preference configuration in the current hotel. The system will retrieve the user's previous check-in records in the hotel or similar hotels, including information such as stay duration, stay frequency, common equipment combinations, typical operation modes, equipment adjustment habits, etc., and combine the preference templates set by the user before check-in (such as the most comfortable air-conditioning temperature, whether to automatically draw the curtains, whether to use silent mode, etc.) to extract the user's high-frequency behavior patterns in the room usage scenario; through multi-dimensional feature fusion (such as frequency vectorization, behavior heat map encoding, etc.) and label embedding processing, a unified user in-hotel portrait is formed. This portrait is the comprehensive in-hotel contextual features. As a carrier for the system to understand the user's long-term behavioral preferences and environmental expectations, it helps to improve the reasoning network's ability to judge the consistency between current behavior and historical habits.

[0027] The comprehensive living situation characteristics and internal behavior feature vector sequence are input into the inference network to analyze the room leaving timestamp T n-1 Time stamp of departure T n The device usage patterns, environmental change trends and human activity patterns within the system are combined to obtain user internal activity prediction data. The inference network uses the Bayesian inference network as the basic network structure to analyze user behavior. The user's internal activity prediction data is the intelligent recognition result of the typical activity status of the user in the guest room output by the system after analyzing the environmental changes, human operation behaviors, historical check-in habits and preference factors during the time period of leaving the room; the device usage pattern is the first type of behavioral feature identified by the inference network, which mainly reflects the user's use combination, frequency, change trend and operation rhythm of the guest room smart devices in the time period before leaving the room; for example, whether the air conditioner is turned on, whether the temperature setting is adjusted, whether the TV is turned on and off repeatedly, whether the lighting is gradually dimmed or turned off, etc. These operation combinations can reflect whether the user is preparing to leave, whether he is resting or doing office activities; through the collaborative control characteristics between devices and the rhythm of the operation sequence, the system can infer whether the user is in active use or in automatic energy-saving response; the environmental change trend is the system's analysis result of the change process of the environmental perception data in the room, reflecting the user's indirect intention to adjust the environment without directly controlling the device. Expression; for example, whether the light intensity gradually decreases, whether the indoor temperature remains within a constant range, whether the noise level shows a downward trend, etc.; the environmental change trend is usually combined with the device response results and the user's historical preferences for comparative analysis to determine whether the current environment has tended to the user's expected state, and whether the environment has stabilized or declined due to the user's impending departure; human activity patterns refer to the purposeful behavior combinations and behavior logic expressed by users through device switch operations, mobile triggers, access control behaviors, etc. Human activity patterns not only consider whether the operations occur, but also focus on the combination sequence, time interval, repetitiveness and typicality of the operations; for example, if a user continuously turns off the air conditioner, turns off the lights, opens the door lock and triggers the door sensor within 10 minutes, it may constitute a typical human activity pattern of preparing to leave the room; and if the user repeatedly adjusts the air conditioner and curtains during the noon period, it may indicate that the user is in the office or in a residential state. These patterns are modeled through probabilistic reasoning and constitute the main clues for identifying the user's internal activity state.

[0028] The process of user behavior analysis by reasoning network is based on the structure of Bayesian reasoning network, which integrates the comprehensive living situation characteristics of users and their T n-1 to T nThe sequence of internal behavioral feature vectors formed within a time period is used to establish conditional probability dependencies between each feature, thereby systematically inferring the user's behavioral status within this time period. The inference network is constructed by using historical user behavior data as training samples, combining artificial expert rules and data statistical analysis results to establish the initial network topology, and then using parameter learning and structure learning algorithms to optimize the dependencies between nodes. During the construction process, the core variable nodes in the network are first determined, including user behavior labels, device operation characteristics, environmental change indicators, and preference constraints. Then, through data analysis, the statistical correlation and causal path between variables are identified to form an initial directed acyclic graph. On this basis, the maximum expectation algorithm or Bayesian scoring function is used to iteratively optimize the network structure to ensure the validity of the conditional independence assumption and the balance between the model's generalization ability. The conditional probability distribution under each node is parameter-fitted through maximum likelihood estimation or Bayesian estimation to ensure that the model can effectively infer the target behavior label based on the status of some variables when actually used.

[0029] Align the user's internal activity prediction data and the user's external activity prediction data in time and space, and identify the room departure timestamp T n Corresponding behavioral intention labels; based on the identified behavioral intention labels, matching the real-time device application update index of the guest room; based on the real-time device application update index, intelligently adjusting the hotel device applications in the guest room and the hotel device applications outside the guest room to achieve intelligent management of the device applications in the target hotel; Identify the departure timestamp T n The specific steps for corresponding behavioral intention labels include: The user's internal activity prediction data and external activity prediction data are aligned in time and space to obtain a user comprehensive context behavior vector. Behavioral intention matching is performed based on the user comprehensive context behavior vector to obtain a behavioral intention label. The user comprehensive context behavior vector generated in the previous stage is first used as input. This vector combines the user's internal activity characteristics and external behavior characteristics before and after leaving the room and has completed semantic integration through a spatiotemporal alignment graph structure. This vector has high-dimensional semantic expression capabilities and can reflect the user's complete behavioral state in the current context. The system then extracts a set of modeled behavioral intention label vector templates from the built-in behavioral intention label prototype library. Each label vector is obtained based on training of a large amount of historical user data or expert semantic modeling, representing a specific behavioral intention. Each behavioral intention label contains a standardized behavioral feature distribution, typical device operation mode, preference consistency weight, and behavioral background feature range.

[0030] Using a behavioral semantic similarity calculation method, a semantic similarity measurement method (such as cosine similarity or Euclidean distance) is used to perform a one-to-one comparison between the user's current behavior vector and each label vector, calculating their similarity values ​​in the semantic space and forming a similarity score list. Based on the similarity sorting results, the system selects the label with the highest score that exceeds the confidence threshold as the primary behavior intention label at the current moment, and can use secondary labels with higher scores as auxiliary behavior references. The specific steps of spatiotemporal alignment are: extracting all behavioral events from the user's internal activity prediction data and the user's external activity prediction data; mapping the user's internal activity prediction data and the user's external activity prediction data to the user's behavior prediction coordinate system, and annotating the spatial position coordinates and absolute timestamps of the behavioral events according to the granularity of the behavioral events; Extracting behavioral events is to identify the smallest event unit with independent behavioral semantics by analyzing the structured behavioral state information contained in the predicted data; the system first deconstructs the predicted data, and parses the fields such as the labels representing user activities, the corresponding time intervals, spatial locations, device operation characteristics, and environmental response states into standardized behavioral description units; each activity with an independent behavioral intention expression, such as turning off the air conditioner, entering the elevator, dimming the lights, and leaving the gym, is regarded as a behavioral event. During the extraction process, the system will use the event trigger time as the main anchor point and arrange all events in chronological order, while eliminating time overlaps or behavioral conflicts to ensure that each event has clear boundaries in time and semantics.

[0031] The method for mapping user internal activity prediction data and user external activity prediction data into the user behavior prediction coordinate system is to establish a two-dimensional or three-dimensional coordinate system structure composed of time and space as the main axes, unifying the expression and aligned analysis of the two types of predicted behavior events. The user behavior prediction coordinate system is a structured data mapping space. The time dimension is usually represented by a standard timestamp, and the spatial dimension is represented by a predefined spatial location code in the hotel scenario. For example, each floor, each functional area, and each guest room has a unique spatial number. At a finer granularity, it can also be annotated as a two-dimensional coordinate. During the mapping process, the system locates each behavior event on the time axis based on the time information it carries, and locates it in the spatial dimension based on the location information of the behavior. At the same time, the specific content of the event is annotated in the coordinate system, including the behavior type, event confidence, triggering method, etc., to complete the projection of the behavior event into the coordinate system. Spatial locations can be converted into position coordinates using a spatial label mapping table, and timestamps can be directly normalized to form a unified alignment basis.

[0032] Define dynamic time windows based on the granularity of behavioral events; establish several event correspondences for user internal activity prediction data and user external activity prediction data based on the dynamic time windows; set a behavioral response tolerance interval for each event correspondence; Defining dynamic time windows at the granularity of behavioral events means that the system automatically adjusts the allowed matching time range for each type of event based on the duration, continuity, and behavioral rhythm of the behavioral event itself, thereby establishing a more adaptable behavioral correspondence; the dynamic time window represents the allowable alignment interval occupied by each behavioral event on the time axis, and is a time period of adjustable length used to determine whether a temporal association can be established between two events; based on the semantic attributes and historical behavioral characteristics of the event, the system uses a sliding window mechanism or a rhythm estimation algorithm to define the initial window range for each type of behavior, and automatically adjusts the window width through model learning.

[0033] In the dynamic window, the system selects events with high temporal overlap and similar behavioral semantics from internal and external events to establish event correspondences, that is, to identify whether two events express the same continuous behavior or causally related behavior. Each event correspondence further defines a behavioral response tolerance interval. This tolerance interval refers to the maximum allowable temporal offset range between two behavioral events. It can not only handle the delay in the perception system, but also adapt to the asynchrony in user behavior. For example, if a user turns off all lights within 5 minutes after leaving the room, it is still considered part of the leaving behavior. The role of the response tolerance interval is to improve the system's tolerance for the uncertainty of real-world behavior timing, prevent misjudgment of unimportant time differences, and improve the robustness and fault tolerance of behavior understanding.

[0034] A spatiotemporally aligned graph structure is constructed with each behavioral event as a node and the event correspondence as an edge. The aligned graph structure is vectorized using a behavioral semantic embedding model to obtain a comprehensive user contextual behavior vector. In graph construction, the system takes each internal or external behavioral event that has been aligned in time and space as a node, and connects the events that are judged to be related through edges to form a graph structure; each edge records the attributes such as the time and space distance, semantic similarity, and response intensity of two events, forming a multi-attribute heterogeneous graph; then, the graph is sent to the behavioral semantic embedding model for learning and vectorization processing. The model can be a graph convolutional neural network or a structured transformation model, which is used to encode the node and its context structure into a fixed-dimensional vector representation; the model compresses the entire graph into a comprehensive contextual behavior vector by capturing the causal relationship, spatial adjacency, temporal sequence logic and semantic coupling characteristics between events. This vector is an overall abstract expression of user behavior in the current time period, which is used for behavioral intention recognition and control strategy matching.

[0035] It should be noted that in all data processing and data analysis processes, when the input data structure is inconsistent, the system uses the data standardization module to perform feature alignment and missing value completion to ensure that all behavioral event data can enter the analysis process in a unified manner, and map the data formats generated from different sources (such as devices, sensors, and behavior recognition modules) into an internal unified behavioral event data structure; if the data granularity is different, for example, some external behavioral events are composite behaviors, the system will decompose the events and remodel and label them according to the standard behavioral granularity; through the above-mentioned standardization and repair mechanism, the system can ensure that all behavioral data can be accurately located in the coordinate system and participate in the unified graph modeling and semantic reasoning process, ensuring that behavioral intention recognition is still consistent and highly accurate under multi-source heterogeneous data conditions.

[0036] The specific steps for matching the real-time device application update index of guest rooms based on the identified behavioral intent tags include: Constructing a device application control factor library containing multiple hotel device applications; the device application control factor library contains several types of device application control factors; wherein the device application control factors include but are not limited to device type identification, user usage frequency, current device operating status, environmental response priority, energy consumption level, and device scenario category label; Based on the behavior intention tag matching associated target device application set, the corresponding device application control factor in the device application control factor library is matched for each target device application in the target device application set as the behavior association weight coefficient; the device application operating parameters currently in the guest room are obtained; and the real-time device application update index is calculated by combining the device application operating parameters and the behavior association weight coefficient. The device application control factor library is constructed by categorizing and modeling all intelligently controllable device resources within the hotel. The application characteristics of each device category are structured based on multiple factors, including functional attributes, user behavior relevance, frequency of use, and energy efficiency. During the system initialization phase, the hotel management system imports a list of all devices. Based on the device's functional category, service scenario, historical user interaction data, and energy consumption data, the system extracts key attributes that influence device control decisions and establishes a standardized set of control factors. Each device application object forms a device control factor record in the library. The record consists of multiple fields, including device type identification (such as lighting, air conditioning, and television), user usage frequency (number of times and duration of use during the current or historical stay), current operating status of the device (on, off, medium, and adjustment), environmental response priority (referring to the weight of the device's contribution to environmental comfort or user satisfaction), energy consumption level (high, medium, and low corresponding to unit time consumption), and the scenario category label to which the device belongs (such as rest scene, room departure scene, and welcome scene). The control factor library adjusts the weight distribution of each factor in real time based on user feedback and usage behavior through a regular update mechanism to ensure a high degree of consistency between device response and user needs.

[0037] The device type identifier in the device application control factor is used to identify the device's functional role in the control logic, facilitating matching with the target behavioral scenario in the behavioral intention label. User usage frequency indicates how often the device has been activated in the current user's historical behavior, which is used to determine the user's degree of dependence on the device and their habitual usage tendency. The device's current operating status reflects the necessity of current control; if the device is already in an ideal state, the response priority can be lowered. The environmental response priority assesses whether the device directly affects comfort or behavioral scenarios, such as air conditioning's contribution to temperature regulation being higher than background music. The energy consumption level measures the impact of device response on energy consumption and can be used for intelligent balancing in energy-saving scenarios. The scenario category label to which the device belongs is used to bind the device to a specific behavioral intention type, such as giving priority to controlling lighting and curtains under the "Ready to Rest" label, rather than television and aromatherapy. These factors constitute the core input of the decision model and provide a quantitative basis for the calculation of the update index.

[0038] To match the associated target device application set, the behavioral intention label is used as the query condition, and the corresponding device list is searched from the preset behavior-device mapping relationship model; this mapping relationship model is built based on historical behavior data mining, user operation path analysis and expert experience knowledge, and maps common behavioral intentions with typical device groups; for example, preparing to leave the room is usually associated with air conditioning, lighting, curtains and door locks, while returning to the guest room prioritizes matching lighting, background music and air conditioning temperature settings; when the user's behavioral intention label is identified, the system extracts the corresponding device set in the relationship model as the response candidate target, and further dynamically filters it according to the user's personalized preferences, room configuration differences and current occupancy status to obtain the target device set that can actually respond in the guest room.

[0039] The real-time device application update index is calculated by extracting the corresponding control factor from the device application control factor library for each device after matching the target device application set associated with the current behavioral intention label. The system then performs a comprehensive calculation based on the device's current operating parameters in the guest room to derive a response priority score for each device in the current behavioral context. Specifically, the system first converts the control factor of each device into a standardized scoring vector, which mainly includes the user usage frequency score, operating status deviation value, environmental response priority weight, energy consumption level adjustment coefficient, and scenario category matching degree. At the same time, the system obtains the current parameter value from the device operating status, such as the switch status, adjustment intensity, operating mode, etc., and weightedly calculates the degree of deviation between it and the state required by the target behavioral intention. After the calculation is completed, the update index of all devices is summarized and sorted. The higher the value, the more the device needs to adjust or respond in the current behavioral scenario. This calculation method takes into account the user's behavioral intention, the actual status of the device, and the energy efficiency strategy, and is the core mechanism for realizing intelligent linkage of guest room equipment under behavior-driven control.

[0040] Intelligently adjust hotel device applications inside and outside guest rooms based on the real-time device application update index to achieve intelligent management of device applications within the target hotel; The specific adjustment method is as follows: For example, in an actual application scenario, a user checks into a smart hotel and is identified by the system as having a short-term vacancy at 12:30 p.m. The system first extracts the user's internal and external activity prediction data for the current time period, performs spatiotemporal alignment based on the granularity of behavioral events, constructs a spatiotemporal alignment graph structure, and generates a user comprehensive contextual behavior vector using a behavioral semantic embedding model. The system then matches this vector with multiple tags in the behavioral intention tag library and determines that the current behavioral intention is a short-term vacancy. Based on the identified behavioral intention label, the system obtains the target device application set associated with the short-time room absence label in the preset behavior-device mapping rules, covering device types such as air conditioners, lighting, curtains, televisions, air purifiers and door locks; the system extracts the corresponding device application control factor for each target device from the device application control factor library, including information such as device type identification, user usage frequency, current operating status, environmental response priority, energy consumption level and the scenario category label to which the device belongs; combined with the current operating parameters of each device, the system calculates the update index of each device and concludes that the real-time device application update index values ​​of devices such as air conditioners, televisions, and curtains are relatively high, indicating that they need to be adjusted first under the current behavioral intention.

[0041] Based on the update index ranking results, the system automatically adjusts the target device applications in sequence. Specifically, this includes: setting the air conditioner to energy-saving mode and raising the set temperature; turning off the TV and all main lighting fixtures, leaving only the entrance sensor light; automatically closing the curtains to their fully closed state; and maintaining the air purifier's original operating state, as the user has a recorded air quality preference. After the user leaves the room, the system automatically locks the door lock controller three seconds after detecting the door magnetic status changes to closed, and pushes room status feedback to the user. Simultaneously, the system determines that the user is heading to the gym area and automatically activates the corridor lighting and gym welcome interface, providing the user with a coherent, low-perception intelligent experience. Through these operations, based on user behavior intention tags, combined with the device application control factor library and device operating status data, a real-time device application update index is calculated. Based on this, hotel device applications both inside and outside the guest rooms are intelligently adjusted, achieving full-process intelligent management of device applications within the target hotel, improving energy efficiency and user service experience.

[0042] Example 2, an intelligent application management system for a hotel, see Figure 1 As shown, including: The hotel user behavior inference module includes a data perception unit and an activity prediction unit; the data perception unit is used to collect user device application data of at least one user staying in a guest room in a target hotel in real time; the user device operation data includes environmental perception data and device application switch status data; the corresponding public user behavior route data and hotel preference data are obtained; the user behavior route data includes n-th behavior data, n=1, 2, ..., N, n represents the n-th behavior event recorded by the user in public places in the hotel; the activity prediction unit is used to perform activity state inference on the current corresponding n-th behavior data to obtain user external activity prediction data; and the departure timestamp T of the n-th behavior data is obtained. n ; Get the time stamp of leaving the room T n-1 Time stamp of departure T nThe user device operation data between them is combined with the inference network to analyze user behavior and obtain the user's internal activity prediction data; The intelligent application management module includes a chip control unit; the chip control unit is used to align the user's internal activity prediction data and the user's external activity prediction data in time and space, and identify the room leaving timestamp T n Corresponding behavioral intention labels; based on the identified behavioral intention labels, match the real-time device application update index of the guest rooms; based on the real-time device application update index, intelligently adjust the hotel device applications in the guest rooms and the hotel device applications outside the guest rooms to achieve intelligent management of the device applications in the target hotel.

[0043] It should be understood that those skilled in the art may make improvements or modifications based on the above description, and all such improvements and modifications shall fall within the scope of protection of the appended claims. Any portion of this specification not described in detail is prior art known to those skilled in the art.

Claims

1. A smart application management method for a hotel, characterized in that: The following steps are involved: For at least one user staying in a guest room of the target hotel, real-time collection of user device application data of the user staying in the hotel room; User device operation data includes environmental perception data and device application switch status data; obtain corresponding public user behavior route data and hotel preference data; user behavior route data includes n-th behavior data, n = 1, 2, ..., N, n represents the n-th behavior event recorded by the user in public places in the hotel; Perform activity state inference on the current corresponding n-th behavior data to obtain the user's external activity prediction data; Get the departure timestamp T of the nth behavior data n ; Get the time stamp of leaving the room T n-1 Time stamp of departure T n The user device operation data between them is combined with the inference network to analyze user behavior and obtain the user's internal activity prediction data; Align the user's internal activity prediction data and the user's external activity prediction data in time and space, and identify the room departure timestamp T n Corresponding behavioral intention labels; based on the identified behavioral intention labels, match the real-time device application update index of the guest rooms; based on the real-time device application update index, intelligently adjust the hotel device applications in the guest rooms and the hotel device applications outside the guest rooms to achieve intelligent management of the device applications in the target hotel.

2. A smart application management method for a hotel according to claim 1, characterized in that: The specific steps for inferring the activity state for the current corresponding n-th behavior data include: Extracting continuous spatiotemporal features based on the nth behavior data to obtain a user spatiotemporal feature set; wherein the user spatiotemporal feature set includes the behavior path, duration, and corresponding location coordinates of public places in the hotel; Obtain the 1st to n-1th behavior data to construct the behavior sequence context features; take the user's spatiotemporal feature set, behavior sequence context features and hotel preference data as input, input them into the pre-trained behavior sequence learning model for analysis, and obtain the user's external activity prediction data.

3. A smart application management method for a hotel according to claim 2, characterized in that: Get the check-out timestamp T n-1 Time stamp of departure T n The specific steps for combining the user device operation data between the two devices with the inference network to perform user behavior analysis include: Identify user intentions for environmental changes based on environmental perception data; identify user intentions for actions based on device application switch status data; construct scenario features based on user intentions for environmental changes and actions to obtain an internal behavior feature vector sequence; Obtain historical stay data of users staying in guest rooms at the target hotel; establish comprehensive stay context features based on historical stay data and hotel preference data; The comprehensive living situation characteristics and internal behavior feature vector sequence are input into the inference network to analyze the room leaving timestamp T n-1 Time stamp of departure T n The device usage pattern, environmental change trend and human activity pattern within the system are combined to obtain the user's internal activity prediction data.

4. A smart application management method for a hotel according to claim 3, characterized in that: Identify the departure timestamp T n The specific steps for corresponding behavioral intention labels include: Align the user's internal activity prediction data and the user's external activity prediction data in time and space to obtain the user's comprehensive context behavior vector; match the user's behavior intention based on the comprehensive context behavior vector to obtain the behavior intention label; The specific steps of spatiotemporal alignment are: extracting all behavioral events from the user's internal activity prediction data and the user's external activity prediction data; mapping the user's internal activity prediction data and the user's external activity prediction data to the user's behavior prediction coordinate system, and annotating the spatial position coordinates and absolute timestamps of the behavioral events according to the granularity of the behavioral events; Define dynamic time windows based on the granularity of behavioral events; establish several event correspondences for user internal activity prediction data and user external activity prediction data based on the dynamic time windows; set a behavioral response tolerance interval for each event correspondence; A spatiotemporally aligned graph structure is constructed with each behavioral event as a node and the event correspondence as an edge. The aligned graph structure is vectorized using a behavioral semantic embedding model to obtain a comprehensive user contextual behavior vector.

5. A smart application management method for a hotel according to claim 4, characterized in that: The specific steps for matching the real-time device application update index of guest rooms based on the identified behavioral intent tags include: Constructing a device application control factor library containing multiple hotel device applications; the device application control factor library contains several types of device application control factors; Based on the target device application set associated with the behavior intention label matching, the corresponding device application control factor is matched in the device application control factor library for each target device application in the target device application set as the behavior association weight coefficient; the device application operating parameters of the current guest room are obtained; and the real-time device application update index is calculated by combining the device application operating parameters and the behavior association weight coefficient.

6. A smart application management method for a hotel according to claim 5, characterized in that: The inference network uses the Bayesian inference network as the basic network structure to analyze user behavior.

7. An intelligent application management system for a hotel, characterized in that: The system applies the intelligent application management method for a hotel according to any one of claims 1 to 6, including: The hotel user behavior inference module includes a data perception unit and an activity prediction unit; the data perception unit is used to collect user device application data of at least one user staying in a guest room in a target hotel in real time; the user device operation data includes environmental perception data and device application switch status data; the corresponding public user behavior route data and hotel preference data are obtained; the user behavior route data includes n-th behavior data, n=1, 2, ..., N, n represents the n-th behavior event recorded by the user in public places in the hotel; the activity prediction unit is used to perform activity state inference on the current corresponding n-th behavior data to obtain user external activity prediction data; and the departure timestamp T of the n-th behavior data is obtained. n ; Get the time stamp of leaving the room T n-1 Time stamp of departure T n The user device operation data between them is combined with the inference network to analyze user behavior and obtain the user's internal activity prediction data; The intelligent application management module includes a chip control unit; the chip control unit is used to align the user's internal activity prediction data and the user's external activity prediction data in time and space, and identify the room leaving timestamp T n Corresponding behavioral intention labels; based on the identified behavioral intention labels, match the real-time device application update index of the guest rooms; based on the real-time device application update index, intelligently adjust the hotel device applications in the guest rooms and the hotel device applications outside the guest rooms to achieve intelligent management of the device applications in the target hotel.

Images