A marketing strategy generation method, electronic device, storage medium and program product

Abstract

The present disclosure provides a marketing strategy generation method, an electronic device, a storage medium and a program product, relates to the field of aviation, and can improve the accuracy of passenger group division, thereby meeting the demand of airlines for passenger group fine operation. The method comprises the following steps: obtaining passenger characteristic data sets of a plurality of passengers, wherein the passenger characteristic data sets comprise identity data, behavior characteristic data and environment characteristic data of each passenger. Based on the passenger characteristic data sets, a passenger comprehensive behavior graph is constructed, the passenger comprehensive behavior graph is used to reflect the association relationship between each passenger and other passengers, the passenger comprehensive behavior graph is input into a passenger segmentation model, and a division result of each passenger is obtained, the division result comprises a high-frequency business trip group, a cross-border tourism family or a price-sensitive individual passenger. Based on the division result of each passenger, a corresponding marketing strategy of each passenger is determined.

Description

Technical Field

[0001] This disclosure relates to the aviation field, and more particularly to a marketing strategy generation method, electronic device, storage medium, and program product. Background Technology

[0002] With intensifying competition in the civil aviation market and increasingly diversified passenger demands, precise passenger segmentation has become a core foundation for airlines to improve operational efficiency, optimize revenue management, and conduct personalized services and targeted marketing. Scientifically segmenting passenger groups helps airlines identify high-value customer segments, understand travel patterns, and match differentiated services, thereby enhancing customer loyalty and market competitiveness.

[0003] However, the current accuracy of passenger group segmentation is low, making it difficult to meet airlines' needs for refined passenger group operations. Summary of the Invention

[0004] This disclosure provides a marketing strategy generation method, electronic device, storage medium, and program product that can improve the accuracy of passenger group segmentation, thereby meeting the needs of airlines for refined passenger group operations.

[0005] To achieve the above objectives, the present disclosure adopts the following technical solution: Firstly, this disclosure provides a marketing strategy generation method, comprising: acquiring a passenger feature dataset of multiple passengers, the passenger feature dataset including each passenger's identity data, behavioral feature data, and environmental feature data; constructing a comprehensive passenger behavior graph based on the passenger feature dataset, the comprehensive passenger behavior graph reflecting the relationships between each passenger and other passengers; inputting the comprehensive passenger behavior graph into a passenger segmentation model to obtain the segmentation results for each passenger, the segmentation results including high-frequency business travel groups, cross-border travel families, or price-sensitive individual travelers; and determining the corresponding marketing strategy for each passenger based on the segmentation results.

[0006] Based on the first aspect, in this embodiment, by constructing a comprehensive passenger behavior graph that integrates multiple features and topological structures, and considering both the node attributes of passengers and the relationships between passenger association edges, the shortcomings of traditional clustering methods that treat passengers as isolated individuals are avoided. This enables accurate identification of passenger groups with stable behavioral patterns, improving the reliability and authenticity of the segmentation results. Based on more reliable segmentation results, direct and reliable data support can be provided for airlines' core decisions such as differentiated pricing, precise promotions, and capacity allocation. This can meet the airlines' needs for refined operation of passenger groups, thereby improving customer relationship management efficiency and market competitiveness.

[0007] The passenger segmentation results are obtained by inputting the data into the passenger segmentation model. Secondly, embodiments of this disclosure provide an electronic device, which includes a processor, a communication interface, and a memory. The memory stores at least one computer program, which is loaded and executed by the processor to implement the marketing strategy generation method described above.

[0008] Thirdly, this disclosure provides a computer-readable storage medium storing at least one computer program, which is loaded and executed by a processor to implement the marketing strategy generation method described above.

[0009] Fourthly, this disclosure provides a computer program product, which includes a computer program or instructions that, when executed by a processor, implement the marketing strategy generation method described above.

[0010] It should be noted that any of the possible implementations of any of the above aspects can be combined, provided that the solutions do not contradict each other. Attached Figure Description

[0011] To more clearly illustrate the technical solutions in this disclosure, the accompanying drawings used in some embodiments of this disclosure will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this disclosure. For those skilled in the art, other drawings can be obtained based on these drawings.

[0012] Figure 1 This is a schematic diagram illustrating an application scenario of a marketing strategy generation method provided in an embodiment of this disclosure; Figure 2 A schematic diagram of the hardware structure of an electronic device provided in an embodiment of this disclosure; Figure 3 A flowchart illustrating a marketing strategy generation method provided in this embodiment of the disclosure; Figure 4 A flowchart illustrating another marketing strategy generation method provided in this embodiment of the disclosure; Figure 5 This is a schematic diagram of the structure of a comprehensive passenger behavior map provided in an embodiment of the present disclosure; Figure 6 A flowchart illustrating yet another marketing strategy generation method provided in this disclosure embodiment; Figure 7 This is a schematic diagram of the structure of a passenger segmentation model provided in an embodiment of the present disclosure; Figure 8 This is a schematic diagram of a marketing strategy generation device provided in an embodiment of the present disclosure. Detailed Implementation

[0013] The following describes in detail a marketing strategy generation method provided by an embodiment of this disclosure, with reference to the accompanying drawings.

[0014] In this article, the term "and / or" is merely a description of the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can represent three situations: A exists alone, A and B exist simultaneously, and B exists alone.

[0015] The terms “first” and “second” in this disclosure and its accompanying drawings are used to distinguish different objects or to distinguish different treatments of the same object, rather than to describe a particular order of objects.

[0016] Furthermore, the terms “comprising” and “having”, and any variations thereof, used in the description of this disclosure are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the steps or units listed, but may optionally include other steps or units not listed, or may optionally include other steps or units inherent to such process, method, product, or apparatus.

[0017] It should be noted that in this disclosure, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" or "for example" in this disclosure should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.

[0018] With intensifying competition in the civil aviation market and increasingly diversified passenger demands, precise passenger segmentation has become a core foundation for airlines to improve operational efficiency, optimize revenue management, and conduct personalized services and targeted marketing. Scientifically segmenting passenger groups helps airlines identify high-value customer segments, understand travel patterns, and match differentiated services, thereby enhancing customer loyalty and market competitiveness.

[0019] Existing passenger segmentation schemes mostly rely on traditional clustering algorithms or shallow graph learning models, which generally have obvious shortcomings: traditional methods only group passengers based on static attributes, ignoring the topological relationships of groups such as passenger social interaction, similar behavior, and mutual influence of decision-making, resulting in distorted segmentation results and failure to uncover stable customer group patterns; existing graph learning segmentation techniques are mostly based on comment sentiment networks to build models, which cannot adapt to the core association features of air travel scenarios, and lack a time-series dynamic update mechanism, resulting in poor timeliness of customer group identification.

[0020] Currently, passenger segmentation in the civil aviation industry involves airlines clustering passengers based on their different attributes, behaviors, needs, and values. A commonly used method in passenger segmentation research is the K-means clustering algorithm. However, this method only analyzes static individual attributes and cannot consider relationships between passengers. For example, it struggles to capture characteristics such as frequency of travel together, social network interactions, route overlap, service usage patterns, and the influence of word-of-mouth recommendations. This results in low accuracy in segmenting passenger groups and fails to meet airlines' needs for refined passenger group management.

[0021] Furthermore, existing methods for segmenting civil aviation passengers lack the ability to systematically map the segmentation results to specific business scenarios of airlines.

[0022] To address the aforementioned technical issues, this disclosure provides a marketing strategy generation method. This method involves constructing a comprehensive passenger behavior graph based on a passenger feature dataset to indicate relationships between passengers, inputting the comprehensive passenger behavior graph into a passenger segmentation model to obtain passenger segmentation results, and determining the corresponding marketing strategy for each passenger based on these results.

[0023] In this solution, specifically in this embodiment, a comprehensive passenger behavior graph integrating multiple features and topological structures is constructed. This considers both the node attributes of passengers and the relationships between passenger association edges, avoiding the shortcomings of traditional clustering methods that treat passengers as isolated individuals. This allows for accurate identification of passenger groups with stable behavioral patterns, improving the reliability and authenticity of the segmentation results. Based on these more reliable segmentation results, direct and reliable data support can be provided for airlines' core decisions such as differentiated pricing, precise promotions, and capacity allocation. This meets airlines' needs for refined passenger group operations, thereby improving customer relationship management efficiency and market competitiveness.

[0024] Figure 1 This is a schematic diagram illustrating an application scenario of a marketing strategy generation method provided in an embodiment of this disclosure. For example... Figure 1 As shown in the figure, the figure includes a first device and a second device.

[0025] The first device is the executing entity of the marketing strategy generation method provided in this embodiment. The second device is used to generate or store a passenger feature dataset of multiple passengers, which includes the passenger's identity data, behavioral feature data, and environmental feature data. The first and second devices can communicate via wired or wireless means.

[0026] The first device is used to acquire passenger characteristic data of multiple passengers in the second device.

[0027] In some embodiments, the first device and the second device may be communication equipment, electronic equipment, or servers, etc. A server may be a single physical server; or it may be two or more physical servers sharing different responsibilities, with multiple servers cooperating to achieve corresponding functions. Depending on the server type, for example, a server may be a blade server, a high-density server, a rack server, or a tower server, etc.

[0028] Let's take an electronic device as an example for explanation. Figure 2 This is a schematic diagram of the hardware structure of an electronic device provided in an embodiment of this disclosure.

[0029] See Figure 2 , Figure 2 The electronic device 200 shown may include a processor 201, a memory 202, a communication interface 203, and a bus 204. The processor 201, the memory 202, and the communication interface 203 can be connected via the bus 204.

[0030] The processor 201 is the control center of the electronic device 200, and can be a general-purpose central processing unit (CPU) or other general-purpose processors. The general-purpose processor can be a microprocessor or any conventional processor.

[0031] As an example, processor 201 may include one or more CPUs, for example Figure 2 CPU 0 and CPU 1 are shown in the diagram.

[0032] The memory 202 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but is not limited thereto.

[0033] In one possible implementation, the memory 202 may exist independently of the processor 201. The memory 202 can be connected to the processor 201 via a bus 204 and is used to store data, instructions, or program code. When the processor 201 calls and executes the instructions or program code stored in the memory 202, it can implement the airfare monitoring method provided in this embodiment of the disclosure.

[0034] In another possible implementation, the memory 202 can also be integrated with the processor 201.

[0035] The communication interface 203 is used for the electronic device 200 to connect with other devices via a communication network, which may be Ethernet, radio access network (RAN), wireless local area network (WLAN), etc. The communication interface 203 may include a receiving unit for receiving data and a transmitting unit for transmitting data.

[0036] Bus 204 can be an industry standard architecture (ISA) bus, a peripheral component interconnect (PCI) bus, or an extended industry standard architecture (EISA) bus. This bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 2 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0037] It should be pointed out that, Figure 2 The structure shown does not constitute a limitation on the first device, except Figure 2 In addition to the components shown, the first device may include more or fewer components than illustrated, or combine certain components, or have different component arrangements.

[0038] The following description uses an example of an electronic device as the executing entity of the airfare monitoring method provided in this embodiment of the disclosure.

[0039] Figure 3 This is a flowchart illustrating a marketing strategy generation method provided in an embodiment of the present disclosure, such as... Figure 3 As shown, the method includes: S301. Electronic devices acquire passenger feature datasets from multiple passengers.

[0040] The passenger feature dataset includes each passenger's identity data, behavioral feature data, and environmental feature data.

[0041] Identity data is data used to characterize the individual features of passengers. This identity data may include passenger attribute data and consumption data. For example, attribute data may include passenger identification, age, gender, occupation, and membership level. Consumption data may include passenger cabin class preferences (e.g., economy, first class, business class), and usage rates of ancillary services.

[0042] Behavioral characteristic data is used to characterize passenger behavior during air travel. For example, behavioral characteristic data may include the flight name, ticket purchase period, complaint number, ticket purchase time distribution, flight combination pattern, monthly / annual travel frequency, complaint type (such as service attitude complaint, flight delay complaint, etc.), and complaint processing status.

[0043] Environmental characteristic data is used to characterize the time information of passengers' flight travel and the activities they participate in. For example, environmental characteristic data includes the season (e.g., spring, summer, autumn, winter), specific month information, name of the activity participated in, type of activity (e.g., discount coupon activity, mileage doubler activity, etc.), and validity period of the activity.

[0044] In some embodiments, the passenger feature dataset may also include passenger relationship data, such as the intensity of social interaction between at least two passengers, frequency of travel together, similarity of routes and service behaviors, and word-of-mouth demand transmission effect.

[0045] In some embodiments, the electronic device may acquire passenger feature datasets of multiple passengers according to a preset period. For example, the preset period may be 1 day, 1 week, 1 month, 3 months, etc.

[0046] For periods with frequent passenger travel (such as July to August, January to February, and Saturdays and Sundays of each month), the preset period can be shorter; for example, the electronic device can acquire passenger feature datasets every other day. For periods with infrequent passenger travel (such as September to December, March to June, and weekdays of each month), the preset period can be longer; for example, the electronic device can acquire passenger feature datasets every week.

[0047] It is understandable that the data in the passenger feature dataset obtained by electronic devices from multiple passengers can be called the raw passenger feature data. The different data in the raw passenger feature data come from various channels such as ticketing, security checks, and online travel platforms. There are problems such as data redundancy, missing fields, chaotic format, abnormal noise, and inconsistent information. It is not easy to directly extract entities and associate relationships to build a knowledge graph. Therefore, before building a comprehensive passenger behavior graph, the raw passenger feature data can be preprocessed to obtain a passenger feature dataset composed of passenger feature data of each passenger with a standardized data structure.

[0048] In one feasible implementation, the electronic device can clean the raw feature data of each passenger, removing duplicate, invalid, and abnormal data, supplementing missing values ​​in key fields, and correcting erroneous data. Secondly, it integrates data from different sources for the same passenger using key fields such as ID number and membership card number. Subsequently, it standardizes and normalizes the data from different sources corresponding to each passenger, unifying the formats of fields such as ID information, travel time, and flight information, and digitally encodes the discrete features of each data point to eliminate differences in the units of measurement between different indicators and solve the problem of heterogeneity in multi-source data. Finally, the electronic device can perform data anonymization and filtering, compliantly handle passenger privacy information, and remove invalid and redundant fields, thereby obtaining passenger feature data for each passenger to form a passenger feature dataset.

[0049] By preprocessing the acquired raw passenger feature dataset, a structured and high-quality passenger feature dataset can be obtained, which can provide reliable data support for the subsequent entity construction and relationship analysis of the comprehensive passenger behavior map.

[0050] S302. Electronic devices construct a comprehensive passenger behavior map based on passenger feature datasets.

[0051] Among them, the passenger comprehensive behavior map is used to reflect the relationships between passengers.

[0052] In some embodiments, the electronic device can determine passenger nodes, behavioral feature nodes, and environmental feature nodes based on a passenger feature dataset, and determine the pairwise connections between passenger nodes, behavioral feature nodes, and environmental feature nodes, thereby constructing a comprehensive passenger behavior graph based on the passenger nodes, behavioral feature nodes, and environmental feature nodes. Subsequent embodiments will describe this in detail, and will not be repeated here.

[0053] S303. The electronic device inputs the comprehensive passenger behavior map into the passenger segmentation model to obtain the passenger segmentation results.

[0054] The classification results include high-frequency business travel groups, cross-border travel families, or price-sensitive individual travelers.

[0055] High-frequency business travel groups are used to characterize passengers who travel frequently by air, are highly sensitive to timeliness, and are low in price sensitivity. This type of passenger's travel behavior shows regular periodicity or fixed route preferences, has high requirements for flight times, punctuality, and flexibility in refunds and changes, and is more tolerant of ticket price fluctuations.

[0056] Cross-border tourism families are used to characterize travelers who travel in large numbers by air, whose travel time is concentrated on holidays, and who have complete route preferences. The travel behavior of this type of traveler is characterized by traveling in groups at fixed time periods, paying high attention to the convenience of direct flights and visa transit, and having relatively moderate price sensitivity.

[0057] Price-sensitive individual travelers are used to characterize passengers who travel by air with high randomness, are greatly affected by price, and have low loyalty. This type of passenger's travel behavior is characterized by uncertainty, short ticket purchase windows, significant influence from promotional activities, sensitivity to fare fluctuations, and relatively limited attention to flight brand affiliation, flight schedule, and service quality.

[0058] S304. Based on the segmentation results of each passenger, determine the corresponding marketing strategy for each passenger.

[0059] In some embodiments, for travelers belonging to high-frequency business travel groups, the focus is more on enhancing and retaining traveler value. Marketing strategies for this type of traveler may focus on providing exclusive service benefits and / or implementing churn prevention and intervention measures.

[0060] For example, exclusive service benefits include priority security checks and check-in for this type of passenger, access to exclusive VIP lounges, and complimentary cabin upgrades.

[0061] For example, churn warning intervention could involve monitoring the behavior of this type of passenger and issuing an alert when the passenger's travel frequency falls below a preset threshold, or having a dedicated account manager follow up with this type of passenger, or providing targeted benefits to this type of passenger.

[0062] In some embodiments, for travelers who are cross-border families, a greater emphasis is placed on scenario-based, precise operations. Marketing strategies for this type of traveler may focus on providing value-added marketing related to flight routes and / or offering differentiated family services.

[0063] For example, associated value-added marketing includes embedding bundled recommendations for family hotels, destination car rentals, and attraction packages in flight booking scenarios.

[0064] For example, differentiated family services include seat reservation, baggage allowance sharing, and optimized capacity deployment. Seat reservation refers to pre-locking adjacent seats for this type of passenger, which can alleviate the anxiety of family travelers about taking care of each other during the journey. Baggage allowance sharing allows family members to flexibly allocate the total weight of checked baggage, which can avoid additional fees caused by limits and thus improve the utilization rate of baggage resources. Optimized capacity deployment refers to flexibly increasing the number of flights during peak family travel periods and on popular family routes, alleviating the difficulties of ticket purchase and overcrowding for family travelers.

[0065] In some embodiments, for price-sensitive individual travelers, a greater emphasis can be placed on dynamic and targeted promotions. Marketing strategies for this type of traveler can focus on providing dynamic and targeted promotions and supplementary revenue generation.

[0066] For example, dynamic and precise promotions include low-occupancy-rate-triggered special offers, limited-time discounts, and remaining seat promotions. Low-occupancy-rate-triggered special offers refer to airlines monitoring the booking status of individual flights in real time. When the remaining seat ratio for a particular flight drops below a preset threshold, the airline triggers a price adjustment for that flight, targeting price-sensitive individual travelers with special offers, thereby improving the sales efficiency of remaining seats. Limited-time discounts refer to offering discounted fares below the regular selling price for designated routes or flights within a specific time period. By setting a clear purchase deadline or inventory limit, this can encourage hesitant travelers to accelerate their purchase decisions. The preset threshold can be an empirical value or a set value; this embodiment does not impose specific limitations on it.

[0067] In this embodiment, by constructing a comprehensive passenger behavior graph that integrates multiple features and topological structures, and considering both the node attributes of passengers and the relationships between passenger association edges, the traditional clustering method avoids treating passengers as isolated individuals. This enables accurate identification of passenger groups with stable behavioral patterns, improving the reliability and authenticity of the segmentation results. Based on more reliable segmentation results, direct and reliable data support can be provided for airlines' core decisions such as differentiated pricing, precise promotions, and capacity allocation, thereby improving customer relationship management efficiency and market competitiveness.

[0068] The following embodiments will provide a detailed description of step S302 above.

[0069] Figure 4 A flowchart illustrating another marketing strategy generation method provided in this disclosure embodiment is shown below. Figure 4 As shown, the method includes: S3021. Electronic devices are based on passenger feature datasets, with passengers as passenger nodes and passenger identity data as node attributes of passenger nodes.

[0070] Passenger nodes are used to represent each independent traveler entity and carry the passenger's static identity information.

[0071] In some embodiments, the electronic device can first extract the unique passenger IDs from the passenger feature dataset, remove invalid and duplicate identifiers, and generate an independent passenger node based on each valid ID obtained, thereby realizing passenger node partitioning.

[0072] S3022. Electronic devices are based on passenger feature datasets, using passenger behavior features as behavior feature nodes and behavior feature data as node attributes of behavior feature nodes.

[0073] Behavioral feature nodes are used to characterize various independent travel behavior dimensions of passengers, carrying quantitative indicators such as travel frequency, consumption, channels, refunds and changes.

[0074] In some embodiments, the electronic device can, based on the behavioral feature data of the passenger feature dataset, perform deduplication and classification of similar behavioral indicators, generating a behavioral feature node for each independent behavioral dimension, including travel frequency, cabin class preference, ticketing habits, refund and change behavior, and channel usage. Subsequently, using the behavioral feature node as the association key, the device retrieves the quantitative indicators, classification labels, behavioral thresholds, and other data corresponding to each passenger under each independent behavioral dimension from the behavioral feature data of the passenger feature dataset. The above data, after being standardized and formatted into a uniform text format, is used as the node attributes corresponding to the behavioral feature node.

[0075] S3023. Electronic devices are based on passenger feature datasets, with passenger environmental features as environmental feature nodes and environmental feature data as node attributes of environmental feature nodes.

[0076] Environmental feature nodes are used to characterize the external objective environmental factors involved in passengers' travel by air, carrying quantitative indicators of environmental dimensions such as routes, airports, travel times, peak and off-peak seasons, market fares, and passenger flow.

[0077] In some embodiments, the electronic device can obtain external environmental fields such as airport, route, travel time, and peak / off-peak season when each passenger travels by flight from the environmental feature data of the passenger feature dataset, and use these as the basic data source for environmental feature nodes; after uniformly normalizing and deduplicating invalid information for the extracted environmental elements, and setting a unique code for each unique environmental element, an independent environmental feature node is generated based on the unique code; then, the corresponding quantitative indicators and classification labels such as route passenger flow, time period congestion level, and peak season average price in the environmental feature data are collected and standardized into key-value pairs and attached to the corresponding nodes to form the node attributes of the environmental feature nodes.

[0078] S3024. Electronic devices construct associated edges based on passenger nodes, behavioral feature nodes, and environmental feature nodes.

[0079] The association edges include those between passenger nodes and other passenger nodes, those between passenger nodes and behavioral feature nodes, and those between behavioral feature nodes and environmental feature nodes. The association edges between passenger nodes and behavioral feature nodes can be used to match frequent business class travel, family standard class travel during holidays, and low-price off-season ticket purchases for individual travelers; the association edges between behavioral feature nodes and environmental feature nodes can be used to match family travel during winter and summer vacations, business travel on weekdays, and ticket purchases during promotional periods for individual travelers.

[0080] In some embodiments, the association edges between passenger nodes can be established by filtering multiple passengers such as those in the same order, the same flight, the same travel company, or the same travel group, and pairing them up to establish different types of association edges such as those for travelers in the same company or group. The association edges can also be attached with attribute information such as travel time, company code, and group number.

[0081] In one feasible implementation, the edges connecting a passenger node to other passenger nodes can be represented as: Equation (1).

[0082] in, This represents the set of edges connecting a passenger node to other passenger nodes. Indicates the first One passenger node; Indicates the first One passenger node; Indicates the first The first passenger node and the first The edge weights of the connections between passenger nodes; This represents the set of passenger nodes.

[0083] In some embodiments, the association edge between a passenger node and a behavioral feature node can be established by anchoring the corresponding passenger node with a unique passenger ID and matching all behavioral tags corresponding to that passenger node based on the same order record. The edge attributes corresponding to the association edge synchronously record quantitative information such as the frequency, consumption amount, preference level, and cancellation / modification records of that behavior. In one feasible implementation, the association edge between the passenger node and the behavior feature node can be represented as: Equation (2).

[0084] in, This represents the set of edges connecting passenger nodes and behavioral feature nodes. Indicates the first One passenger node; Indicates the first Each behavioral characteristic node; Indicates the first The first passenger node and the first The edge weights of the associated edges between nodes with behavioral characteristics; Represents the set of passenger nodes; Represents a set of behavioral feature nodes.

[0085] In some embodiments, the association edge between behavioral feature nodes and environmental feature nodes can be constructed by anchoring the corresponding passenger node with a unique passenger ID, and matching the corresponding environmental feature node based on external scenario information such as the flight, departure and arrival airport, and travel time associated with the same order. At the same time, environmental information such as travel time, route, peak and off-peak season, holidays, winter and summer vacations, promotions, and flight capacity can be marked on the association edge.

[0086] In one feasible implementation, the association edge between behavioral feature nodes and environmental feature nodes can be represented as: Equation (3).

[0087] in, This represents the set of edges connecting behavioral feature nodes and environmental feature nodes. Indicates the first Each behavioral characteristic node; Indicates the first One environmental feature node; The edge weight represents the connection between behavioral feature nodes and environmental feature nodes; Represents a set of behavioral feature nodes; This represents a set of environmental feature nodes.

[0088] S3025. Electronic devices construct a comprehensive passenger behavior graph based on passenger nodes, their node attributes, behavioral feature nodes, environmental feature nodes, and associated edges.

[0089] In this embodiment, by integrating multiple entities and various relationships of passengers, behavioral characteristics, and environmental characteristics to obtain a comprehensive passenger behavior map, multi-source information can be fused into a unified network within an entity and relationship framework. This allows for a global perspective on the relationships between passengers, passenger behavior, and the travel environment, avoiding the limitations of traditional statistical analysis in uncovering implicit group relationships and behavioral-environment linkage patterns among passengers. This provides a foundation for subsequent segmentation based on the comprehensive passenger behavior map.

[0090] In some embodiments, the passenger behavior graph also includes edge weights of the association edges between passenger nodes and other passenger nodes.

[0091] Among them, edge weights are used to characterize the degree of convergence of behavior between passengers and other passengers, that is, the degree of similarity between two passengers in their travel decisions, habits or preferences.

[0092] In some embodiments, the edge weights of the associated edges between passenger nodes and other passenger nodes are determined based on the quantified values ​​of the frequency of travel between passengers, the quantified values ​​of social interaction, and the quantified values ​​of behavioral similarity between passengers.

[0093] The frequency of travel together is used to characterize the frequency of travel between two travelers, such as the number of times they travel together as a family or as business travelers.

[0094] Social interaction quantification is used to characterize the intensity of social interaction between two travelers, such as daily social connections with family, friends, or colleagues.

[0095] Behavioral similarity metrics are used to characterize the similarity of behavior between two passengers when traveling via an air route, including the comprehensive similarity of passenger routes, ticketing habits, cabin class selection, and travel time. Behavioral similarity metrics can be used to quantify the convergence of routes and ticketing habits among business travelers, the similarity of travel patterns among family tourists, and the similarity of ticketing behavior among individual travelers during promotional periods.

[0096] In one feasible implementation, the edge weights of the connections between passenger nodes and other passenger nodes satisfy the following relationship: Equation (4).

[0097] in, Indicates the first The first passenger node and the first The edge weights between each passenger node range from 0 to 1; the larger the value, the greater the weight of the edge between the i-th and j-th passenger nodes. The first passenger node and the first The higher the degree of similarity in the behavior of individual passengers at each node; This represents the peer frequency weight coefficient, which can be dynamically adjusted to control the contribution ratio of peer relationships in the edge weight. Indicates the first The first passenger node and the first Quantitative value of the frequency of travel between individual passenger nodes; This represents the social interaction weight coefficient, which can be dynamically adjusted to control the contribution ratio of social interactions in the edge weights. Indicates the first The first passenger node and the first Quantitative value of social interaction between individual passenger nodes; This represents the behavioral similarity weight coefficient, which can be dynamically adjusted to control the contribution of travel behavior convergence in the edge weights. Indicates the first The first passenger node and the first A metric for behavioral similarity among passenger nodes.

[0098] In some embodiments, after obtaining the comprehensive passenger behavior map, a node feature matrix can be obtained based on the comprehensive passenger behavior map. This node feature matrix is ​​used to characterize the attribute features of each passenger node, reflects the inherent information of each node, and is a digital feature carrier of the node.

[0099] Specifically, by uniformly assigning global numbers to all nodes in the comprehensive passenger behavior graph, each node is transformed into a fixed-dimensional numerical vector. Then, all vectors are concatenated according to the node numbering order to obtain a node feature matrix where each row corresponds to a single node and each column corresponds to a feature dimension.

[0100] In one feasible implementation, the node feature matrix , denoted as the node feature matrix; N represents the total number of all nodes in the passenger comprehensive behavior graph, including passenger nodes, behavior feature nodes, and environmental feature nodes; F represents the feature dimension of a single node.

[0101] In some embodiments, an adjacency matrix can be obtained based on the passenger comprehensive behavior graph. This adjacency matrix is ​​used to characterize the association between passenger nodes and other passenger nodes, that is, the topological connection relationship between nodes in the passenger comprehensive behavior graph. By measuring whether there is an associated edge between any two nodes through the matrix value, all association relationships between passenger nodes, behavioral feature nodes and environmental feature nodes can be completely digitally restored, reflecting the overall network topology of the passenger comprehensive behavior graph.

[0102] Specifically, an initial zero matrix is ​​created with the total number of all passenger nodes in the passenger comprehensive behavior graph as the dimension. Node pairs with associated edges are filled with weight values ​​at the corresponding positions. If there is an associated edge, the actual edge weight is filled in; if there is no associated edge, 1 is filled in. Finally, a graph adjacency matrix that can represent the connection relationship between nodes is formed, which intuitively carries the overall topological structure information of the passenger comprehensive behavior graph.

[0103] In one feasible implementation, the adjacency matrix , Represents the adjacency matrix; This represents the total number of passenger nodes.

[0104] In summary, this comprehensive passenger behavior profile can be represented as follows: Equation (5).

[0105] in, Represents a comprehensive passenger behavior profile; This represents the set of all nodes in the comprehensive passenger behavior graph, including passenger nodes. Behavioral feature nodes Environmental feature nodes ; This represents the set of all associated edges in the passenger behavior graph; Represents the node feature matrix; This represents the graph adjacency matrix.

[0106] Figure 5 This is a schematic diagram of the structure of a comprehensive passenger behavior map provided in an embodiment of this disclosure. Figure 5 As shown, the graph includes passenger nodes, passenger nodes (other passengers), behavioral feature nodes and environmental feature nodes, passenger node-passenger node association edges, passenger node-behavioral feature node association edges, and behavioral feature node-environment feature node association edges.

[0107] Understandably, in reality, travel via air routes will involve new passengers, new behavioral events, and new environmental factors, thus necessitating updates to the comprehensive passenger behavior map. The following examples detail a method for dynamically and incrementally updating the comprehensive passenger behavior map.

[0108] Figure 6 This is a flowchart illustrating yet another marketing strategy generation method provided in an embodiment of this disclosure. Figure 6 As shown, it includes: S601. Electronic devices acquire passenger feature datasets of newly added passengers.

[0109] The newly added passenger characteristic dataset includes the identity data, behavioral characteristic data, and environmental characteristic data of each newly added passenger.

[0110] The above step S301 provides a detailed explanation of identity data, behavioral characteristic data, and environmental characteristic data, which will not be repeated here.

[0111] In some embodiments, the electronic device can acquire a dataset of passenger characteristics of the new passengers when the number of new passengers exceeds a preset threshold. The preset threshold can be a set value or an empirical value.

[0112] S602. Electronic devices determine new passenger nodes, new behavioral feature nodes, and new environmental feature nodes based on passenger feature datasets.

[0113] This step is the same as the process in steps S3021-S3023 above where the electronic device determines passenger nodes, behavioral feature nodes, and environmental feature nodes based on the passenger feature dataset, and will not be repeated here.

[0114] S603. Electronic devices construct new associated edges based on newly added passenger nodes, newly added behavioral feature nodes, and newly added environmental feature nodes.

[0115] This step is the same as the process described above for electronic devices to construct associated edges based on passenger nodes, behavioral feature nodes, and environmental feature nodes, and will not be repeated here.

[0116] S604. The electronic device updates the associated edges in the passenger comprehensive behavior graph based on the time decay function to obtain the updated associated edges.

[0117] In some embodiments, the formula for the time decay function can be expressed as: Equation (6).

[0118] in, Indicates historical behavior over time The effective edge weights after decay; This represents the attenuation coefficient, used to control the rate of attenuation. Used to represent the time interval between the time when the action occurs and the current time, that is, the difference between the time when the action occurs and the current time.

[0119] This formula relies on passenger time-series behavior data and combines environmental variables such as seasons and holidays to complete the weight amplitude. The associated edges are updated through this time decay function. Following the business logic that recent travel behavior has higher weight and the weight of older behavior naturally decays, it can weaken the interference of outdated historical records, accurately match the current real travel preferences and needs of passengers, and improve the accuracy and real-time performance of the knowledge graph.

[0120] S605: The electronic device incrementally updates the passenger comprehensive behavior graph based on newly added passenger nodes, newly added behavioral feature nodes, newly added environmental feature nodes, newly added associated edges, and updated associated edges, to obtain the updated passenger comprehensive behavior graph.

[0121] It should be noted that incremental updates to the passenger behavior graph are only lightweight graph update operations, that is, only the nodes and relationships that have changed are updated, and it is not necessary to reconstruct the entire passenger behavior graph, thus avoiding the resource consumption caused by a full recalculation.

[0122] Specifically, new passenger nodes are added and node attribute vectors are initialized. Then, the weights of the edges connecting passenger nodes with other passenger nodes are updated synchronously. Edges connecting passenger nodes and behavioral feature nodes are added in batches, and edges connecting behavioral feature nodes and environmental feature nodes are updated. Only nodes and relationships that have changed are locally modified. After local incremental updates, the complete passenger behavior graph for the next time step is output.

[0123] In some embodiments, after obtaining the updated passenger comprehensive behavior graph, a subdivision incremental update logic can be initiated to select only the local subgraphs that have undergone structural and weight changes for subsequent passenger segmentation; after the segmentation is completed, the updated passenger comprehensive behavior graph can be used as the historical graph data source for the next cycle, and a new round of data flow can be connected to form a continuously iterative time-series graph update closed loop.

[0124] For the updated passenger behavior graph, a threshold for the number of new additions can be set. If the number of new additions exceeds the threshold, only the local subgraphs can be re-clustered, eliminating the need for full retraining and thus enabling low-cost real-time maintenance of passenger segmentation results.

[0125] In one feasible implementation, the node feature matrix corresponding to the addition of passenger nodes, behavioral feature nodes, and environmental feature nodes can be represented as follows: Equation (7).

[0126] in, Indicates the current time The node feature matrix; Indicates the time before the current time. The node feature matrix; This represents the node feature matrix corresponding to newly added passenger nodes, newly added behavioral feature nodes, and newly added environmental feature nodes.

[0127] In one feasible implementation, the adjacency matrix corresponding to the addition of passenger nodes, behavioral feature nodes, and environmental feature nodes can be represented as: Equation (8).

[0128] in, Indicates the current time The adjacency matrix; Indicates the time before the current time. The adjacency matrix; This represents the adjacency matrix corresponding to newly added passenger nodes, newly added behavioral feature nodes, and newly added environmental feature nodes.

[0129] In this embodiment, a time decay function and an incremental learning mechanism are introduced to dynamically update node attributes, associated edges, and their weights in the passenger comprehensive behavior graph without requiring re-clustering of the entire dataset. By dynamically updating the passenger comprehensive behavior graph incrementally, it is not necessary to recalculate all nodes and matrices; only local changes need to be modified, significantly reducing data maintenance costs and ensuring that the subdivided model can continuously and in real-time reflect changes in passenger behavior patterns. At the same time, by dynamically adjusting edge weights in conjunction with the time decay function, the influence of outdated behaviors can be weakened, thereby more accurately reflecting the current travel status of passengers.

[0130] In some embodiments, step S303 may include: inputting the adjacency matrix and node feature matrix corresponding to the passenger comprehensive behavior map into the passenger segmentation model to obtain the passenger segmentation result.

[0131] In some embodiments, the adjacency matrix and node feature matrix corresponding to the passenger comprehensive behavior map can be input into the passenger segmentation model to obtain the embedding vector of each passenger node. Then, based on the embedding vector of each passenger node, the passenger nodes can be clustered using a preset clustering algorithm to obtain the segmentation result of each passenger.

[0132] In some embodiments, the passenger segmentation model includes a random graph enhancement module for performing random graph destruction operations on the input original passenger graph.

[0133] In one feasible implementation, before obtaining the embedding vectors of each passenger node, the passenger segmentation model can be based on an original passenger graph G=(A, X) composed of an adjacency matrix and a node feature matrix, where A represents the adjacency matrix and X represents the node feature matrix. The random graph enhancement module can perform two independent random graph destruction operations on this original passenger graph, generating two views G1=(A1, X1) and G2=(A2, X2) with perturbed structures but consistent semantics. Samples from passengers with the same name are considered positive sample pairs, while samples from different passengers are considered negative sample pairs.

[0134] The random graph disruption operation can include two types of random perturbations: node feature perturbation and graph topology perturbation. Node feature perturbation can add Gaussian noise to a small number of features by randomly masking part of the passenger feature dimensions; graph topology perturbation can randomly delete some associated edges between passengers and randomly add a small number of weakly associated edges.

[0135] In this embodiment, there is no need to collect the original passenger feature data of new passengers. Instead, multiple usable samples are generated by random perturbation within the passenger segmentation model, which can enrich the diversity of passenger graph data.

[0136] In some embodiments, the passenger segmentation model includes a two-layer graph attention encoder module for obtaining the embedding vector of each passenger node.

[0137] Specifically, the dual-layer graph attention encoder takes two sets of views G1=(A1,X1) and G2=(A2,X2) obtained by random graph destruction as input, and first performs feature linear transformation on the node features of each passenger node in each view through a learnable weight matrix.

[0138] In one feasible implementation, the learnable weight matrix is: .in, Represents the learnable weight matrix; This indicates the feature dimension after the transformation at this layer; This indicates the output dimension of the previous layer.

[0139] Next, the attention weight scores of passenger nodes to neighboring passenger nodes are calculated. The attention weight coefficients are then weighted and aggregated with the features of the passenger node and its neighbors using exponential (soft maximum, softmax) normalization to obtain the standardized attention weights.

[0140] In one feasible implementation, the formula corresponding to the attention weight score can be expressed as: Equation (9).

[0141] in, Indicates the first The first passenger node pairs with its neighbors. The attention weight score of each passenger node can amplify the node weight of strong connections between business groups and families traveling together; Represents the feature edge weight matrix; Represents the attention parameter vector; Used to represent nonlinear activation functions; Indicates feature concatenation operation; Indicates the first A set of first-order neighbors of each passenger node; Indicates the first The node characteristics of each passenger node; Indicates the first The neighbor of the first passenger node The node characteristics of each passenger node.

[0142] Finally, after stacking two layers of graph attention network (GAT), the embedding vector Z1 of the passenger node corresponding to view G1 and the embedding vector Z2 of the passenger node corresponding to view G2 are obtained respectively.

[0143] In one feasible implementation, the first-layer graph node embedding formula can be expressed as: Equation (10).

[0144] in, This represents the first layer of the graph attention encoder. Embedsion vectors of each passenger node; Indicates the activation function; Indicates the first A set of first-order neighbors of each passenger node; Indicates the first layer corresponding to the first layer. The first passenger node pairs with its neighbors. Attention weight score of each passenger node; This represents the learnable parameters of the first layer; Indicates the first The neighbor of the first passenger node The node characteristics of each passenger node.

[0145] In one feasible implementation, the second-layer graph node embedding formula can be expressed as: Equation (11).

[0146] in, This represents the second layer of the graph attention encoder. Embedsion vectors of each passenger node; Indicates the activation function; Indicates the first A set of first-order neighbors of each passenger node; Indicates the second layer corresponding to the first The first passenger node pairs with its neighbors. Attention weight score of each passenger node; This represents the learnable parameters of the second layer; Indicates the first layer The neighbor of the first passenger node The node characteristics of each passenger node.

[0147] In some embodiments, the passenger segmentation model includes a contrastive learning optimization module. Before clustering passenger nodes based on the embedding vectors of each passenger node using a preset clustering algorithm to obtain the segmentation results for each passenger, the GAT encoder weights can be optimized based on the contrastive learning optimization module. This allows the dual-view embedding vectors corresponding to the same passenger node to be as close as possible, and the embedding vectors corresponding to different passenger nodes to be as far apart as possible, thereby widening the embedding distance between business travelers, families, and individual travelers, narrowing the distance between similar customer groups, and improving the segmentation differentiation.

[0148] Specifically, positive and negative sample pairs are first constructed based on the dual-view embedding vectors corresponding to each passenger node. Positive sample pairs consist of the embedding vectors of the same passenger node in both views, while negative samples consist of the embedding vectors of all other passenger nodes in the current batch, excluding the original passenger node. Subsequently, a graph contrastive learning loss function is used to minimize the contrastive loss of all positive sample pairs, thereby optimizing the encoder parameters. This increases the embedding distance between business travelers, families, and individual travelers, while decreasing the distance between similar passenger groups, thus improving the segmentation and discrimination.

[0149] In one feasible implementation, the contrastive learning loss function can be expressed as: Equation (12).

[0150] in, This represents the contrastive learning loss function; Indicates embedded cosine similarity; This represents the temperature coefficient, used to control the smoothness of the distribution. Used to represent the dual-view embedding vector corresponding to the same passenger node; This is used to represent the embedding vectors corresponding to all other passenger nodes in the current batch, excluding this passenger node.

[0151] In some embodiments, the passenger segmentation model includes a self-expressive subspace clustering module, which is used to mine the group subspace by relying on the linear self-expressive properties of the embedding vectors of passenger nodes, thereby realizing passenger segmentation and obtaining the segmentation results of each passenger.

[0152] Specifically, a linear modeling relationship is first established, assuming that the embedding vector of any passenger node can be obtained by a weighted linear combination of the embedding vectors of the other passenger nodes. For the th... Embedded vectors of each passenger node It can be represented as: Equation (13).

[0153] in, Indicates the first Embedsion vectors of each passenger node; Represents the self-expression coefficient, representing the first... The embedding vector of the _th passenger node is reconstructed by linear combination. The weight assigned to the embedding vector of each passenger node. The larger the value, the higher the number. The first passenger node and the first The higher the probability that multiple passenger nodes belong to the same subspace, the greater the probability. All self-expression coefficients constitute a self-expression matrix. N represents the total number of passenger nodes.

[0154] Passengers are then segmented based on a pre-defined clustering algorithm. This pre-defined clustering algorithm can be a self-expressive subspace clustering objective function, which can be represented as: Equation (14).

[0155] in, This represents the matrix consisting of the embedding vectors of all passenger nodes; Represents the self-expression coefficient matrix; The Frobenius norm is used to characterize the reconstruction error. Represents the regularization coefficient; express Norms are used to control the sparsity of coefficients.

[0156] By constructing a reconstruction loss that simultaneously incorporates the Frobenius norm and The joint optimization objective function with norm sparse constraints balances the weights of the two losses by relying on the hyperparameter λ, filtering out irrelevant weak correlations while ensuring the accuracy of the embedded linear reconstruction. The optimal self-representation matrix is ​​solved iteratively by gradient descent. Once the passenger segmentation model converges, passengers with strong correlations are selected based on the magnitude of their coefficients and assigned to the same linear subspace, thus completing the passenger segmentation.

[0157] In this embodiment, by employing a passenger segmentation model based on a self-expressive attention graph neural network, the unsupervised two-layer graph attention encoder and the self-expressive module work together to adaptively learn the association weights between passenger nodes and construct a semantically consistent passenger subspace in a high-dimensional embedding space. This automatically clusters three market segments: high-frequency business travel groups, cross-border travel families, and price-sensitive individual travelers. This achieves high-quality passenger clustering without the need for manual annotation, significantly reducing segmentation costs.

[0158] In conclusion, Figure 7 This is a schematic diagram of the structure of a passenger segmentation model provided in an embodiment of this disclosure. Figure 7 As shown, the passenger segmentation model includes an input module, a random graph augmentation module, a two-layer graph attention encoder module, a contrastive learning optimization module, a self-expressive subspace clustering module, and an output module.

[0159] The input module is used to receive the adjacency matrix and node feature matrix as input.

[0160] The random graph enhancement module, the two-layer graph attention encoder module, the contrastive learning optimization module, and the self-expressive subspace clustering module have been described in detail in the above embodiments, and will not be repeated here.

[0161] The output module is used to output the segmentation results for each passenger.

[0162] The foregoing primarily describes the solutions of the embodiments of this disclosure from a methodological perspective. It is understood that, in order to achieve the above-mentioned functions, the marketing strategy generation apparatus includes at least one of the hardware structures and software modules corresponding to the execution of each function. Those skilled in the art should readily recognize that, in conjunction with the units and algorithm steps of the various examples described in the embodiments disclosed herein, the embodiments of this disclosure can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the embodiments of this disclosure.

[0163] This disclosure embodiment can divide the marketing strategy generation device into functional modules according to the above method embodiment. For example, each function can be divided into a separate functional module, or two or more functions can be integrated into one functional module. The integrated module can be implemented in hardware or software. It should be noted that the module division in this disclosure embodiment is illustrative and only represents one logical functional division. In actual implementation, there may be other division methods. The following description uses the example of dividing each functional module according to each function.

[0164] Figure 8 This is a schematic diagram of a marketing strategy generation device provided in an embodiment of the present disclosure. This device is applied to a first device and can execute the marketing strategy generation method provided in the above-described method embodiments. Figure 8 As shown, the marketing strategy generation device 800 includes: a processing module 801, a communication module 802, and a storage module 803.

[0165] The communication module 802 is used to acquire passenger feature datasets of multiple passengers. The passenger feature datasets include each passenger's identity data, behavioral feature data, and environmental feature data.

[0166] Processing module 801 is used to construct a comprehensive passenger behavior map based on the passenger feature dataset. The comprehensive passenger behavior map is used to reflect the relationships between each passenger and other passengers.

[0167] The processing module 801 is also used to input the comprehensive behavior map of passengers into the passenger segmentation model to obtain the segmentation results of each passenger, including high-frequency business travel groups, cross-border travel families, or price-sensitive individual travelers.

[0168] The processing module 801 is also used to determine the corresponding marketing strategy for each passenger based on the segmentation results of each passenger.

[0169] The processing module 801 may be a central processing unit (CPU), a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. The processor may include an application processor and a baseband processor. It may implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with this disclosure. The processor may also be a combination that implements computational functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc.

[0170] The communication module 802 is also used to establish a communication connection between the processing module 801 and the storage module 803.

[0171] In some embodiments, the storage module 803 is used to store passenger feature datasets, segmentation results of each passenger, and marketing strategies corresponding to each passenger.

[0172] For example, the processing module 801 can be as follows: Figure 2 The processor 201 shown; the communication module 802 can be as follows: Figure 2 The communication interface 203 shown; the storage module 803 can be as follows: Figure 2 The internal memory 202 shown.

[0173] Through the above description of the embodiments, those skilled in the art will clearly understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The specific working process of the system, device, and unit described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0174] This disclosure provides a computer program product containing instructions that, when run on a computer, cause the computer to execute the marketing strategy generation method described in the above method embodiments.

[0175] This disclosure also provides a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the marketing strategy generation method in the method flow shown in the above method embodiments.

[0176] The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires; a portable computer disk drive; a hard disk drive; random access memory (RAM); read-only memory (ROM); erasable programmable read-only memory (EPROM); a register; a hard disk drive; an optical fiber; a compact disc read-only memory (CD-ROM); an optical storage device; a magnetic storage device; or any suitable combination thereof; or any other form of computer-readable storage medium known in the art. An exemplary storage medium is coupled to a processor, enabling the processor to read information from and write information to the storage medium. Of course, the storage medium may also be a component of the processor. The processor and the storage medium may reside in an application-specific integrated circuit (ASIC).

[0177] In this embodiment of the disclosure, the computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0178] Since the marketing strategy generation device, computer-readable storage medium, and computer program product in the embodiments of the present invention can be applied to the above method, the technical effects obtained can also be referred to the above method embodiments. The embodiments of the present invention will not be repeated here.

[0179] In the embodiments provided in this disclosure, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the shown or discussed mutual couplings, direct couplings, or communication connections may be through some interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0180] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0181] In addition, the functional units in the various embodiments of this disclosure can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0182] The above are merely specific embodiments of this disclosure, but the scope of protection of this disclosure is not limited thereto. Any changes or substitutions within the technical scope disclosed in this disclosure should be included within the scope of protection of this disclosure. Therefore, the scope of protection of this disclosure should be determined by the scope of the claims.

Claims

1. A marketing strategy generation method, characterized in that, The method includes: Obtain passenger feature datasets for multiple passengers; the passenger feature datasets include the identity data, behavioral feature data, and environmental feature data of each passenger; Based on the passenger feature dataset, a comprehensive passenger behavior graph is constructed; the comprehensive passenger behavior graph is used to reflect the relationship between each passenger and other passengers. The comprehensive passenger behavior map is input into the passenger segmentation model to obtain the segmentation results of each passenger, which include high-frequency business travel groups, cross-border travel families, or price-sensitive individual travelers. Based on the segmentation results of each passenger, a corresponding marketing strategy is determined for each passenger.

2. The method according to claim 1, characterized in that, The step of constructing a comprehensive passenger behavior map based on the passenger feature dataset includes: Based on the passenger feature dataset, the passenger is taken as a passenger node, and the passenger's identity data is taken as the node attribute of the passenger node; Based on the passenger feature dataset, the passenger's behavioral features are used as behavioral feature nodes, and the behavioral feature data is used as the node attribute of the behavioral feature nodes. Based on the passenger feature dataset, the passenger's environmental features are used as environmental feature nodes, and the environmental feature data is used as the node attribute of the environmental feature nodes. Based on the passenger node, the behavior feature node, and the environment feature node, an association edge is constructed, which includes the association edge between the passenger node and other passenger nodes, the association edge between the passenger node and the behavior feature node, and the association edge between the behavior feature node and the environment feature node; Based on the passenger node, the node attributes of the passenger node, the behavioral feature node, the node attributes of the behavioral feature node, the environmental feature node, the node attributes of the environmental feature node, and the associated edges, the comprehensive passenger behavior graph is constructed.

3. The method according to claim 2, characterized in that, The comprehensive passenger behavior graph includes the edge weights of the association edges between the passenger node and other passenger nodes, and the edge weights are used to characterize the degree of behavioral similarity between the passenger and other passengers.

4. The method according to claim 3, characterized in that, The edge weights of the connections between the passenger nodes and other passenger nodes satisfy the following relationship: ； in, Indicates passenger and passengers Edge weights between them; This represents the peer frequency weighting coefficient; Indicates passenger and passengers The quantification value of peer frequency between them; This represents the social interaction weighting coefficient; Indicates passenger and passengers Quantitative value of social interaction between them; This represents the weighting coefficient for behavioral similarity. Indicates passenger and passengers The behavioral similarity metric between them.

5. The method according to claim 3, characterized in that, The method further includes: Obtain the passenger characteristic dataset of newly added passengers; Based on the passenger feature dataset, new passenger nodes, new behavioral feature nodes, and new environmental feature nodes are identified. Based on the newly added passenger node, the newly added behavioral feature node, and the newly added environmental feature node, construct new related edges; The associated edges in the passenger comprehensive behavior graph are updated based on the time decay function to obtain the updated associated edges; Based on the newly added passenger nodes, newly added behavioral feature nodes, newly added environmental feature nodes, newly added associated edges, and the updated associated edges, the passenger comprehensive behavior graph is incrementally updated to obtain the updated passenger comprehensive behavior graph.

6. The method according to claim 1, characterized in that, The method further includes: Based on the passenger comprehensive behavior graph, an adjacency matrix and a node feature matrix corresponding to the passenger comprehensive behavior graph are obtained; the adjacency matrix is ​​used to characterize the association relationship between the passenger node and other passenger nodes; the node feature matrix includes the attribute features of each passenger node; The step of inputting the comprehensive passenger behavior map into the passenger segmentation model to obtain the passenger segmentation results includes: The adjacency matrix and node feature matrix corresponding to the comprehensive passenger behavior graph are input into the passenger segmentation model to obtain the passenger segmentation result.

7. The method according to claim 6, characterized in that, The step of inputting the adjacency matrix and node feature matrix corresponding to the passenger comprehensive behavior map into the passenger segmentation model to obtain the passenger segmentation result includes: The adjacency matrix and node feature matrix corresponding to the passenger comprehensive behavior graph are input into the passenger segmentation model to obtain the embedding vector of each passenger node; Based on the embedding vectors of each passenger node, the passenger nodes are clustered using a preset clustering algorithm to obtain the segmentation results of each passenger.

8. An electronic device, characterized in that, The electronic device includes a processor, a communication interface, and a memory, wherein the memory stores at least one computer program, and the at least one computer program is loaded and executed by the processor to implement the marketing strategy generation method as described in any one of claims 1 to 7.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores at least one computer program, which is loaded and executed by a processor to implement the marketing strategy generation method as described in any one of claims 1 to 7.

10. A computer program product, characterized in that, The computer program product includes a computer program or instructions that, when executed by a processor, implement the marketing strategy generation method as described in any one of claims 1 to 7.

Images