Flight trip recommendation system and method

Abstract

The invention discloses a flight trip recommendation system and method. The method comprises the following steps: obtaining trip demand information of a user; obtaining a candidate flight set comprising a plurality of candidate flights according to the user travel demand information, and identifying and determining at least one feasible potential travel mode based on the place of departure and the geographic position of the airport corresponding to each candidate flight; collecting and fusing dynamic influence parameters, special affiliated data and static geographic information from the departure place to the airport related to each potential travel mode to evaluate the potential travel mode corresponding to each candidate flight, and obtaining an evaluation result of each potential travel mode; and generating a comprehensive recommendation value of each candidate flight in combination with the flight attribute data of each candidate flight and the evaluation result, and outputting a flight recommendation scheme according to the comprehensive recommendation value. Through the recommended scheme, the real demand and value orientation of the user can be more accurately met.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of traffic travel, and particularly relates to a flight travel recommendation system and method. BACKGROUND

[0002] With the popularity of air travel and the development of Internet technology, flight query and booking services have become an important part of public travel planning. At present, the flight query systems or platforms on the market mainly rely on traditional lists or tables to show users the basic information such as flight schedules, ticket prices and cabin space provided by ticket vendors. Users usually need to manually compare the advantages and disadvantages of different flights in terms of price, time, transfer times, etc. on the basis of these information, combined with personal experience, to make travel decisions.

[0003] However, the prior art solution has the following disadvantages in actual application: the information is scattered in different platforms, and the user needs to repeatedly switch to obtain flight, punctuality rate, weather and other data, which is cumbersome to operate; the interaction mode is single, and the list form is mainly used, lacking visual presentation and intelligent recommendation based on user preferences and real-time data; more importantly, the existing solution only focuses on the "airport to airport" flight segment, and does not include the ground connection mode from the actual departure location of the user to the airport in the overall evaluation, resulting in an incomplete and inaccurate recommended travel plan. SUMMARY

[0004] To solve the technical problems in the background art, the present application provides a flight travel recommendation system and method.

[0005] The flight travel recommendation method provided by the present application comprises: S1, obtaining travel demand information of a user, the travel demand information comprising a departure location, a destination and a travel date; S2, obtaining a candidate flight set comprising a plurality of candidate flights according to the travel demand information of the user, and identifying and determining at least one feasible potential travel mode based on the geographical position of the departure location and the airport corresponding to each candidate flight; S3, collecting and fusing dynamic influence parameters, special auxiliary data related to each potential travel mode, and static geographical information from the departure location to the airport, to evaluate the potential travel mode corresponding to each candidate flight, and obtain evaluation results of each potential travel mode; S4, generating a comprehensive recommendation value of each candidate flight in combination with flight attribute data of each candidate flight and the evaluation results, and outputting a flight recommendation plan according to the comprehensive recommendation value.

[0006] Preferably, in S1, the obtaining of the travel demand information of the user specifically comprises: receiving a geographic point selection operation of the user on a departure location and a destination location through a map interaction interface, and generating initial travel demand information containing the departure location and the destination location; receiving a travel date input by the user on the map interaction interface, and merging the initial travel demand information to form complete travel demand information.

[0007] Preferably, S2 specifically comprises: performing format standardization processing on the travel demand information to generate structured request data; calling a ticket service interface based on the structured request data to obtain a corresponding candidate flight set; calling a map service interface or a traffic planning service in combination with user departure location information for a geographic location of a departure airport corresponding to each candidate flight in the candidate flight set to obtain multiple potential travel modes from the departure location to the airport; the potential travel modes include but are not limited to self-driving, online car hailing, taxi, bus, and subway.

[0008] Preferably, S3 specifically comprises: for each potential travel mode, collecting and fusing dynamic influence parameters, static geographic information, and special auxiliary data to form a multi-dimensional feature vector of the potential travel mode, wherein the dynamic influence parameters are current and predicted traffic information obtained by calling a real-time traffic data interface and a weather data interface, the static geographic information is path information from the departure location to the airport obtained by a map service interface, and the special auxiliary data is additional parameters obtained according to the type of the travel mode; according to a preset priority configuration, evaluating feature components representing efficiency, cost, convenience, and reliability dimensions in the multi-dimensional feature vector; based on an importance relationship defined by the priority configuration, comprehensively evaluating the results of each dimension to generate a final evaluation result of each potential travel mode.

[0009] Preferably, the preset priority configuration can be dynamically adjusted according to the travel preference data of the user, specifically comprising: obtaining the travel preference data of the user; analyzing the travel preference data to extract preference parameters for quantitatively representing the degree of attention of the user in the efficiency, cost, convenience, and reliability dimensions; based on the preference parameters, generating a group of correction instructions for adjusting the importance relationship of each dimension; and updating the importance relationship of each dimension defined in the priority configuration according to the correction instructions.

[0010] Preferably, the travel preference data is derived from at least one of the following ways: configuration information actively submitted by the user through a system interface or default configuration information generated by analyzing historical orders and query records of the user.

[0011] Preferably, S4 specifically comprises: S41, for each candidate flight in the candidate flight set, obtain its flight attribute data and associated weather data, wherein the flight attribute data at least includes ticket price, total journey time, number of transfers, historical punctuality rate and seat availability, and the associated weather data at least includes weather forecast information of the departure airport and the destination airport within a preset time period before and after the planned takeoff time of the flight; S42, based on the obtained flight attribute data and associated weather data, comprehensively considering the economy, reliability and experience dimensions, the flight attribute comprehensive score of the candidate flight is calculated; S43, obtain the evaluation result of all potential travel modes corresponding to the candidate flight; S44, overall analysis is performed on the flight attribute comprehensive score and the evaluation result, and the best matching relationship between the flight and the ground connection mode before flight is judged according to the preset decision logic, and a comprehensive recommendation value representing the overall advantages and disadvantages of the flight and its connected travel mode is generated accordingly; S45, according to the comprehensive recommendation value of all candidate flights, the top ranked scheme is selected and added with a highlighted mark and output as a flight recommendation scheme.

[0012] Preferably, the calculation of the flight attribute comprehensive score can be dynamically adjusted according to the personal flight preference parameters of the user, specifically: Obtain the personal flight preference parameters of the user; analyze the personal flight preference parameters to extract personalized tendency information representing the user's preference in the economy, reliability and experience dimensions; adjust the comprehensive consideration of the flight attribute data and associated weather data in the economy, reliability and experience dimensions according to the personalized tendency information; and calculate the flight attribute comprehensive score that integrates the user's personalized tendency based on the adjusted comprehensive consideration.

[0013] Preferably, it further comprises S5, which is specifically: obtaining ticketing platform booking interface information corresponding to the flight recommendation scheme; generating and displaying a user interface containing an interactive booking control; in response to the triggering operation of the interactive booking control, calling the corresponding ticketing platform booking interface and jumping to the ticket purchase page.

[0014] The application also provides a flight travel recommendation system, comprising: A demand obtaining interaction module is used to receive the departure place, destination and travel date input by the user through an interactive map; A flight connection scheme generation module is used to obtain a candidate flight set according to the demand, and to identify at least one potential travel mode to the departure airport corresponding to each candidate flight; The connection scheme evaluation module is configured to fuse multi-source data to form feature information for each potential travel mode, perform multi-dimensional evaluation based on a preset priority configuration, fuse the results, and output evaluation results of each potential travel mode. The comprehensive recommendation generation module is configured to calculate a comprehensive score of each flight attribute based on flight attribute data, associated meteorological data, and user preference data, analyze and calculate the comprehensive recommendation value of each flight according to a preset decision logic based on the evaluation results, and output a flight recommendation scheme. The display output module is configured to visually display the flight recommendation scheme and provide a jump interface to a ticketing platform.

[0015] In the present application, by collecting, fusing, and calculating data dispersed on multiple platforms, and through intelligent sorting and visual presentation, users can quickly choose from a small number of selected, high-confidence recommendation schemes in a single interface, simplifying user operations and greatly shortening decision-making time. Through map interaction design, users can intuitively select the departure and destination on the map, and the system renders the route in real time, changing the abstract interaction mode of traditional lists and reducing the difficulty of operation, providing a more intuitive user experience. By creatively including various potential travel modes (such as self-driving and subway) from the user's actual departure location to the airport in the evaluation system and dynamically injecting user preference-related data into the core evaluation process, the recommended scheme can more accurately meet the user's real needs and value orientation, providing a customized service experience for "thousands of people with thousands of faces". BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 A flowchart of a flight travel recommendation method according to the present application. DETAILED DESCRIPTION

[0017] REFERENCE Figure 1 The flight travel recommendation method according to the present application includes the following steps: S1, obtaining user travel demand information, the travel demand information including departure location, destination, and travel date.

[0018] In this step, the system presents an interactive map to the user, and the user specifies the departure location and destination by clicking, long-pressing, or dragging the selected area on the map. The system captures the two geographic coordinate points (or areas) and parses them into available location identifiers (such as city names or airport codes) to generate initial travel demand information containing the departure location and destination. This approach greatly improves the intuitiveness and convenience of operation, especially for users who are not familiar with airport names or want to plan their trip from a macro geographical perspective.

[0019] After determining the geographic location of the departure location, the user usually needs to input or select the travel date (one-way date or round-trip date) in the control interface within or associated with the map interface. The system receives this date information, merges it with the initial travel demand information generated in the previous step, and structures and packages it to form a complete travel demand information that can be processed in subsequent steps.

[0020] By using intuitive spatial point selection instead of abstract code input, the user's usage threshold is lowered, and the accuracy and machine readability of the obtained "departure location, destination, and travel date" information are ensured, which helps to improve user experience and decision-making efficiency.

[0021] S2, according to the user travel demand information, a candidate flight set containing several candidate flights is obtained, and based on the geographic location of the departure location and the airport corresponding to each candidate flight, at least one feasible potential travel mode is identified and determined.

[0022] The specific process is as follows: First, demand standardization and structuring: the system performs format standardization processing on the complete travel demand information obtained in step S1. This includes uniform geographic location coding, date format, and possible additional preference filtering conditions, which are converted into an internal recognizable structured request data.

[0023] Next, flight information acquisition: the system calls one or more ticketing interfaces based on the above structured request data. These interfaces can access publicly available airline direct sales systems, global distribution systems, or inventory of online travel agencies. The interface returns a candidate flight set that meets the user's departure location, destination, and date, where each flight entry contains at least flight number, departure and arrival airport, time, and other flight information.

[0024] Finally, generate flight-related pre-flight ground transfer scheme: for each candidate flight in the candidate flight set, the system performs the following sub-steps to realize the binding analysis of flight and pre-flight ground transportation. The analysis process is as follows: (1) Positioning analysis: according to the departure airport code of the candidate flight, the accurate geographic location is obtained.

[0025] (2) Connection mode discovery: combined with the specific departure location information provided by the user, call the map service interface or professional traffic planning service. Based on the spatial relationship between the two points and real-time traffic network data, the system automatically calculates and obtains multiple potential travel modes from the user's departure location to the specific departure airport.

[0026] (3) Mode enumeration: these potential travel modes are a collection of multiple transportation modes, typically including but not limited to: self-driving, online car-hailing, taxi, bus, subway. The system aims to exhaust all possible transfer options at this stage, providing a basis for subsequent quantitative evaluation.

[0027] S3, collect and fuse dynamic influence parameters, special auxiliary data related to each potential travel mode, and static geographic information from the departure place to the airport to evaluate the potential travel modes corresponding to each candidate flight, and obtain the evaluation results of each potential travel mode.

[0028] The specific process of this step includes the following steps: Step one, build a multi-dimensional feature vector: for each potential travel mode identified (such as self-driving, subway), the system performs multi-source data fusion: (1) Collect dynamic influence parameters: by calling real-time traffic data interfaces and weather data interfaces publicly available on the web, obtain current and predicted traffic information (including congestion level, estimated travel time), which integrates weather influence factors and time period characteristics (such as morning and evening peak hours); Collect static geographic information: through the map service interface publicly available on the web, obtain the path distance from the departure place to the airport, the geographic location information of the key nodes (such as highway entrances, transportation hubs) and public transportation stations (bus stops, subway stations); Collect special auxiliary data: according to the type of travel mode, additional parameters are obtained, and additional parameters are different for different travel modes, for example: if it is a self-driving mode: call the parking lot data interface publicly available on the web to obtain the real-time vacancy number of parking lots around the target airport, parking fees and walking distance to the terminal; The collection of this additional parameter can help users judge whether it is convenient to park, the related parking fee and the possible time needed to reach the terminal from the parking point when using self-driving mode to go to the airport, so as to help users judge whether to use this mode to go to the airport; If it is a bus or subway mode: calculate the walking distance and estimated walking time from the departure place to the nearest bus stop or subway station, and obtain the real-time interval of the line and the operation schedule, the collection of this additional parameter can help users judge the time needed to go to the airport by subway or bus.

[0029] (2) Align all collected data with the current evaluation of the departure place-airport path and the user's planned travel time window. For example, map the weather prediction data to the geographical area of the path and the planned travel period, ensuring that all data reflect the same space-time scenario.

[0030] (3) Extract quantitative features from real-time / predicted data: For example, extract "estimated travel time (minutes)" and "congestion delay index (0-1)" from traffic data; extract "weather impact coefficient" (mapped to a numerical value based on precipitation, visibility, etc.) from weather data.

[0031] Extract features that are constant or slowly changing from geographic data: For example, "total path distance (km)", "number of required transfers", "walking distance from origin to subway station (m)".

[0032] Extract specific features according to mode of transportation type: For example, extract "parking lot fee (yuan)" for "self-driving" mode; extract "next train waiting time (minutes)" for "subway" mode; extract "current dynamic premium rate" for "ride-hailing" mode.

[0033] (4) Data cleaning and normalization First, handle missing values and outliers (such as unreasonable negative times or extremely high fees).

[0034] Next, map feature values of different dimensions and ranges to a unified numerical interval (such as [0, 1] or conform to standard normal distribution) through linear scaling (such as Min-Max), Z-score standardization, etc. This is a key step to achieve comparability of different dimensional features and enable subsequent rule-based judgments. For example, normalize "fee (0-200 yuan)" and "time (0-120 minutes)" to the [0, 1] interval.

[0035] (5) According to the pre-set evaluation dimensions (efficiency, cost, convenience, reliability), select the key features that best represent each dimension from all extracted and processed features. For example, "efficiency" dimension features may include "normalized estimated time" and "congestion delay index".

[0036] Then, arrange the selected feature values in order according to the fixed dimension sequence to form a one-dimensional array, i.e., a multi-dimensional feature vector.

[0037] The above data is cleaned, normalized, and fused to form a multi-dimensional feature vector representing the overall state of the potential mode of transportation.

[0038] Step 2: Quantitative evaluation based on priority configuration The system initializes a default priority configuration that explicitly sets the relative importance of efficiency (such as time), cost (such as fee), convenience (such as walking distance), and reliability (such as delay risk) in comprehensive judgment in the form of rules. For example, the default configuration may give "efficiency" and "reliability" higher base weights, and set "cost" and "convenience" as medium.

[0039] Specifically, the system is built-in a hierarchical rule base, containing the following two types of core rules: (1) Dimensional filtering rules: define the basic criteria for "qualified" and "unqualified" for each evaluation dimension (efficiency, cost, convenience, reliability). Each rule independently acts on a specific component of the multi-dimensional feature vector.

[0040] For example: In the efficiency dimension, if the estimated travel time > baseline time threshold, it is marked as "inefficient"; if the travel time fluctuation range > acceptable fluctuation threshold, it is marked as "low time reliability".

[0041] In the cost dimension, if the estimated total cost > budget threshold, it is marked as "cost overruns".

[0042] In the reliability dimension, if the current weather warning level ≥ severe weather threshold, it is marked as "high risk weather"; if the current traffic congestion index > severe congestion threshold, it is marked as "high delay risk".

[0043] (2) Comprehensive decision rules: define how to make the final evaluation based on the results of dimensional filtering. These rules explicitly define the priority relationship between different dimensions.

[0044] Rule A (safety / reliability veto): if there is "high risk weather" or "high delay risk", the final evaluation result is "not recommended"; Rule B (cost and efficiency trade-off): if marked as "cost overruns" and not marked as "inefficient", the final evaluation result is "average" (unless the user shows preference for cost); Rule C (efficiency first): if marked as "inefficient" and not triggered by Rule A, the final evaluation result is "not recommended"; Rule D (convenience supplement): if there is no negative marking and the number of transfers is 0, the final evaluation result is "strongly recommended"; Rule E (default pass): if none of the above rules are triggered, the final evaluation result is "recommended".

[0045] In this embodiment, user preference data directly affects the execution logic of the rule base, realizing personalized evaluation. The system obtains user preference data through two channels: one is the explicit channel, which directly receives the configuration information submitted by the user in the map interaction interface (such as setting "lowest cost" as the priority option), which is more suitable for first-time login users; the other is the implicit channel, which intelligently generates default configurations representing the long-term behavior tendency of the user by analyzing the user's historical order and query records (such as frequent selection of early flights can infer that the user is a "time-sensitive" user), which is more suitable for users with multiple flight experiences and multiple uses of the system. This design also enables the system to have learning and adaptation capabilities, allowing it to automatically update default configuration information.

[0046] It is worth noting that the source of the above "user historical order and query record" is authorized by the user.

[0047] This link is based on the above-mentioned dimension filtering rules and comprehensive decision rules, and the specific operation process is as follows: (1) The system generates a modification instruction to enable or disable part of the comprehensive decision rules according to the user preference parameters parsed.

[0048] For example: when the system determines that the user is "extremely cost-sensitive", it disables rule C (efficiency first) and significantly strengthens the conclusion of rule B (cost and efficiency trade-off), that is, even if the efficiency is not qualified, as long as the cost is extremely low, it may still obtain a "recommended" evaluation.

[0049] (2) Dynamically adjust the rule threshold: the various thresholds in the dimension filtering rules (such as "baseline time threshold" and "budget threshold") are individually set according to user preference data.

[0050] For example: for "time-sensitive" users, the system will automatically generate a modification instruction to lower the "baseline time threshold" of the efficiency dimension, making the same travel time more likely to trigger the "inefficient" label, and thus more likely to be excluded in subsequent decision-making.

[0051] (3) Adjust the rule execution order: the application order of the comprehensive decision rules can be adjusted according to the preferences in the modification instruction. For example, for "convenience first" users, the execution order of the decision rules related to convenience (such as rule D) is advanced.

[0052] Now, in combination with the above two links, the evaluation process performed in S3 for each potential travel mode to be evaluated is described as follows: ① Feature extraction and input: convert the multi-dimensional feature vector formed into "fact" data that can be recognized by the rule engine.

[0053] ② Execute dimension filtering rules: the rule engine traverses all dimension filtering rules, matches the input facts, and outputs a set of "tags" (e.g. "cost over budget", "high risk of delay").

[0054] ③ Load personalized rule set: according to the current user's preference data, load the matching comprehensive decision rule set from the rule library, adjusted by state and parameters.

[0055] ④ Execute comprehensive decision: the rule engine reasons the "tag" set generated in step ② according to the loaded rule set. The rules are triggered in sequence until a final evaluation result (e.g. "strongly recommended", "recommended", "average", "not recommended") is generated.

[0056] ⑤ Result quantization mapping (optional): for the convenience of subsequent comprehensive calculation with flight score, the system internally maps the qualitative evaluation result (e.g. "strongly recommended") output by the rule into a score within a standardized numerical value interval.

[0057] The core of this evaluation process is to convert "priority configuration" into a configurable and adjustable rule system. By obtaining user preference data, the rule activation state, decision threshold and execution logic in this rule system are dynamically modified, so that the same set of rule engines can execute completely different evaluation strategies for different users, realizing personalized evaluation based on business logic without relying on numerical weights, highly transparent and interpretable. The recommended results are more in line with the real needs and value orientation of users, providing "thousand faces" customized service experience.

[0058] S4, generate a comprehensive recommendation value for each candidate flight based on the flight attribute data and the evaluation results of each candidate flight, and output a flight recommendation scheme according to the comprehensive recommendation value.

[0059] The specific process of this step is as follows: First, the calculation of flight attribute comprehensive score: for each candidate flight, the system real-time obtains its flight attribute data (such as ticket price, total travel time, number of transfers, historical punctuality rate, seat availability) and associated weather data (weather forecast of the departure and arrival airports before and after the flight plan time).

[0060] Based on the above data, the system performs quantitative calculation on three preset dimensions of economy (mainly associated with ticket price and total time), reliability (mainly associated with punctuality rate, number of transfers and weather influence), and experience (mainly associated with service evaluation and seat availability), to generate preliminary dimension item scores.

[0061] Meanwhile, the system will automatically obtain and analyze the user's personal flight preference parameters (such as "price sensitive", "extremely averse to transfer", "prefer a specific airline") from the user's historical flight records. For users who are flying for the first time, the system can obtain the personal flight preference parameters by manually inputting them through the interactive interface. These parameters are not simply filtered, but are used to dynamically adjust the way the three dimensions are considered when the final synthesis is made. For example, for "extremely price-sensitive" users, the system will significantly increase the weight of the economy dimension when synthesizing, and even tolerate a slight decrease in reliability to some extent. Finally, a personalized flight attribute synthesis score that combines objective data and user subjective preferences is calculated for each candidate flight.

[0062] Next, the comprehensive recommendation value is generated: for each candidate flight, the personalized flight attribute synthesis score generated above is analyzed overall with the evaluation results of all potential ground travel modes associated with the departure airport of the flight: the system performs in-depth analysis based on a pre-set decision logic. The decision logic defines the complex matching and mutual exclusion relationship between flight attributes and ground connection modes. For example: (1) Complementary reinforcement logic: a flight score is high, but the only feasible ground connection mode is evaluated as "high delay risk", the decision logic may determine that the overall scheme has shortcomings, and thus adjust the comprehensive recommendation value downward.

[0063] (2) Bottleneck arbitration logic: if a flight score is average, but there is a connection mode evaluated as "high efficiency and low cost" (such as a direct subway), the decision logic may determine that "the ground segment advantage significantly makes up for the air segment deficiency", and thus adjust the comprehensive recommendation value upward.

[0064] (3) Risk aversion logic: for early flights, the decision logic will more strictly examine the "reliability" evaluation of the connection mode, and unreliable connection modes will cause the comprehensive recommendation value of the early flight to be significantly reduced.

[0065] Through the design of the above decision logic, a comprehensive recommendation value representing the overall pros and cons of the "flight + connection" combination can be output. This value quantifies the expected experience of the complete travel scheme from the user's actual departure location to the final destination.

[0066] Finally, the system sorts the comprehensive recommendation values calculated for all candidate flights, automatically filters the top-ranked schemes as the optimal solution set. These schemes are added with highlighted identifiers (such as "intelligent first recommendation", "best connection") as the final flight recommendation scheme output to the user.

[0067] The application effectively avoids the decision-making pitfalls of "flight on time but missed connection" or "cheap flight but expensive and complicated connection" by taking the reliability and time cost of ground connection before flight into overall evaluation, and ensures that the recommended "door-to-door" scheme has better real feasibility and overall superiority.

[0068] S5, ticketing platform booking interface information corresponding to the flight recommendation scheme is acquired; a user interface containing an interactive booking control is generated and displayed; in response to a triggering operation on the interactive booking control, the corresponding ticketing platform booking interface is called and a ticket purchase page is jumped to.

[0069] The specific process of this step is as follows: When the system generates and filters the final flight recommendation scheme according to step S4, the corresponding ticketing platform booking interface information is parsed and associated for each recommendation scheme. This information is not a simple web link, but a structured calling credential, which usually includes: the application programming interface address of the target ticketing platform, the session identifier or offer code that uniquely identifies this query or cabin price in the target platform, and the necessary identity verification parameters (such as encrypted partner token). These information has been pre-extracted and encapsulated from the metadata or response of the ticketing interface in the data aggregation stage (corresponding to the data aggregation module of the system), and is bound with the flight data as an attached information when generating the recommendation scheme.

[0070] At the same time, during the output of the system, the interface rendering engine dynamically generates and embeds an interactive booking control (for example, the "immediate booking" or "view details" button at the bottom of each recommendation scheme card). The generation logic of this control is: ① state binding: each control has been strongly associated with the ticketing platform booking interface information corresponding to its own recommendation scheme in the internal logic; ② context awareness: the visual style (such as color, text) of the control may be dynamically adjusted according to the attributes of the recommendation scheme (such as whether it is "smart first choice") or the member status of the user, in order to improve the guiding effect.

[0071] Finally, the system listens to the triggering operation (such as clicking, touching) of the user on the above-mentioned "interactive booking control" in real time through the front-end interaction logic. Once triggered, the system will perform a secure calling process. It will extract the interface information bound to the control and organize it into an interface calling request that meets the specifications of the target ticketing platform. The key point is that this request will pass the identifier containing the complete context of the specific flight selection, cabin, price, etc. to the target platform, ensuring that the user sees the booking page consistent with the recommendation scheme after jumping, rather than a blank page that needs to be searched again. After the calling is successful, the system guides the user to jump to the corresponding ticket purchase page of the target ticketing platform according to the response.

[0072] The application further provides a flight travel recommendation system, comprising a demand acquisition interaction module, a flight connection scheme generation module, a connection scheme evaluation module, a comprehensive recommendation generation module and a display output module.

[0073] The demand acquisition interaction module has a map interaction unit, a geographic point selection capture unit and a structured packaging unit, the map interaction unit is responsible for loading and rendering a map interface with geographic information visualization capability, and provides basic interaction functions such as zooming and dragging; the geographic point selection capture unit listens to and responds to user point selection, drawing or search operations on the map interaction interface, and analyzes user actions into accurate geographic position coordinates; the structured packaging unit receives position information from the interaction unit and user-supplemented travel date parameters, integrates, verifies and packages them into a machine-readable structured travel demand object.

[0074] The flight connection scheme generation module has an interface scheduling unit, a data analysis processing unit and a connection scheme identification unit, the interface scheduling unit is used to coordinate and call external ticket data interfaces according to the structured travel demand, request and receive raw data of candidate flight sets; the data analysis processing unit is used to clean and analyze the obtained raw flight data, extract key attributes (such as flight number, departure and arrival airport, time) and convert them into an internal unified data format; the connection scheme identification unit is used to call a map or traffic planning service interface for each candidate flight departure airport in combination with the user's actual departure coordinates, automatically discover and list all feasible ground traffic connection modes, and form an initial association pair of flight and connection mode.

[0075] The connection scheme evaluation module has a multi-source data fusion unit, a feature extraction unit, an evaluation unit and a personalized rule adjustment unit, the multi-source data fusion unit is used to collect real-time traffic status, static path information and mode-specific parameters for each discovered potential travel mode in parallel, and perform time and space alignment and fusion processing; the feature extraction unit is used to extract quantitative features reflecting four dimensions of efficiency, cost, convenience and reliability from the fused data, and construct them into a standardized multi-dimensional feature vector; the evaluation unit is embedded with a configurable rule engine, which loads a preset priority configuration (reflected as a business rule set), takes the feature vector as an input fact, executes rule reasoning, and outputs a qualitative evaluation result of each travel mode; the personalized rule adjustment unit is used to dynamically adjust the activation state, threshold parameter or execution priority of related rules in the rule engine according to the obtained user travel preference data, to realize personalized adaptation of the evaluation standard.

[0076] The comprehensive recommendation generation module has a flight multi-dimensional scoring unit, a collaborative analysis unit, a comprehensive recommendation value calculation unit and a scheme generation unit. The flight multi-dimensional scoring unit is used for obtaining flight attribute data and associated meteorological data of each candidate flight, and referring to user personal flight preferences, quantitatively scoring in multiple dimensions such as economy, reliability, experience, etc., and finally fusing into personalized flight attribute comprehensive score. The collaborative analysis unit is used for receiving flight attribute comprehensive score and corresponding ground connection scheme evaluation result set. The unit is built-in with preset collaborative decision logic (such as complementary reinforcement, bottleneck arbitration, risk avoidance, etc.), which performs integrated analysis on the "flight-connection" combination and evaluates its overall advantages and disadvantages. The comprehensive recommendation value calculation unit is used for calculating a unified comprehensive recommendation value for each candidate flight (and its associated best or typical connection scheme) based on the results of collaborative decision analysis, and sorting all candidate schemes in descending order according to the value to form a sorted list. The scheme generation unit is used to filter the top-ranked schemes from the sorted list, add a recommendation identifier to them, and assemble them into a final flight recommendation scheme list.

[0077] The display output module has a visual rendering unit, an interface binding unit and a response and jump unit. The visual rendering unit is used for visual rendering of the final flight recommendation scheme list in a user-friendly form (such as card, list), and highlighting the recommendation identifier and key information. The interface binding unit is used for dynamically generating "booking" or "view details" interactive controls for each recommendation scheme, and securely binding the control with the specific booking interface information of the corresponding ticketing platform. The response and jump unit is used to listen to the triggering event of the interactive control by the user. Once triggered, the bound ticketing interface is extracted and called, carrying complete context information (such as flight selection, offer code), seamlessly guiding the user to the corresponding ticket purchase page of the cooperative ticketing platform.

[0078] The above is only the preferred specific embodiment of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can make equivalent replacement or change according to the technical solution and inventive concept of the present application within the technical range disclosed by the present application, which should be covered within the protection scope of the present application.

Claims

1. A flight trip recommendation method, characterized in that, The method comprises the following steps: S1, obtaining the travel demand information of the user, the travel demand information comprising the departure place, the destination and the travel date; S2, obtaining a candidate flight set comprising a plurality of candidate flights according to the travel demand information of the user, and identifying and determining at least one feasible potential travel mode based on the geographical position of the departure place and the airport corresponding to each candidate flight; S3, collecting and fusing the dynamic influence parameters, special auxiliary data and static geographical information from the departure place to the airport related to each potential travel mode to evaluate the potential travel mode corresponding to each candidate flight, and obtaining the evaluation result of each potential travel mode; S4, generating the comprehensive recommendation value of each candidate flight in combination with the flight attribute data of each candidate flight and the evaluation result, and outputting the flight recommendation scheme according to the comprehensive recommendation value. 2.The flight travel recommendation method of claim 1, wherein, In S1, the obtaining of the travel demand information of the user specifically comprises: receiving the geographical point selection operation of the user on the departure place and the destination through a map interactive interface to generate initial travel demand information comprising the departure place and the destination; receiving the travel date input by the user on the map interactive interface, and merging the initial travel demand information to form complete travel demand information. 3.The flight travel recommendation method of claim 1, wherein, S2 specifically comprises: performing format standardization processing on the travel demand information to generate structured request data; calling a ticket service interface to obtain the corresponding candidate flight set based on the structured request data; calling a map service interface or a traffic planning service to obtain a plurality of potential travel modes from the departure place to the airport in combination with the user departure place information for the geographical position of the departure airport corresponding to each candidate flight in the candidate flight set; the potential travel modes include but are not limited to self-driving, online car-hailing, taxi, bus and subway. 4.The flight travel recommendation method of claim 1, wherein, S3 specifically comprises: for each potential travel mode, collecting and fusing the dynamic influence parameters, static geographical information and special auxiliary data to form a multi-dimensional feature vector of the potential travel mode, wherein the dynamic influence parameters are current and predicted traffic information obtained by calling a real-time traffic data interface and a weather data interface, the static geographical information is path information from the departure place to the airport obtained by a map service interface, and the special auxiliary data is additional parameters obtained according to the type of travel mode; according to a preset priority configuration, the feature components representing the efficiency, cost, convenience and reliability dimensions in the multi-dimensional feature vector are evaluated; based on the importance relationship defined by the priority configuration, the evaluation results of each dimension are comprehensively generated to obtain the final evaluation result of each potential travel mode. 5.The flight travel recommendation method of claim 4, wherein, The preset priority configuration can be dynamically adjusted according to the travel preference data of the user, specifically: obtaining the travel preference data of the user; analyzing the travel preference data to extract the preference parameters for quantitatively representing the attention degree of the user on the efficiency, cost, convenience and reliability dimensions; based on the preference parameters, a group of correction instructions for adjusting the importance relationship of each dimension are generated; according to the correction instructions, the importance relationship of each dimension defined in the priority configuration is updated. 6.The flight travel recommendation method of claim 5, wherein, The travel preference data is derived from at least one of the following: configuration information actively submitted by the user through a system interface or default configuration information generated by analyzing historical orders and query records of the user.

7. The flight travel recommendation method of claim 1, wherein, The S4 specifically includes: S41, for each candidate flight in the candidate flight set, obtaining its flight attribute data and associated meteorological data, wherein the flight attribute data at least includes ticket price, total travel time, number of transfers, historical punctuality rate and seat availability, and the associated meteorological data at least includes weather forecast information of the departure airport and the destination airport within a preset time period before and after the planned takeoff time of the flight; S42, based on the obtained flight attribute data and associated meteorological data, comprehensively considering the economy, reliability and experience dimensions, the flight attribute comprehensive score of the candidate flight is calculated; S43, obtaining the evaluation results of all potential travel modes corresponding to the candidate flight; S44, overall analysis of the flight attribute comprehensive score and the evaluation results, judging the best matching relationship between the flight and the ground connection mode before flight according to the preset decision logic, and generating a comprehensive recommendation value representing the overall advantages and disadvantages of the flight and its connected travel mode; S45, sorting all candidate flights according to the comprehensive recommendation value, selecting the top ranked scheme and adding a highlight display mark as the flight recommendation scheme output. 8.The flight travel recommendation method of claim 7, wherein, The calculation of the flight attribute comprehensive score can be dynamically adjusted according to the user's personal flight preference parameters, specifically: Obtaining the user's personal flight preference parameters; analyzing the personal flight preference parameters to extract personalized tendency information representing the user's preferences in economy, reliability and experience dimensions; adjusting the comprehensive consideration of the flight attribute data and associated meteorological data in the economy, reliability and experience dimensions according to the personalized tendency information; Based on the adjusted comprehensive consideration, the flight attribute comprehensive score is calculated, which is fused with the user's personalized tendency. 9.The flight travel recommendation method of claim 1, wherein, After S4, S5 is also included, which specifically includes: obtaining ticketing platform reservation interface information corresponding to the flight recommendation scheme; generating and displaying a user interface containing an interactive booking control; in response to the triggering operation of the interactive booking control, calling the corresponding ticketing platform reservation interface and jumping to the ticket purchase page. 10.A flight trip recommendation system, characterized in that, It includes: A demand acquisition interaction module for receiving the departure location, destination and travel date input by the user through an interactive map; A flight connection scheme generation module for obtaining a candidate flight set according to the demand and identifying at least one potential travel mode to the departure airport corresponding to each candidate flight; A connection scheme evaluation module for each potential travel mode, fusing multi-source data to form feature information, performing multi-dimensional evaluation based on preset priority configuration and fusing the results, and outputting the evaluation results of each potential travel mode; A comprehensive recommendation generation module for calculating each flight attribute comprehensive score based on flight attribute data, associated meteorological data and user preference data, combining the evaluation results to analyze and calculate according to the preset decision logic, generating a comprehensive recommendation value for each flight and outputting a flight recommendation scheme; The display output module is configured to visualize the flight recommendation scheme and provide a jump interface to a ticketing platform.

Images