A Method and System for Precision Marketing Recommendation in Scenic Spot Ticket Management by Integrating User Behavior Graphs

By acquiring multi-dimensional behavioral data and user interaction data from scenic spot ticket digital platforms, and combining this with user behavior graphs, we can quantify real-time inventory pressure, scenic spot preferences, consumption sensitivity, and the stability of user mobile behavior to conduct precise marketing recommendations. This solves the problems of resource mismatch and excessive costs in existing strategies and achieves dynamic matching between user demand and scenic spot inventory.

CN121746039BActive Publication Date: 2026-11-17浙江携途玩乐旅游控股集团有限公司
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
CN202511966440.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2025-12-24
Publication Date
2026-11-17
Estimated Expiration
2045-12-24

AI Technical Summary

Technical Problem

Existing strategies for recommending and operating scenic spot tickets have failed to effectively balance inventory reduction and marketing cost control, resulting in resource misallocation or excessive costs.

Method used

By acquiring multi-dimensional behavioral data from scenic spot ticket digital platforms, including platform basic data and user interaction data, and combining it with user behavior graphs, we can quantify real-time inventory pressure, scenic spot preferences, consumption sensitivity, and user mobile behavior stability to conduct precise marketing recommendations.

Benefits of technology

It enables dynamic matching of user demand and scenic area inventory, avoiding resource mismatch and excessive costs, and improving marketing conversion rate and user experience.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121746039B_ABST
    Figure CN121746039B_ABST
Patent Text Reader

Abstract

This invention relates to the field of scenic spot ticket technology, specifically disclosing a method and system for precise marketing and recommendation of scenic spot tickets through user behavior mapping. The invention first obtains multi-dimensional behavioral data, including platform basic data and user interaction data, through the backend database of the scenic spot ticket digital platform and the user behavior log system. This lays the foundation for accurately depicting the states of both parties. Then, based on this, a real-time inventory pressure coefficient is calculated to quantify the risks of unsold stock and overcrowding; scenic spot preference is calculated to reflect user interest in the scenic spot; consumption sensitivity is calculated based on the user behavior graph to reflect price sensitivity; and a mobility behavior stability coefficient is calculated in conjunction with user status data to determine the likelihood of travel. Finally, a comprehensive recommendation based on these four coefficients is made to balance the needs of the scenic spot and the user, such as recommending discounted tickets to users with high inventory, high preference, high stability, and price sensitivity, thus avoiding resource misallocation and excessive costs.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of scenic spot ticket technology, and in particular to a method and system for precise marketing recommendation of scenic spot ticket agency operations that integrates user behavior graphs. Background Technology

[0002] With the deepening of digital transformation in the cultural tourism industry, scenic spot ticket sales models are gradually shifting from traditional window ticketing to online agency operation models. Scenic spot ticket agency operation services, by integrating multi-channel traffic resources and optimizing the marketing chain, have become an important means to improve ticket sales and user experience. In this process, precise marketing recommendations are a core element, aiming to achieve personalized matching of ticket products based on user characteristics and scenic spot resource characteristics, thereby improving conversion rates, reducing customer acquisition costs, and optimizing scenic spot inventory turnover efficiency. Existing recommendation strategies often focus solely on conversion rates without considering operational goals such as inventory clearance and marketing cost control, which can easily lead to resource misallocation or excessive costs. Therefore, a precise marketing recommendation method for scenic spot ticket management that integrates user behavior mapping is needed to solve these problems. Summary of the Invention

[0003] The purpose of this invention is to provide a method and system for precise marketing recommendations for scenic spot ticket agency operations that integrates user behavior graphs, so as to solve the technical problems mentioned in the background art above.

[0004] To achieve the above objectives, the present invention provides the following technical solution: A method for precise marketing recommendations in scenic spot ticket operation that integrates user behavior graphs includes: Acquire multi-dimensional behavioral data from the scenic area's digital ticketing platform. This multi-dimensional behavioral data includes both platform basic data and user interaction data with the platform. The real-time inventory pressure coefficient is obtained based on the platform's basic data. The scenic area preference is obtained based on the user's interaction data with the platform. User behavior data and user status data are obtained based on user behavior graphs, and browsing behavior data and search behavior data are obtained based on the user behavior data, and consumption sensitivity is obtained based on the browsing behavior data and the search behavior data. Geographic mobility status data is obtained based on the user status data, and a user mobility behavior stability coefficient is obtained based on the geographic mobility status data. Based on the real-time inventory pressure coefficient, the scenic spot preference, the consumption sensitivity, and the user mobile behavior stability coefficient, precise marketing recommendations for ticket agency operations are made to users.

[0005] Preferably, the step of obtaining the real-time inventory pressure coefficient based on the platform's basic data includes: Obtain the current inventory and historical sales volume of the scenic area's digital platform to understand the basic inventory pressure; The first moment is obtained based on the current inventory level, and the time decay coefficient is obtained based on the first moment and the preset end moment. Obtain the difference between time-sharing reservation data and the scenic area's carrying capacity, and obtain the time period saturation based on the difference between the time-sharing reservation data and the scenic area's carrying capacity; The real-time inventory pressure coefficient is obtained based on the basic inventory pressure, the time decay coefficient, and the time period saturation.

[0006] Preferably, the step of obtaining scenic spot preference based on the user's interaction data with the platform includes: Based on the user's interaction data with the platform, we obtain the user's first dwell time, page scroll depth, and number of image clicks on the scenic spot details page on the platform. The browsing preference coefficient is obtained based on the first dwell time, the page scrolling depth, and the number of image clicks. Obtain the frequency of keyword repetition and the click-through rate of search results entered by the user, and obtain the search preference coefficient based on the frequency of keyword repetition and the click-through rate of search results; The scenic area preference degree is obtained based on the browsing preference coefficient and the search preference coefficient.

[0007] Preferably, the step of obtaining consumer sensitivity based on the browsing behavior data and the search behavior data includes: Based on the browsing behavior data, the user's second dwell time in the price section and the frequency of price filtering operations are obtained; The browsing behavior coefficient is obtained based on the second dwell time and the frequency of the price filtering operation; The frequency of comparison keywords and the number of historical low-price queries are obtained based on the search behavior data, and the search behavior coefficient is obtained based on the frequency of comparison keywords and the number of historical low-price queries. Consumer sensitivity is obtained based on the browsing behavior coefficient and the search behavior coefficient.

[0008] Preferably, the step of obtaining the user's mobility behavior stability coefficient based on the geographic mobility status data includes: The total length of repeated paths and the total length of all current paths are obtained based on the geographic movement status data. The path repetition rate is obtained based on the total length of the repeated paths and the total length of all current paths. The user's set of dwell points is obtained based on the geographic mobility status data, and multiple dwell clusters are obtained based on the K-means algorithm; Obtain the start time and end time of each dwell cluster, calculate multiple dwell durations based on multiple start times and multiple end times, and select the dwell duration with the longest duration as the main dwell duration; The stability coefficient of user mobile behavior is obtained based on the main dwell time and the path repetition rate.

[0009] Preferably, the step of providing precise marketing recommendations for ticket agency services based on the real-time inventory pressure coefficient, the scenic spot preference, the consumption sensitivity, and the user mobile behavior stability coefficient includes: The real-time inventory pressure coefficient, the scenic spot preference, the consumption sensitivity, and the user mobile behavior stability coefficient are normalized to obtain the real-time inventory pressure normalization coefficient, the scenic spot preference normalization coefficient, the consumption sensitivity normalization coefficient, and the user mobile behavior stability normalization coefficient. The user's comprehensive recommendation priority score is calculated by normalizing the real-time inventory pressure coefficient, the scenic spot preference coefficient, the consumption sensitivity coefficient, and the user's mobile behavior stability coefficient. Determine whether the user's overall recommendation priority score is within a preset range; If it is not within the preset range but is greater than the preset range, it is determined to be a high recommendation level, and a high-priority recommendation strategy including core scenic spots and discounted tickets is generated. If the information falls within a preset range, it is classified as a medium recommendation level, and a medium-priority recommendation strategy is generated that includes relevant scenic spots and regular ticket prices. If it is not within the preset range and is smaller than the preset range, it is judged as a low recommendation level, and a low-priority recommendation strategy containing potential interest scenic spots and promotional information is generated.

[0010] This application also provides a scenic spot ticket agency operation and precision marketing recommendation system that integrates user behavior graphs, including: The first acquisition module is used to acquire multi-dimensional behavioral data from the scenic spot ticket digital platform. The multi-dimensional behavioral data includes platform basic data and user interaction data with the platform. The second acquisition module is used to acquire the real-time inventory pressure coefficient based on the platform's basic data. The third acquisition module is used to acquire scenic spot preference based on the user's interaction data with the platform. The fourth acquisition module is used to acquire user behavior data and user status data based on the user behavior graph, and to acquire browsing behavior data and search behavior data based on the user behavior data, and to acquire consumption sensitivity based on the browsing behavior data and the search behavior data. The fifth acquisition module is used to acquire geographic mobility status data based on the user status data, and to acquire the user mobility behavior stability coefficient based on the geographic mobility status data. The recommendation module is used to make precise marketing recommendations for ticket agency operations to users based on the real-time inventory pressure coefficient, the scenic spot preference, the consumption sensitivity, and the user mobile behavior stability coefficient.

[0011] Preferably, the second acquisition module includes: The first acquisition unit is used to obtain the current inventory and historical sales volume of the scenic area's digital platform to obtain the basic inventory pressure. The second acquisition unit is used to acquire the corresponding first moment based on the current inventory level, and to acquire the time decay coefficient based on the first moment and the preset end moment. The third acquisition unit is used to acquire the difference between the time-sharing reservation data and the scenic area's carrying capacity, and to acquire the time period saturation based on the difference between the time-sharing reservation data and the scenic area's carrying capacity. The fourth acquisition unit is used to acquire the real-time inventory pressure coefficient based on the basic inventory pressure, the time decay coefficient, and the time period saturation.

[0012] This application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the above-described method.

[0013] This application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method.

[0014] The beneficial effects of this application are as follows: This invention first obtains multi-dimensional behavioral data, including platform basic data and user interaction data, through the background database of the scenic spot ticket digital platform and the user behavior log system, laying the foundation for accurately depicting the state of both parties. Then, based on this, the real-time inventory pressure coefficient is calculated to quantify the risk of unsold goods and overcrowding, the scenic spot preference is calculated to reflect the user's interest in the scenic spot, the consumption sensitivity is calculated based on the user behavior graph to reflect the degree of price sensitivity, and the mobility behavior stability coefficient is calculated in combination with the user status data to judge the probability of travel. Finally, the four coefficients are combined to make recommendations to balance the needs of the scenic spot and the user, such as pushing discounted tickets to users with high inventory, high preference, high stability and price sensitivity, so as to avoid resource misallocation and excessive costs. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of a method flow according to an embodiment of this application.

[0016] Figure 2 This is a schematic diagram of the system structure according to an embodiment of this application.

[0017] Figure 3 This is a schematic diagram of the internal structure of a computer device according to an embodiment of this application.

[0018] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0020] like Figures 1-3 As shown, this application provides a method for precise marketing recommendation of scenic spot ticket operation based on user behavior graph, including: S1. Obtain multi-dimensional behavioral data from the scenic spot ticket digital platform, including platform basic data and user interaction data with the platform; S2. Obtain the real-time inventory pressure coefficient based on the platform's basic data; S3. Obtain scenic area preference based on the user's interaction data with the platform; S4. Obtain user behavior data and user status data based on the user behavior graph, and obtain browsing behavior data and search behavior data based on the user behavior data, and obtain consumption sensitivity based on the browsing behavior data and the search behavior data; S5. Obtain geographic mobility status data based on the user status data, and obtain the user mobility behavior stability coefficient based on the geographic mobility status data. S6. Based on the real-time inventory pressure coefficient, the scenic spot preference, the consumption sensitivity, and the user mobile behavior stability coefficient, conduct precise marketing recommendations for ticket agency operations for users.

[0021] As described in steps S1-S6 above, existing recommendation strategies often solely pursue conversion rates without comprehensively considering operational goals such as inventory clearance and marketing cost control, easily leading to resource misallocation or excessive costs. This invention first acquires multi-dimensional behavioral data from the scenic area ticketing digital platform. This multi-dimensional behavioral data includes platform basic data and user interaction data. Data is collected through the platform's backend database and user behavior log system. This multi-dimensional behavioral data forms the basis for subsequent calculations; only by comprehensively acquiring the raw data from both the scenic area and the user side can the state of both parties be accurately depicted. For example, without acquiring "time-slot reservation data," it is impossible to determine whether a scenic area is crowded during a certain time period, potentially recommending users to go to a saturated time slot. Then, based on the platform's basic data, a real-time inventory pressure coefficient is obtained. This real-time inventory pressure coefficient quantifies the risk of unsold tickets and overcrowding during peak hours for a scenic spot. For example, if a scenic spot's current inventory is twice that of the same period in history (high basic pressure), and there are only 3 days left until the tickets expire (high time decay coefficient), while weekend reservations have reached 90% of its capacity (high time saturation), then the real-time inventory pressure coefficient is high, and priority should be given to recommending tickets for non-weekend periods to clear the inventory. Next, the user's attraction preference is obtained based on the user's interaction data with the platform. This attraction preference quantifies the user's level of interest in a specific attraction. For example, if a user spends 10 minutes on the "Forbidden City" details page (far exceeding the average of 2 minutes), scrolls to the bottom of the page (100% depth), clicks on 10 images, and repeatedly searches for "Forbidden City tickets" and clicks on the search results, it indicates a high level of preference for the Forbidden City, and Forbidden City tickets should be given priority in recommendations. Secondly, user behavior data and user status data are obtained based on the user behavior graph. Browsing behavior data and search behavior data are then obtained based on the user behavior data, and consumption sensitivity is obtained based on the browsing behavior data and search behavior data. The user behavior graph is a structured graph integrating user behavior records within the platform (such as browsing and searching) and outside the platform (such as behavior on other travel apps), as well as user status (such as geographical location). Thus, consumption sensitivity reflects the user's sensitivity to ticket prices. For example, if a user spends 5 minutes in the price section (far exceeding the average of 1 minute), repeatedly filters for tickets "under 100 yuan," and frequently searches for "lowest price for XX scenic spot," it indicates high consumption sensitivity. Recommendations should focus on discounted tickets or promotional activities to increase conversion rates. Subsequently, geographic movement status data is obtained based on the user status data, and a user movement behavior stability coefficient is obtained based on the geographic movement status data. The geographic movement status data includes the user's movement path (e.g., GPS trajectory) and stopover points (e.g., residence, workplace, frequently visited business districts) (obtained from the user's authorized geographic location information). This user movement behavior stability coefficient reflects the regularity of the user's movement and the likelihood of visiting the scenic spot. For example, if a user travels between their residence and the vicinity of scenic spot A every Saturday for the past month (path repetition rate 80%), and spends 5 hours near scenic spot A (main stopover duration), it indicates high movement stability and a high probability of visiting scenic spot A in the near future; therefore, the priority for recommending tickets to scenic spot A should be increased. Finally, based on the real-time inventory pressure coefficient, the scenic spot preference, the consumption sensitivity, and the user's mobile behavior stability coefficient, precise marketing recommendations for ticket operation are made to users. This comprehensive approach, using these four dimensions, achieves a balance between "scenic spot demand" and "user demand." For example, if a scenic spot has a high real-time inventory pressure coefficient (needs to be cleared), and users have high preference for it, high mobile stability (likely to visit), and high consumption sensitivity (price-conscious), then the overall score is high, and discounted tickets for that scenic spot are recommended. This satisfies users' demand for low prices while helping the scenic spot clear its inventory, minimizing resource misallocation or excessive costs.

[0022] In one embodiment, step S2, which involves obtaining the real-time inventory pressure coefficient based on the platform's basic data, includes: S201. Obtain the current inventory and historical sales volume of the scenic area's digital platform to understand the basic inventory pressure; S202. Obtain the corresponding first moment based on the current inventory level, and obtain the time decay coefficient based on the first moment and the preset end moment; S203. Obtain the difference between the time-sharing reservation data and the scenic area's carrying capacity, and obtain the time period saturation based on the difference between the time-sharing reservation data and the scenic area's carrying capacity. S204. Obtain the real-time inventory pressure coefficient based on the basic inventory pressure, the time decay coefficient, and the time period saturation.

[0023] As described in steps S201-S204 above, this invention first obtains the current inventory and historical sales figures from the scenic area's digital platform to determine the basic inventory pressure. The data sources are: current inventory from the platform's inventory management module (e.g., the system's real-time record of remaining tickets for a scenic area, such as "800 tickets remaining for the Forbidden City on July 15th"); and historical sales figures from the platform's historical sales database (e.g., "1200 tickets sold for the Forbidden City on July 15th, 2023"). The basic inventory pressure is calculated as: current inventory ÷ historical sales figures (the higher the ratio, the higher the current inventory relative to historical demand, and the greater the basic pressure). This, combined with the historical sales baseline, determines whether the current inventory is "too high." For example, if the current inventory is 800 tickets and historical sales are 1200 tickets, the ratio is approximately 0.67 (low basic pressure, because the current inventory is less than historical sales, and it is highly likely to sell out). If the sales volume in the same period in history was only 300 units, the ratio is approximately 2.67 (the basic pressure is high, the current inventory is far higher than the historical demand, and there is a risk of unsold inventory). At the same time, it can avoid judging the inventory pressure solely by the absolute quantity and solve the problem that "the meaning of inventory is different in different seasons / time periods". Then, based on the current inventory level, the corresponding first moment is obtained, and the time decay coefficient is obtained based on the first moment and the preset end moment. The data sources are: the first moment is the start time of the ticket validity period corresponding to the current inventory (e.g., for "tickets valid on July 15th," the first moment is July 15th 00:00); the preset end moment is the end time of the ticket validity period (e.g., July 15th 24:00, valid for that day) or a longer period (e.g., for tickets "valid from July 1st to July 31st," the preset end moment is July 31st 24:00). The data comes from the platform's ticket rule setting module, and the time decay coefficient = (preset end moment - current time) ÷ (preset end moment - first moment). The coefficient ranges from [0,1]. The closer to the preset end moment, the larger the coefficient (e.g., if the current time is only 1 hour away from the end moment, the coefficient ≈ 1; just past the first moment, the coefficient ≈ 0). This quantifies the "time urgency" of the inventory. For example: A ticket is valid from July 1st to July 10th, and the current time is July 9th (1 day left until the end). The time decay coefficient = 1 ÷ 10 ≈ 0.1 (assuming a total duration of 10 days). This indicates that the stock is nearing its expiration date and is highly urgent. If the current time is July 2nd (8 days left until the end), the coefficient = 8 ÷ 10 = 0.8, indicating low urgency. At the same time, it can prioritize identifying inventory nearing its expiration date to avoid resource waste due to expiration. Next, the difference between the time-sharing reservation data and the scenic area's carrying capacity is obtained, and the time-saturation is calculated based on this difference. The data sources are: time-sharing reservation data from the platform's reservation system (e.g., "300 people have already booked the Palace Museum for 9:00-11:00 on July 15th"); and scenic area carrying capacity from the maximum capacity for each time period as approved by the scenic area management department (e.g., "The Palace Museum's carrying capacity for 9:00-11:00 is 500 people"). The time-saturation is calculated as: time-sharing reservation data ÷ scenic area carrying capacity (the larger the ratio, the closer the number of people who have booked for that time period is to the upper limit, and the higher the saturation). This reflects the "congestion risk" of a specific time period and avoids recommending tickets to saturated time periods, ensuring the visitor experience. It also guides users to choose time periods with insufficient reservations to reduce inventory. Finally, the real-time inventory pressure coefficient is obtained based on the basic inventory pressure, the time decay coefficient, and the time period saturation. A weighted summation is used to integrate the three indicators (the weights can be adjusted according to the scenic area's operational goals; for example, basic inventory pressure accounts for 40%, time decay coefficient for 30%, and (1-time period saturation) for 30%, where "1-time period saturation" quantifies the remaining capacity pressure of a time period; a larger value indicates that the time period needs more recommendations). For example: a scenic area currently has 800 tickets in stock, and historical sales during the same period were 300 (basic inventory pressure = 800 ÷ 300 ≈ 2.67); there is 1 day left until the expiration date (time decay coefficient = 0.1); the time period saturation is 0.6 (1-0.6 = 0.4). The real-time inventory pressure coefficient is calculated as follows: 2.67 × 0.4 + 0.1 × 0.3 + 0.4 × 0.3 ≈ 1.068 + 0.03 + 0.12 = 1.218 (the higher the value, the greater the real-time pressure). This coefficient, combining "quantity pressure," "time urgency," and "time slot capacity," provides a quantitative value reflecting the overall urgency of the current inventory. For example, a coefficient of 1.218 indicates that the ticket inventory is high, nearing expiration, and there is still capacity available for the corresponding time slot. Therefore, tickets for that time slot should be prioritized to clear the inventory. This also provides a clear priority basis for subsequent marketing recommendations (high-coefficient inventory is recommended first), achieving a dynamic match between inventory and demand.

[0024] In one embodiment, step S3, which involves obtaining scenic spot preference based on the user-platform interaction data, includes: S301. Based on the user-platform interaction data, obtain the user's first dwell time, page scrolling depth, and number of image clicks on the scenic spot details page on the platform; S302. Obtain a browsing preference coefficient based on the first dwell time, the page scrolling depth, and the number of image clicks; S303. Obtain the keyword repetition frequency and search result click rate input by the user, and obtain the search preference coefficient based on the keyword repetition frequency and the search result click rate; S304. Obtain the scenic area preference degree based on the browsing preference coefficient and the search preference coefficient.

[0025] As described in steps S301-S304 above, this invention first obtains the user's first dwell time, page scrolling depth, and image click count on the scenic spot details page based on the user-platform interaction data. The user-platform interaction data comes from the platform's front-end behavior tracking system and log recording module. Specifically, the first dwell time is the time from when the user enters the scenic spot details page to when they leave (e.g., "the user stayed for 7 minutes after entering the 'Summer Palace' details page"); the page scrolling depth is the proportion of the distance the user scrolls within the page to the total page length (e.g., if the total page length is 1000 pixels, and the user scrolls to the 800-pixel position, the depth is 80%); and the image click count is the number of times the user clicks on scenic spot images, guide maps, and other image elements within the details page (e.g., clicking the "Summer Palace Seventeen-Arch Bridge" image 5 times). This allows for the extraction of core details reflecting user interest in the browsing behavior. A long dwell time indicates that the user is willing to spend time learning about the scenic spot; a high scrolling depth indicates that the user is interested in more information about the scenic spot (e.g., not only looking at the homepage but also at guides, transportation, etc.); and a high number of image clicks indicates that the user is interested in the visual content of the scenic spot (e.g., the appearance and environment of the scenic spot). For example, users who spend 12 minutes, scroll 90% of the way, and click on 10 pictures on the "Zhangjiajie" details page are significantly more interested in Zhangjiajie than users who spend 1 minute, scroll 10% of the way, and click on 0 pictures. Then, a browsing preference coefficient is obtained based on the first dwell time, the page scroll depth, and the number of image clicks. A weighted summation formula is used to quantify browsing preference, with weights set according to the intensity of the behavior's indication of "interest" (e.g., first dwell time weight 0.4, page scroll depth weight 0.3, image click count weight 0.3). Example formula: Browsing preference coefficient = (first dwell time ÷ industry average dwell time) × 0.4 + page scroll depth × 0.3 + (image click count ÷ total number of images on the page) × 0.3. All indicators are normalized to the [0,1] range to avoid the influence of dimensional differences. This transforms scattered browsing behaviors into a single quantifiable value, intuitively reflecting the intensity of user interest during browsing. It also avoids misjudging preferences due to a single browsing behavior, and more accurately captures the user's true interests during browsing through multi-dimensional weighting. Next, the frequency of keyword repetition and the click-through rate (CTR) of search results are obtained from the user's input. A search preference coefficient is then calculated based on these metrics. Keyword repetition frequency is derived from the platform's search logs (e.g., a user searches for "West Lake tickets" 3 times in one day). The search result CTR is the percentage of times a user clicks on a specific scenic spot in the search results out of the total number of searches (e.g., a user searches for "Hangzhou scenic spots" 5 times, and clicks on "West Lake" 4 times → CTR 80%). Then, a weighted summation formula is used to quantify search preference, with weights emphasizing the expression of proactive demand (e.g., keyword repetition frequency weight 0.5, search result CTR weight 0.5). Example formula: Search Preference Coefficient = (Keyword Repetition Frequency ÷ Maximum Historical Repetition Frequency) × 0.5 + Search Result CTR × 0.5. The maximum historical repetition frequency is the maximum number of searches recorded by the platform, such as a user searching for a keyword a maximum of 5 times in one day, used for normalization. This converts proactive search behavior into a quantifiable value, reflecting the intensity of the user's proactive demand for scenic spots. High keyword repetition frequency indicates that users are continuously paying attention to the scenic spot; high search result click rate indicates that users recognize the scenic spot in the search results and have a high degree of matching with their needs. For example, User C: Searched for "Huangshan" 4 times in 1 day (the maximum historical number was 5 times → 0.8), and the probability of clicking on the Huangshan result was 100% (1), with a coefficient of 0.8×0.5+1×0.5=0.4+0.5=0.9 (high search preference). User D: Searched for "Taishan" once (0.2), and the probability of clicking on the Taishan result was 20% (0.2), with a coefficient of 0.2×0.5+0.2×0.5=0.1+0.1=0.2 (low search preference). Thus, the preference judgment is strengthened through active search behavior, which makes up for the possible accidental defects of passive browsing. Finally, the scenic area preference score is obtained based on the browsing preference coefficient and the search preference coefficient. A weighted average is used to integrate the two coefficients (the weights can be adjusted according to the business scenario; for example, 0.6 is set when browsing behavior better reflects potential interest, and 0.6 is set when search behavior better reflects explicit needs; in this example, each is 0.5). Example formula: Scenic Area Preference Score = Browsing Preference Coefficient × 0.5 + Search Preference Coefficient × 0.5. Example: User A's browsing preference coefficient is 0.79, and search preference coefficient is 0.9, then the scenic area preference score = 0.79 × 0.5 + 0.9 × 0.5 ≈ 0.395 + 0.45 = 0.845 (high preference); User B's browsing preference coefficient is 0.22, and search preference coefficient is 0.2, then the scenic area preference score = 0.22 × 0.5 + 0.2 × 0.5 = 0.11 + 0.1 = 0.21 (low preference). This combines the user's passive browsing and active search behavior signals to obtain a quantitative indicator reflecting the user's overall interest in a specific scenic area. The higher the value, the stronger the user's preference for the scenic spot, and the higher the priority should be given when making recommendations. At the same time, it can also achieve dual verification of "browsing behavior (potential interest) + search behavior (explicit needs)," which greatly improves the accuracy of user preference assessment and provides a reliable interest basis for subsequent recommendations. In summary, by constructing scenic spot preference index through a two-dimensional approach of "browsing preference index (integrating browsing behavior details) + search preference index (integrating search behavior details)," we can capture both the interest signals of users in passive browsing and the demand expressions in active searching, thereby achieving a multi-dimensional and quantifiable evaluation of user preferences.

[0026] In one embodiment, step S4, which involves obtaining consumer sensitivity based on the browsing behavior data and the search behavior data, includes: S401. Obtain the user's second dwell time in the price section and the frequency of price filtering operations based on the browsing behavior data; S402. Obtain the browsing behavior coefficient based on the second dwell time and the frequency of the price filtering operation; S403. Obtain the frequency of comparison keywords and the number of historical low-price queries based on the search behavior data, and obtain the search behavior coefficient based on the frequency of comparison keywords and the number of historical low-price queries; S404. Obtain consumption sensitivity based on the browsing behavior coefficient and the search behavior coefficient.

[0027] As described in steps S401-S404 above, this invention first obtains the user's second dwell time and price filtering frequency in the price section based on the browsing behavior data. The browsing behavior data comes from the platform interaction logs in the user behavior graph (consistent with "obtaining user behavior data from the user behavior graph" in claim 1), specifically including: Second dwell time: the time the user spends in price sections such as "Price Description" and "Ticket Price Details" on the scenic spot details page (e.g., the user spends 4 minutes in the "Jiuzhaigou Ticket Price List" section); Price filtering frequency: the number of times the user uses filtering conditions such as "Price Range" and "Low Price Priority" when filtering tickets on the platform (e.g., the user filters "Tickets under 100 yuan" 3 times within 1 hour). The core details reflecting price concerns in the browsing behavior are extracted. A longer second dwell time indicates that the user is actively understanding the price composition (e.g., distinguishing between adult and child ticket prices); a high price filtering frequency indicates that the user is narrowing down the price range through filtering and has specific requirements for the price range. For example, a user who spends 5 minutes in the price section and filters for "50-150 yuan" tickets 4 times is clearly more price-sensitive than a user who spends 1 minute and filters for prices 0 times. Then, the browsing behavior coefficient is obtained based on the second dwell time and the frequency of price filtering operations. A weighted summation formula is used to quantify price-sensitive behavior during browsing. The weights are set according to the intensity of the behavior's indication of "price attention" (e.g., the second dwell time has a weight of 0.6, as dwell time better reflects active attention; the price filtering frequency has a weight of 0.4, as the operation is more purposeful). Example formula: Browsing behavior coefficient = (Second dwell time ÷ Industry average price column dwell time) × 0.6 + (Price filtering frequency ÷ Total user filtering frequency) × 0.4, and all indicators are normalized to the [0,1] range. This transforms scattered price operations during browsing into a single quantitative value, reflecting the intensity of user attention to prices during browsing, while avoiding misjudgment of price sensitivity due to a single browsing behavior. The combination of "dwell time + filtering frequency" more accurately captures the user's price preferences during browsing. Next, the frequency of price comparison keywords and the number of historical low price queries are obtained based on the search behavior data. Then, a search behavior coefficient is obtained based on the frequency of price comparison keywords and the number of historical low price queries. The search behavior data also comes from the search logs in the user behavior graph (consistent with "obtaining search behavior data from the user behavior graph" in section 1), specifically including: Price comparison keyword frequency: the number of times a user's search contains price-related terms such as "lowest price," "price comparison," and "discount" (e.g., a user searches for "lowest price for Forbidden City tickets" twice in one day); Number of historical low price queries: the number of times a user clicks on the platform's "historical price" section. The search behavior coefficient is calculated by tracking the number of times users use the "price comparison" function (e.g., checking the lowest price of a ticket in the past 30 days) (e.g., a user searches for "historical low price of Great Wall tickets" 3 times). A weighted summation formula is used to quantify price-sensitive behavior during searches, with weights emphasizing proactive demand expression (e.g., the frequency of price comparison keywords has a weight of 0.5, and the number of historical low price queries has a weight of 0.5). The formula is: Search Behavior Coefficient = (Frequency of Price Comparison Keywords ÷ Total User Search Frequency) × 0.5 + (Number of Historical Low Price Queries ÷ Total User Queries) × 0.5. This transforms price-related behavior in proactive searches into a quantifiable value, reflecting the intensity of users' proactive price exploration. A high frequency of price comparison keywords indicates that users are clearly looking for price advantages; a high number of historical low price queries indicates that users are judging whether the current price is worthwhile and may be "waiting for a lower price." Furthermore, proactive search behavior strengthens price sensitivity judgment and compensates for the potential for accidental errors in passive browsing (e.g., accidentally clicking on the price bar). Finally, the consumption sensitivity is obtained based on the browsing behavior coefficient and the search behavior coefficient. A weighted average is used to integrate the two coefficients (the weights can be adjusted according to the business scenario; for example, if search behavior better reflects a clear need, it is set to 0.6; in this example, each is 0.5). Example formula: Consumption Sensitivity = Browsing Behavior Coefficient × 0.5 + Search Behavior Coefficient × 0.5. Example: User A's browsing behavior coefficient is 0.8, and search behavior coefficient is 0.775, then consumption sensitivity = 0.8 × 0.5 + 0.775 × 0.5 = 0.4 + 0.3875 = 0.7875 (high sensitivity); User B's browsing behavior coefficient is 0.34, and search behavior coefficient is 0.3875, then consumption sensitivity = 0.34 × 0.5 + 0.3875 × 0.5 = 0.17 + 0.19375 = 0.36375 (low sensitivity). This combines the price-related signals from passive browsing and active searching to obtain a quantitative indicator reflecting the user's overall price sensitivity. The higher the value, the more price-sensitive the user is, and recommendations should focus on discounted tickets and promotional activities; the lower the value, the less price-sensitive the user is, and recommendations can be made for regular-priced tickets or high-end experience packages. It can also achieve dual verification of "browsing price behavior (potential interest) + searching price behavior (explicit needs)," which greatly improves the accuracy of price sensitivity assessment and provides a clear basis for subsequent recommendations on "what price users can accept," thereby improving marketing conversion rates. In summary, by constructing consumer sensitivity through a two-dimensional approach of "browsing behavior coefficient (integrating price-related actions during browsing) + search behavior coefficient (integrating price-related needs during search)," we can capture users' attention to prices during passive browsing and incorporate their explicit price needs during active searches, thereby achieving a dynamic and multi-dimensional quantitative assessment of users' price sensitivity.

[0028] In one embodiment, step S5, which involves obtaining the user's mobility behavior stability coefficient based on the geographic mobility state data, includes: S501. Obtain the total length of repeated paths and the total length of all current paths based on the geographic movement status data; S502. Obtain the path repetition rate based on the total length of the repeated paths and the total length of all current paths; S503. Obtain the user's set of dwell points based on the geographic mobility status data, and obtain multiple dwell clusters based on the K-means algorithm; S504. Obtain the start time and end time of each dwelling cluster, calculate multiple dwelling durations based on multiple start times and multiple end times, and select the dwelling duration with the longest duration as the main dwelling duration. S505. Obtain the stability coefficient of user mobile behavior based on the main dwell time and the path repetition rate.

[0029] As described in steps S501-S505 above, the present invention first obtains the total length of repeated paths and the total length of all current paths based on the geographic movement status data. The geographic movement status data comes from user status data (consistent with "obtaining user status data based on user behavior map" in claim 1), specifically user-authorized GPS trajectory records, map navigation history, etc., including all movement paths of the user within a preset period (such as the trajectory segment from point A to point B). Repeated paths refer to overlapping trajectory segments that the user has traversed at different times (such as the user traveling from home to the company every Monday to Friday, this trajectory segment is a repeated path). The total length of repeated paths is the sum of the lengths of all overlapping trajectory segments (such as the above-mentioned home to company path being 5km long, 5 times a week, the total repeated length is 5km × 5 = 25km). The total length of all current paths is the total length of all movement paths of the user within the preset period (such as the user's total movement path length in the past 30 days being 200km). In this way, the regularity of user movement is reflected by the degree of path repetition. The higher the percentage of repeated path length, the more fixed the user's activity range and the stronger the movement pattern. Then, the path repetition rate is obtained based on the total length of the repeated paths and the total length of all current paths, where the path repetition rate = total length of repeated paths ÷ total length of all current paths (value range [0,1], the higher the value, the greater the proportion of repeated paths). This quantifies the degree of path repetition into a single indicator, intuitively reflecting the regularity of user movement. A high repetition rate means that the user's activity range is relatively fixed, and potential travel destinations can be predicted through their frequently visited areas (e.g., if a scenic spot is near their frequently visited route, the probability of travel is high). At the same time, it avoids judging the user's activity range based solely on a single path, and captures stable movement patterns through the repetition rate of long-term paths, providing a basis for assessing the probability of travel. Next, based on the geographic mobility data, a set of user stop points is obtained, and multiple stop clusters are obtained based on the K-means algorithm. The set of stop points comes from the "stop records" of the user's movement trajectory in the geographic mobility data (e.g., if a user stays at a location for more than a preset threshold (e.g., 10 minutes), it is marked as a stop point, and its latitude, longitude, and time of stay are recorded). The core logic of the K-means algorithm is to cluster stop points that are spatially close together into one category ("stop cluster"). Each cluster represents a core activity area of ​​the user (e.g., residence, workplace, frequently visited shopping mall). Parameter settings: the number of clusters (k value) can be adjusted according to... The number of user dwell points is dynamically adjusted (k=3 when the number of dwell points is ≤20, k=5 when the number of dwell points is 20-50). The distance metric uses Euclidean distance (calculating the straight-line distance between latitude and longitude coordinates). For example, if a user has 50 dwell points in the past 30 days, K-means clustering results in 3 clusters: Cluster 1 (latitude and longitude concentrated near the residential area, 25 points), Cluster 2 (near the office building, 15 points), and Cluster 3 (near the business district, 10 points), corresponding to the residence, workplace, and frequently visited business district, respectively. This clustering algorithm transforms scattered dwell points into meaningful "core activity areas," avoiding the limitations of analyzing individual dwell points in isolation. For example, multiple dwell points of a user "near a certain residential area" are clustered into one cluster, which clearly indicates that the user's residence is in this area. At the same time, it can also achieve accurate identification of the user's core activity area, providing a basis for subsequent analysis of "which area is the user stably active in." Secondly, the start and end times of each stay cluster are obtained. Multiple stay durations are calculated based on these start and end times, and the longest stay duration is selected as the primary stay duration. Specifically, for each stay cluster, the start time (e.g., "8:00") and end time (e.g., "20:00") of all stay points are extracted. The duration of a single stay is calculated as: end time - start time (e.g., 12 hours). The average (or sum) of all stay durations within the same cluster is taken, and the results of all clusters are compared. The longest duration is selected as the primary stay duration. This primary stay duration reflects which core area the user's activity is most stable and where they stay the longest. This area is usually the core location where the user lives or works. If a scenic spot is near this area, the user is more likely to visit it in the short term. For example, if the primary stay duration is at the residence (12 hours) and the scenic spot is within 3km of the residence, it indicates that the user has a high probability of visiting the scenic spot for "nearby travel." Simultaneously, the "longest stay duration" identifies the user's most stable activity area, strengthening the identification of "high-travel-probability areas." Finally, the user's mobile behavior stability coefficient is obtained based on the main dwell time and the path repetition rate. A weighted summation formula is used to integrate the two indicators (weights are set according to their impact on "stability," such as path repetition rate reflecting overall movement patterns with a weight of 0.5; main dwell time reflecting the depth of regional activity with a weight of 0.5). The user mobile behavior stability coefficient = path repetition rate × 0.5 + (main dwell time ÷ 24 hours) × 0.5, with the main dwell time normalized to the [0,1] interval, such as 1 for 24 hours and 0.5 for 12 hours. For example, a path repetition rate of 0.7. 5. If the main dwell time is 12 hours (normalized to 0.5), then the coefficient = 0.75 × 0.5 + 0.5 × 0.5 = 0.375 + 0.25 = 0.625 (moderate to high stability); if the path repetition rate is 0.15 and the main dwell time is 2 hours (normalized to ≈ 0.08), then the coefficient = 0.15 × 0.5 + 0.08 × 0.5 = 0.075 + 0.04 = 0.115 (low stability). This combines the overall user movement pattern (path repetition rate) and the activity depth in the core area (main dwell time) to obtain a quantitative indicator reflecting user movement stability. A higher coefficient indicates stronger user activity patterns and stable dwell time in the core area. If a scenic spot is within their activity range, the likelihood of travel is higher. This also allows for accurate prediction of whether a user is likely to visit a certain scenic spot, providing a clear basis for subsequent recommendations on the "feasibility of user travel" (e.g., highly stable users with scenic spots near their core area are given priority in recommendations). In summary, by constructing a user mobility behavior stability coefficient through a two-dimensional approach of "path repetition rate (reflecting movement patterns) plus main dwell time (reflecting regional activity stability)," we can capture the long-term movement patterns of users and incorporate the depth of dwell time in specific areas, thereby achieving a multi-dimensional and dynamic quantitative assessment of the feasibility of user travel.

[0030] In one embodiment, step S6, which involves providing precise marketing recommendations for ticket agency services to users based on the real-time inventory pressure coefficient, the scenic spot preference, the consumption sensitivity, and the user mobile behavior stability coefficient, includes: S601. Normalize the real-time inventory pressure coefficient, the scenic spot preference, the consumption sensitivity, and the user mobile behavior stability coefficient respectively to obtain the real-time inventory pressure normalization coefficient, the scenic spot preference normalization coefficient, the consumption sensitivity normalization coefficient, and the user mobile behavior stability normalization coefficient. S602. Calculate the user's comprehensive recommendation priority score by normalizing the real-time inventory pressure coefficient, the scenic spot preference coefficient, the consumption sensitivity coefficient, and the user's mobile behavior stability coefficient. S603. Determine whether the user's comprehensive recommendation priority score is within a preset range; If it is not within the preset range but is greater than the preset range, it is determined to be a high recommendation level, and a high-priority recommendation strategy including core scenic spots and discounted tickets is generated. If the information falls within a preset range, it is classified as a medium recommendation level, and a medium-priority recommendation strategy is generated that includes relevant scenic spots and regular ticket prices. If it is not within the preset range and is smaller than the preset range, it is judged as a low recommendation level, and a low-priority recommendation strategy containing potential interest scenic spots and promotional information is generated.

[0031] As described in steps S601-S603 above, the present invention first normalizes the real-time inventory pressure coefficient, the scenic spot preference, the consumption sensitivity, and the user mobile behavior stability coefficient to obtain the real-time inventory pressure normalization coefficient, the scenic spot preference normalization coefficient, the consumption sensitivity normalization coefficient, and the user mobile behavior stability normalization coefficient. The real-time inventory pressure coefficient (calculated using weight 2), scenic spot preference (calculated using weight 3), consumption sensitivity (calculated using weight 4), and user mobile behavior stability coefficient (calculated using weight 5) are normalized using the following methods: The system employs min-max standardization (mapping coefficients to the [0,1] interval), with the formula: Normalized coefficient = (Original coefficient - Minimum value of the coefficient) ÷ (Maximum value of the coefficient - Minimum value of the coefficient). Parameter explanation: The minimum and maximum values ​​are the extreme values ​​of the coefficient in the platform's historical data (e.g., if the historical minimum value of the real-time inventory pressure coefficient is 0.2 and the maximum value is 5, then the normalized value of the current coefficient 2 for a certain scenic spot = (2-0.2) ÷ (5-0.2) = 1.8 ÷ 4.8 = 0.375). This eliminates the dimensional differences between different coefficients, allowing for direct comparison and weighting of the four dimensions. For example, before normalization, "Inventory Pressure Coefficient 5" and "Preference Degree 0.8" cannot be directly added; after normalization, both fall within the [0,1] interval, allowing for weighting to reflect the importance of each dimension. This also prevents a single coefficient from dominating the overall score due to its large numerical range, ensuring fair integration of multi-dimensional information. Next, the user's comprehensive recommendation priority score is calculated using the normalized coefficients of real-time inventory pressure, scenic spot preference, consumption sensitivity, and user mobile behavior stability. A weighted summation formula is used, with weights set according to the scenic spot's operational goals (e.g., when inventory clearance is prioritized, the normalized coefficient of real-time inventory pressure has a higher weight; when user experience is prioritized, the normalized coefficient of scenic spot preference has a higher weight). Example weight allocation: Comprehensive Score = Normalized Coefficient of Real-Time Inventory Pressure × 0.3 + Normalized Coefficient of Scenic Spot Preference × 0.3 + (1 - Normalized Coefficient of Consumption Sensitivity) × 0.2 + Normalized Coefficient of User Mobile Behavior Stability × 0.2, where "1 - Consumption Sensitivity" represents "price acceptance." A higher value indicates less price sensitivity, allowing for priority recommendation of regular-priced tickets. For example, a user with four normalized coefficients of 0.8 (high inventory pressure), 0.9 (high preference), 0.3 (low consumption sensitivity → 1 - 0.3 = 0.7), and 0.8 (high mobile stability) would have different values. If the overall score is 0.8×0.3+0.9×0.3+0.7×0.2+0.8×0.2=0.24+0.27+0.14+0.16=0.81, then the overall score quantifies the priority of "recommending the user to buy a certain ticket" - the higher the score, the more the ticket meets the comprehensive conditions of "the scenic spot needs to promote it (high inventory), the user wants it (high preference), the user can accept it (price matching), and the user can go there (high stability)". At the same time, the weight allocation balances the needs of the scenic spot and the user, avoiding the recommendation imbalance caused by a single goal (such as only clearing inventory). Finally, it is determined whether the user's overall recommendation priority score is within a preset range; If it is not within the preset range but is greater than the preset range, it is determined to be a high recommendation level, and a high-priority recommendation strategy including core scenic spots and discounted tickets is generated. If the information falls within a preset range, it is classified as a medium recommendation level, and a medium-priority recommendation strategy is generated that includes relevant scenic spots and regular ticket prices. If the value is not within the preset range and is smaller than the preset range, it is judged as a low recommendation level, and a low-priority recommendation strategy containing potential interest scenic spots and promotional information is generated. The preset range is set based on the conversion rate statistics of the platform's historical recommendation data (e.g., a score ≥0.7 is a high level, 0.3-0.7 is a medium level, and ≤0.3 is a low level, which can be dynamically adjusted).

[0032] Tiered strategy generation: High recommendation level (score > preset range upper limit): Recommend "core scenic spots + discounted tickets". Core scenic spots refer to the top 3 scenic spots with the user's preference. Discounted tickets are combined with consumption sensitivity (high sensitivity means large discount, such as 20% off; low sensitivity means small discount, such as 10% off). Medium recommendation level (score within the preset range): Recommend "related scenic spots + regular tickets". Related scenic spots refer to scenic spots of the same type as those highly preferred by the user (e.g., if the user prefers mountain and water scenic spots, recommend other mountain and water scenic spots). Regular tickets are at the original price or with a small discount (e.g., 9.5% off). Low recommendation level (score < preset lower limit): Recommend "potential interest scenic spots + promotional information". Potential interest scenic spots refer to scenic spots that users have browsed but have a moderate level of preference for. Promotional information includes "discount coupons" and "limited-time offers", which can stimulate users to try them with low barriers to entry.

[0033] For example: High-level users (score 0.81): Recommend 20% off tickets to "Huangshan Scenic Area" (high inventory pressure), which is their number 1 preference (due to low consumer sensitivity, the discount is moderate). Intermediate-level users (score 0.5): Recommend a 9.5% discount regular ticket for Lushan Scenic Area (with moderate inventory pressure), which is similar to Huangshan. For low-level users (score 0.2): Recommend the "spend 200 get 50 off" promotional information for "Sanqingshan Scenic Area" (moderate preference) that they have previously viewed; This tiered strategy directs recommended resources toward users with "high conversion potential"—high-tier users are matched with core resources and discounts to ensure high conversion rates; mid-tier users are matched with relevant resources to maintain stable conversions; and low-tier users are used to uncover potential demand through promotional information, reducing marketing costs. At the same time, it avoids the inefficient marketing of "pushing the same content to all users," thereby improving the return on investment (ROI) through precise tiering, while also taking into account the digestion of scenic area inventory and user experience. In summary, by employing the processes of "normalization (eliminating incomparability of coefficients), comprehensive score calculation (integrating multi-dimensional priorities), and hierarchical recommendation (matching user and scenic area features)," a multi-dimensional weighted and hierarchical recommendation mechanism is constructed, enabling the recommendation strategy to simultaneously meet the dual objectives of "scenic area inventory reduction" and "user acceptance."

[0034] This application also provides a scenic spot ticket agency operation and precision marketing recommendation system that integrates user behavior graphs, including: The first acquisition module is used to acquire multi-dimensional behavioral data from the scenic spot ticket digital platform. The multi-dimensional behavioral data includes platform basic data and user interaction data with the platform. The second acquisition module is used to acquire the real-time inventory pressure coefficient based on the platform's basic data. The third acquisition module is used to acquire scenic spot preference based on the user's interaction data with the platform. The fourth acquisition module is used to acquire user behavior data and user status data based on the user behavior graph, and to acquire browsing behavior data and search behavior data based on the user behavior data, and to acquire consumption sensitivity based on the browsing behavior data and the search behavior data. The fifth acquisition module is used to acquire geographic mobility status data based on the user status data, and to acquire the user mobility behavior stability coefficient based on the geographic mobility status data. The recommendation module is used to make precise marketing recommendations for ticket agency operations to users based on the real-time inventory pressure coefficient, the scenic spot preference, the consumption sensitivity, and the user mobile behavior stability coefficient.

[0035] In one embodiment, the second acquisition module includes: The first acquisition unit is used to obtain the current inventory and historical sales volume of the scenic area's digital platform to obtain the basic inventory pressure. The second acquisition unit is used to acquire the corresponding first moment based on the current inventory level, and to acquire the time decay coefficient based on the first moment and the preset end moment. The third acquisition unit is used to acquire the difference between the time-sharing reservation data and the scenic area's carrying capacity, and to acquire the time period saturation based on the difference between the time-sharing reservation data and the scenic area's carrying capacity. The fourth acquisition unit is used to acquire the real-time inventory pressure coefficient based on the basic inventory pressure, the time decay coefficient, and the time period saturation.

[0036] This application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the above-described method.

[0037] This application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method.

[0038] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in this application and in the embodiments can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual-speed SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0039] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, apparatus, article, or method. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, apparatus, article, or method that includes that element.

[0040] The above description is merely a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for precise marketing recommendation of scenic spot ticket operation services that integrates user behavior graphs, characterized in that, include: Acquire multi-dimensional behavioral data from the scenic area's digital ticketing platform. This multi-dimensional behavioral data includes both platform basic data and user interaction data with the platform. The real-time inventory pressure coefficient is obtained based on the platform's basic data, including: Obtain the current inventory and historical sales volume of the scenic area's digital platform to understand the basic inventory pressure; The first moment is obtained based on the current inventory level, and the time decay coefficient is obtained based on the first moment and the preset end moment. Obtain the difference between time-sharing reservation data and the scenic area's carrying capacity, and obtain the time-saturation based on the difference between the time-sharing reservation data and the scenic area's carrying capacity; The real-time inventory pressure coefficient is obtained based on the basic inventory pressure, the time decay coefficient, and the time period saturation. The scenic area preference is obtained based on the user's interaction data with the platform. User behavior data and user status data are obtained based on user behavior graphs, and browsing behavior data and search behavior data are obtained based on the user behavior data, and consumption sensitivity is obtained based on the browsing behavior data and the search behavior data. Geographic mobility status data is obtained based on the user status data, and a user mobility behavior stability coefficient is obtained based on the geographic mobility status data. Based on the real-time inventory pressure coefficient, the scenic spot preference, the consumption sensitivity, and the user mobile behavior stability coefficient, precise marketing recommendations for ticket agency operations are made to users.

2. The method for precise marketing recommendation of scenic spot ticket agency operation based on user behavior graphs as described in claim 1, characterized in that, The step of obtaining scenic spot preference based on the user's interaction data with the platform includes: Based on the user's interaction data with the platform, we obtain the user's first dwell time, page scroll depth, and number of image clicks on the scenic spot details page on the platform. The browsing preference coefficient is obtained based on the first dwell time, the page scrolling depth, and the number of image clicks. Obtain the frequency of keyword repetition and the click-through rate of search results entered by the user, and obtain the search preference coefficient based on the frequency of keyword repetition and the click-through rate of search results; The scenic area preference degree is obtained based on the browsing preference coefficient and the search preference coefficient.

3. The method for precise marketing recommendation of scenic spot ticket agency operation based on user behavior graphs as described in claim 1, characterized in that, The step of obtaining consumption sensitivity based on the browsing behavior data and the search behavior data includes: Based on the browsing behavior data, the user's second dwell time in the price section and the frequency of price filtering operations are obtained; The browsing behavior coefficient is obtained based on the second dwell time and the frequency of the price filtering operation; The frequency of comparison keywords and the number of historical low-price queries are obtained based on the search behavior data, and the search behavior coefficient is obtained based on the frequency of comparison keywords and the number of historical low-price queries. Consumer sensitivity is obtained based on the browsing behavior coefficient and the search behavior coefficient.

4. The method for precise marketing recommendation of scenic spot ticket agency operation based on user behavior graph as described in claim 1, characterized in that, The step of obtaining the user's mobility behavior stability coefficient based on the geographic mobility status data includes: The total length of repeated paths and the total length of all current paths are obtained based on the geographic movement status data. The path repetition rate is obtained based on the total length of the repeated paths and the total length of all current paths. The user's set of dwell points is obtained based on the geographic mobility status data, and multiple dwell clusters are obtained based on the K-means algorithm; Obtain the start time and end time of each dwell cluster, calculate multiple dwell durations based on multiple start times and multiple end times, and select the dwell duration with the longest duration as the main dwell duration; The stability coefficient of user mobile behavior is obtained based on the main dwell time and the path repetition rate.

5. The method for precise marketing recommendation of scenic spot ticket agency operation based on user behavior graphs as described in claim 1, characterized in that, The step of providing precise marketing recommendations for ticket agency services based on the real-time inventory pressure coefficient, the scenic spot preference, the consumption sensitivity, and the user mobile behavior stability coefficient includes: The real-time inventory pressure coefficient, the scenic spot preference, the consumption sensitivity, and the user mobile behavior stability coefficient are normalized to obtain the real-time inventory pressure normalization coefficient, the scenic spot preference normalization coefficient, the consumption sensitivity normalization coefficient, and the user mobile behavior stability normalization coefficient. The user's comprehensive recommendation priority score is calculated by normalizing the real-time inventory pressure coefficient, the scenic spot preference coefficient, the consumption sensitivity coefficient, and the user's mobile behavior stability coefficient. Determine whether the user's overall recommendation priority score is within a preset range; If it is not within the preset range but is greater than the preset range, it is determined to be a high recommendation level, and a high-priority recommendation strategy including core scenic spots and discounted tickets is generated. If the information falls within a preset range, it is classified as a medium recommendation level, and a medium-priority recommendation strategy is generated that includes relevant scenic spots and regular ticket prices. If it is not within the preset range and is smaller than the preset range, it is judged as a low recommendation level, and a low-priority recommendation strategy containing potential interest scenic spots and promotional information is generated.

6. A scenic spot ticket agency operation and precision marketing recommendation system integrating user behavior graphs, characterized in that, include: The first acquisition module is used to acquire multi-dimensional behavioral data from the scenic spot ticket digital platform. The multi-dimensional behavioral data includes platform basic data and user interaction data with the platform. The second acquisition module is used to acquire the real-time inventory pressure coefficient based on the platform's basic data. The third acquisition module is used to acquire scenic spot preference based on the user's interaction data with the platform. The fourth acquisition module is used to acquire user behavior data and user status data based on the user behavior graph, and to acquire browsing behavior data and search behavior data based on the user behavior data, and to acquire consumption sensitivity based on the browsing behavior data and the search behavior data. The fifth acquisition module is used to acquire geographic mobility status data based on the user status data, and to acquire the user mobility behavior stability coefficient based on the geographic mobility status data. The recommendation module is used to make precise marketing recommendations for ticket agency operations to users based on the real-time inventory pressure coefficient, the scenic spot preference, the consumption sensitivity, and the user mobile behavior stability coefficient.

7. The scenic spot ticket agency operation and precision marketing recommendation system integrating user behavior graphs as described in claim 6, characterized in that, The second acquisition module includes: The first acquisition unit is used to obtain the current inventory and historical sales volume of the scenic area's digital platform to obtain the basic inventory pressure. The second acquisition unit is used to acquire the corresponding first moment based on the current inventory level, and to acquire the time decay coefficient based on the first moment and the preset end moment. The third acquisition unit is used to acquire the difference between the time-sharing reservation data and the scenic area's carrying capacity, and to acquire the time period saturation based on the difference between the time-sharing reservation data and the scenic area's carrying capacity. The fourth acquisition unit is used to acquire the real-time inventory pressure coefficient based on the basic inventory pressure, the time decay coefficient, and the time period saturation.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 5.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 5.

Citation Information

Patent Citations

  • Member shopping data mining algorithm comprehensive engine

    CN110619552A

  • Product marketing service platform system and method thereof

    CN119693051A

  • Bank user-based point exchange mall commodity on-shelf intelligent commodity selection method and system

    CN120146924A