Oversale management method and device, storage medium and product

Abstract

The invention provides an oversale management method and device, a storage medium and a product, and relates to the technical field of data processing, and the method comprises the steps: obtaining first information corresponding to a selling object, and the first information comprises selling information of the selling object and / or information affecting the selling number of the selling object; on the basis of the first information, the deterministic waste probability of the selling object is predicted, and the deterministic waste probability refers to the probability that the selling object is reserved but the selling object is idle; and on the basis of the deterministic waste probability and the first information, predicting the oversale quantity of the selling objects to obtain the predicted oversale quantity of the selling objects. According to the method, on the premise that the oversale risk is effectively controlled and the idle waste risk is reduced, the predicated oversale quantity is determined by predicting the deterministic waste probability to carry out oversale management, so that the maximization of the net income is realized.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to an overselling management method, device, storage medium and product. Background Technology

[0002] Overbooking refers to a situation where a merchant sells more reservations than the total number of available resources. Taking airline overbooking as an example, it means that an airline sells more tickets than the actual number of seats on the plane when all seats are sold out. This is done to mitigate the risk of wasted seats due to some passengers not boarding, thereby increasing overall flight revenue.

[0003] In existing technologies, overselling management often uses a fixed overselling ratio or a static statistical model to determine the overselling quantity. Therefore, existing technologies generally suffer from one-sided decision-making and insufficient flexibility. Either insufficient overselling leads to waste of resources, or excessive overselling increases compensation costs and damages brand reputation, failing to achieve a dynamic balance between maximizing profits and controlling risks. Summary of the Invention

[0004] This application provides a method, device, storage medium, and product for managing overbooking of airline tickets. Under the premise of effectively controlling the risk of overbooking and reducing the risk of idle waste, it can determine the predicted overbooking quantity by predicting the probability of deterministic waste and thus manage overbooking to maximize net revenue.

[0005] To address the above problems, the embodiments of this application provide the following technical solutions: In a first aspect, this application provides an overselling management method, which includes: obtaining first information corresponding to a sales object; predicting the deterministic waste probability of the sales object based on the first information; and predicting the overselling quantity of the sales object based on the deterministic waste probability and the first information to obtain the predicted overselling quantity of the sales object; wherein, the first information includes sales information of the sales object and / or information affecting the sales quantity of the sales object; the deterministic waste probability refers to the probability that a sales object is reserved but is left idle.

[0006] Based on the aforementioned technical means, by obtaining the sales information of the sales object and related information affecting the sales quantity, the probability of certain waste due to idleness after reservation is predicted. Then, by combining the probability of certain waste with the first information, the oversold quantity is dynamically predicted, avoiding the problem of under-selling / over-selling caused by relying on experience or static data, and minimizing the waste of idle sales objects.

[0007] In one possible implementation, the oversold quantity of a sales object is predicted based on the deterministic waste probability and first information, resulting in the predicted oversold quantity. Specifically, this can be achieved by processing the deterministic waste probability and first information to obtain the sales status characteristics of the sales object, which describe the sales status of the sales object; predicting the oversold quantity of the sales object based on the sales status characteristics to obtain a candidate oversold quantity range; and selecting the candidate oversold quantity that meets the selection criteria from the candidate oversold quantity range as the predicted oversold quantity.

[0008] Based on the aforementioned technical means, by performing targeted data processing on the deterministic waste probability and first information, a sales status feature that accurately describes the actual sales status of the sales object is proposed. By first defining the range of candidate oversold quantities and then screening the candidate oversold quantities, the rigidity of single numerical prediction is avoided, and the optimal solution can be accurately determined through screening conditions. This ensures that the oversold quantity is highly compatible with the idle risk and sales status, and minimizes the conflict between resource idleness and oversold.

[0009] In one possible implementation, the candidate oversold quantity that meets the screening criteria is selected from the candidate oversold quantity range as the predicted oversold quantity. Specifically, this can be achieved by: calculating the revenue value corresponding to each candidate oversold quantity, where the revenue value refers to the revenue brought by the oversold quantity of the sold object; and selecting the candidate oversold quantity from the revenue value corresponding to each candidate oversold quantity as the predicted oversold quantity.

[0010] Based on the aforementioned technical means, by calculating the quantitative revenue value corresponding to each candidate oversold quantity, the oversold revenue is visualized, avoiding the blindness of relying on experience for screening. With revenue as the core screening criterion, the optimal oversold quantity can be accurately identified. At the same time, the revenue calculation implicitly considers the oversold conflict costs and idle risks, achieving a dynamic balance between maximizing revenue and risks and costs. This ensures that the oversold quantity matches the idle scale corresponding to the probability of deterministic waste, while also avoiding oversold without revenue.

[0011] In one possible implementation, the oversold quantity of a sales object is predicted based on the sales status characteristics to obtain a candidate oversold quantity range. Specifically, the upper limit of the candidate oversold quantity range is obtained by multiplying the deterministic waste probability in the sales status characteristics with the rated quantity of the sales object. The rated quantity of the sales object refers to the maximum number of sales objects that can be used. The value between zero and the upper limit value is determined as the candidate oversold quantity range.

[0012] Based on the aforementioned technical means, the upper limit of candidate oversales is determined by multiplying the product of the deterministic waste probability and the fixed quantity, thus anchoring the range to the actual idle scale and avoiding blind oversales that are out of line with risk. The value range from 0 to the upper limit covers both the conservative choice of not overselling and precisely limits the reasonable boundary of overselling, ensuring that the scale of overselling is highly compatible with the idle risk. Defining the range based on quantitative data eliminates reliance on experience, lays a solid scientific foundation for subsequent selection of the optimal overselling quantity, takes into account resource utilization and overselling conflict prevention, and improves the rationality and operability of overselling decisions.

[0013] In one possible implementation, predicting the deterministic waste probability of the sale object based on the first information can be specifically achieved by processing the first information to obtain the first information feature corresponding to the first information; inputting the first information feature into the probability prediction model for prediction to obtain the deterministic waste probability.

[0014] Based on the aforementioned technical means, by processing the first information, the first information features that accurately reflect the idle risk are extracted, redundant and interfering data are eliminated, and high-quality input is provided for probability prediction. By using a probability prediction model to replace traditional experience-based judgment, the prediction accuracy and objectivity of the probability of deterministic waste are significantly improved, and subjective biases are avoided. This lays a solid foundation for accurate risk basis for subsequent overselling quantity prediction, ensuring that the scale of overselling is highly compatible with the idle risk.

[0015] In one possible implementation, the overselling management method provided in this application can also be specifically implemented as follows: when the number of sold objects already acquired exceeds the rated number of sold objects, overselling management is performed on users who have acquired sold objects but have not used them.

[0016] Based on the aforementioned technical means, when the acquired quantity exceeds the quota, targeted overselling management can be carried out for users who have reserved but not used the funds, thereby achieving precise response to overselling scenarios; avoiding user conflicts and compliance risks caused by disorderly overselling, protecting user rights and reducing negative feedback through flexible methods such as voluntary adjustments; at the same time, maximizing the activation of idle resources, taking into account resource utilization, operational order and user experience, and improving the completeness and effectiveness of overselling management.

[0017] In one possible implementation, overselling management for users who have acquired but not used the sale objects can be specifically implemented by sending compensation invitations to users who have acquired but not used the sale objects; identifying target users from among the users who sent the compensation invitations based on the users' responses to the compensation invitations; and reclaiming the sale objects owned by the target users.

[0018] Based on the aforementioned technical means, compensation invitations are sent to users who have reserved but not used the service. Target users are identified and resold based on voluntary responses, avoiding user conflicts and brand reputation damage caused by forced repossession. This approach accurately selects highly suitable users, improving the success rate of invitations and the efficiency of handling overselling, effectively alleviating overselling pressure. It protects users' right to choose and their rights while also revitalizing idle resources and ensuring sales order, forming a flexible overselling response mechanism that balances operational needs and user experience, significantly improving the compliance and effectiveness of overselling management.

[0019] Secondly, this application provides an overselling management device, which includes: an acquisition module, a first prediction module, and a second prediction module.

[0020] The acquisition module is used to acquire the first information corresponding to the sales object. The first information includes the sales information of the sales object and / or information affecting the sales quantity of the sales object.

[0021] The first prediction module is used to predict the probability of certainty of waste of the sale object based on the first information. The probability of certainty of waste refers to the probability that the sale object is reserved but is left idle.

[0022] The second prediction module is used to predict the oversold quantity of the sales object based on the deterministic waste probability and the first information, and obtain the predicted oversold quantity of the sales object.

[0023] The first prediction module is also used to process the first information to obtain the first information feature corresponding to the first information; input the first information feature into the probability prediction model for prediction to obtain the deterministic waste probability.

[0024] The second prediction module is also used to process the deterministic waste probability and the first information to obtain the sales status characteristics of the sales object. The sales status characteristics are used to describe the sales status of the sales object. Based on the sales status characteristics, the oversold quantity of the sales object is predicted to obtain the candidate oversold quantity range. The candidate oversold quantity that meets the screening conditions is selected from the candidate oversold quantity range as the predicted oversold quantity.

[0025] The second prediction module is also used to calculate the revenue value corresponding to each candidate oversold quantity in the candidate oversold quantity. The revenue value refers to the revenue brought by the oversold quantity of the sold object; the candidate oversold quantity is selected from the revenue value corresponding to each candidate oversold quantity as the predicted oversold quantity.

[0026] The second prediction module is also used to obtain the upper limit of the candidate oversold quantity range based on the product of the deterministic waste probability in the sales status characteristics and the rated quantity of the sales object. The rated quantity of the sales object refers to the maximum number of sales objects that can be used. The value between zero and the upper limit value is determined as the candidate oversold quantity range.

[0027] This application provides an overselling management device, which further includes an overselling management module.

[0028] The overselling management module is used to manage overselling to users who have acquired but not used the sales objects when the number of acquired sales objects exceeds the rated number of sales objects.

[0029] The overselling management module is also used to send compensation invitations to users who have acquired sales objects but have not used them; based on the users' responses to the compensation invitations, to identify target users from the users who sent the compensation invitations; and to reclaim the sales objects owned by the target users.

[0030] Thirdly, this application provides an electronic device comprising a processor and a memory. The memory stores processor-executable instructions, and when the processor is configured to execute the instructions, the electronic device implements the method described in the first aspect above.

[0031] Fourthly, this application provides a readable storage medium comprising software instructions. When the software instructions are executed in an electronic device, they cause the electronic device to implement the method described in the first aspect above.

[0032] Fifthly, this application provides a computer program product comprising computer instructions. When the computer instructions are executed on an electronic device, the electronic device performs the method described in the first aspect.

[0033] The beneficial effects of the second to fifth aspects mentioned above can be referred to the first aspect, and will not be repeated here. Attached Figure Description

[0034] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0035] Figure 1 This is a schematic diagram of the structure of an overselling management system provided in an embodiment of this application; Figure 2 A flowchart illustrating an overselling management method provided in an embodiment of this application; Figure 3 A flowchart illustrating yet another overselling management method provided in this application embodiment; Figure 4 A flowchart illustrating yet another overselling management method provided in this application embodiment; Figure 5 A flowchart illustrating yet another overselling management method provided in this application embodiment; Figure 6 A schematic diagram of the layered architecture of an overselling management system provided in this application embodiment; Figure 7 A schematic diagram of the logical structure of an overselling management method provided in an embodiment of this application; Figure 8 This is a schematic diagram of the structure of an overselling management device provided in an embodiment of this application; Figure 9 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0036] Hereinafter, the terms "first," "second," and "third," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, a feature defined as "first," "second," or "third," etc., may explicitly or implicitly include one or more of that feature. In the description of this embodiment, unless otherwise stated, "a plurality of" means two or more.

[0037] It should be noted that, in this application, the terms "exemplary" or "for example" are used to indicate that something is being described as an example, illustration, or illustration. Any embodiment or design described as "exemplary" or "for example" in this application should not be construed as being more preferred or advantageous than other embodiments or design solutions. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.

[0038] In addition, for ease of understanding, the technical terms involved in the embodiments of this application will be introduced below.

[0039] For example, with the recovery of cultural and tourism consumption, the popularization of online booking models, and the diversification of sales scenarios, such as flight tickets, hotels, performance tickets, and pre-sales of scarce goods, the booking volume of sales targets has experienced explosive growth. Enterprises are increasingly eager to maximize resource utilization and revenue. Overbooking, as a key means to alleviate idle bookings and improve resource turnover efficiency, has become a core operational strategy in industries such as travel services and retail. However, at the same time, factors such as increased fluctuations in demand for sales targets, the diversification of booking channels, and increasingly stringent compliance supervision have significantly increased the complexity of overbooking management.

[0040] In related technologies, the prediction of oversold quantities relies on human experience or simple statistical models, and is calculated statically based solely on historical average idle data. This makes it difficult to adapt to demand fluctuations, and often results in under-selling or over-selling.

[0041] Against this backdrop, how to accurately match the dynamic idle risk with the oversold quantity, balance the resource utilization rate with the oversold conflict risk, and at the same time reduce compensation costs and improve user experience through refined handling, so as to maximize long-term benefits, has become an urgent problem to be solved.

[0042] Therefore, in order to overcome the above problems, this application provides an overselling management method. By obtaining the sales information of the sales object and the relevant first information affecting the sales quantity, the method predicts the deterministic waste probability that represents the risk of idleness after reservation. Based on reinforcement learning, an action space integrating multi-dimensional state features and compliance threshold constraints is constructed to output the optimal predicted overselling quantity. When overselling occurs, an elastic scoring mechanism is constructed based on the user's reservation characteristics, attribute data, historical behavior and other related information. Users with high suitability are invited to participate in voluntary adjustment first, so as to minimize the waste of idle resources, control the cost of overselling conflicts, and improve user experience and long-term cumulative benefits.

[0043] The overselling management method provided in this embodiment will be described below, starting with an introduction to the relevant technologies.

[0044] The overselling management method provided in this application embodiment can be applied to, for example, Figure 1 The overselling management system shown is Figure 1 This is a schematic diagram of the structure of an overbooking management system provided in an embodiment of this application. The overbooking management system can be applied to travel service scenarios, life service scenarios, and cultural and tourism consumption scenarios. Figure 1 The overselling management system 100 shown includes: a data acquisition module 101, an overselling management module 102, and an execution module 103.

[0045] The aforementioned travel service scenarios include: flight seat reservations, high-speed rail / bullet train seat pre-sales, long-distance bus seat reservations, and cruise ship cabin sales. The core of these scenarios is the limited availability of seats. After reservation, users may not actually use the seats due to itinerary changes or last-minute cancellations, resulting in idle resources. Lifestyle service scenarios include hotel room pre-sales, guesthouse bookings, time-sharing rentals of shared items such as shared power banks / umbrellas, and the sale of gym group classes / yoga studio reservation slots. These scenarios often involve limited allocation of space or item usage rights, posing a risk of idleness due to no attendance or unused items after reservation. Cultural and tourism consumption scenarios include concert / drama / sports event ticket sales, limited-edition scenic spot ticket pre-sales, museum special exhibition reservation slots, and theme park fast-track access slots. Resources in these scenarios are characterized by high time sensitivity and non-reusability; once idle, they cannot be resold, making overbooking management particularly urgent.

[0046] It should be understood that the above scenarios are merely illustrative examples and are not intended to limit the scope of protection of this application. The overselling management method of this application is applicable to any sales scenario with limited available resources, a pre-ordering model, and the risk of idleness. Furthermore, each scenario can be adapted to the specific needs of the scenario by flexibly adjusting the collection dimension of the first information, the overselling compliance threshold, model parameters, etc., thus fully demonstrating the universality and scalability of this method.

[0047] The data acquisition module 101, the over-selling management module 102, and the execution module 103 are connected via a preset network interface.

[0048] It should be understood that the above communication connection methods may include wired or wireless communication, and may also achieve real-time data transmission and command interaction through message queues, etc., to ensure the efficiency and stability of inter-module collaboration.

[0049] The overbooking management system 100 can be a standalone server cluster, a cloud-based distributed system, or a functional module embedded in an existing reservation system, adapting to the technical architecture needs of different enterprises.

[0050] Specifically, the oversale management system supports various sales scenarios such as flights, hotels, performance tickets, and pre-sales of scarce goods. It can flexibly adjust parameter configurations according to the attributes of the sales objects, and has strong versatility and scalability.

[0051] It should be understood that Figure 1 This is just an example structure. In actual applications, data storage modules, monitoring modules, etc. can be added as needed. The division of each module is not limited to the above form, and the core functions can be split or merged to implement them.

[0052] The data acquisition module 101 can be a hardware device or software component with data capture and interface connection capabilities.

[0053] It should be understood that the data acquisition module 101 can cover both internal business systems and external data sources, and the acquisition frequency supports real-time acquisition or timed incremental acquisition to ensure the timeliness and integrity of the data.

[0054] Specifically, the data acquisition module 101 is used to acquire the first information of the sales object, including internal sales information and external influence information, and after cleaning and format conversion of the acquired raw data, it is transmitted to the oversale management module 102.

[0055] The overselling management module 102 can be a core computing unit with data processing and model calculation capabilities.

[0056] It should be understood that the overselling management module 102 is the core decision-making unit of the system, which integrates a deterministic waste prediction model and a quantity prediction model, and has the functions of model training, real-time reasoning and decision generation.

[0057] Specifically, the oversold management module 102 receives the first information transmitted by the data acquisition module 101, predicts the probability of deterministic waste based on the information, and combines the probability with the first information to output the predicted oversold quantity through the quantity prediction model. At the same time, it generates an oversold management instruction and sends it to the execution module 103.

[0058] The execution module 103 can be a software functional module or hardware interface component with the ability to execute instructions and implement business processes, such as a reservation system interface, a user notification gateway, or an inventory management plugin.

[0059] It should be understood that the execution module 103 directly connects to the enterprise's existing business system, which can transform oversold management instructions into specific business operations to ensure that the strategy is implemented quickly. Specifically, the execution module 103 is used to adjust the inventory configuration in the reservation system according to the predicted oversold quantity. When oversold situations occur, it sends voluntary adjustment invitations to target users according to preset strategies, and simultaneously tracks user response results and feeds them back to the oversold management module 102.

[0060] like Figure 2 As shown, Figure 2 This is a flowchart illustrating an overselling management method provided in an embodiment of this application. The specific steps of the overselling management method provided in this embodiment are as follows: S201: Obtain the first information corresponding to the sales target.

[0061] The first piece of information includes the sales information of the target product and / or information affecting the sales quantity of the target product.

[0062] The aforementioned sales targets refer to products or services with fixed quotas for use, clear usage time constraints, and the risk that users may not actually use the products after booking them during the sales period. These include, but are not limited to, air tickets, hotel rooms, performance / event tickets, high-speed rail seats, car rental services, and reservation-based courses.

[0063] The sales information of the aforementioned sales targets refers to core business data directly related to the sales targets, including but not limited to the total basic inventory, current real-time booked volume, unit sales revenue, sales cycle, distribution of booking channels, and refund and change rules. Taking air tickets as an example, the sales information of the sales targets includes: the total number of seats corresponding to the flight number, the number of sold seats and the number of locked seats, the unit price of economy class / business class / first class, the sales period from the time the ticket goes on sale to the time before the flight takes off, the distribution of booking channels, the deadline for free refunds and changes, the handling fee ratio for different refund and change periods, the number of remaining available seats, and the distribution of seat classes.

[0064] The information affecting the sales volume of the aforementioned products refers to external and related data that indirectly affect sales demand, including but not limited to market demand, holiday / seasonal factors, competitor pricing strategies, regional economic activities, and transportation hub dynamics. Taking air tickets as an example, this specifically includes: real-time search volume for routes, market demand, peak travel periods during holidays, seasonal factors such as rainy / winter seasons, fare ranges of other airlines on the same route, promotional activities, competitor pricing strategies, arrangements for large-scale conferences, exhibitions, and sporting events at the departure or destination, airport flight take-off and landing capacity, runway construction impact, opening of high-speed rail / intercity rail lines and fare adjustments, destination visa policies, epidemic prevention policies, and weather warning information.

[0065] S202: Based on the first piece of information, predict the probability of certainty waste of the target for sale.

[0066] Among them, the probability of certainty waste refers to the probability that a product or service is reserved but is left unused. Taking air tickets as an example, it specifically refers to the probability that a passenger, after booking and confirming the reservation, fails to process a refund or change before the flight's departure and fails to arrive at the airport on time to complete the check-in process, resulting in the corresponding seat being left unused.

[0067] S203: Based on the deterministic waste probability and first information, predict the oversold quantity of the sales object to obtain the predicted oversold quantity of the sales object.

[0068] The predicted overbooking quantity refers to the overbooking scale calculated based on idle risk and multi-dimensional operational data under the constraint of overbooking compliance threshold. In other words, it represents the candidate overbooking quantity that maximizes the reduction of resource idleness while controlling overbooking conflict costs. This is generated by a quantity prediction model constructed through reinforcement learning, using deterministic waste probability and initial information as input. Taking airfare as an example, when 6 seats are overbooked, the idle probability covers most of the overbooking scale, and the overbooking ratio meets compliance requirements. Ultimately, 6 seats are determined as the predicted overbooking quantity for that flight. like Figure 3 As shown, Figure 3 The following is a flowchart illustrating another overselling management method provided in this application embodiment. The specific steps of the overselling management method provided in this application embodiment are as follows: S301: Obtain the first information corresponding to the sales object.

[0069] S302: Process the first information to obtain the first information feature corresponding to the first information.

[0070] Specifically, data processing includes encoding and converting discrete data in the first information, standardizing and normalizing continuous data, removing outliers and filling in missing values, and finally extracting core dimensional features that are strongly correlated with the probability of deterministic waste, forming structured first information features.

[0071] S303: Input the first information feature into the probability prediction model for prediction to obtain the deterministic waste probability.

[0072] Among them, the probability prediction model is a model trained based on the historical first information of the sale object and the historical idle result data.

[0073] The aforementioned historical first information includes: historical sales information of the sales object and / or information affecting the historical sales volume of the sales object within the historical sales period. Taking air tickets as an example, it specifically includes historical sales data of past flights on the route and related data affecting the ticket sales volume of the route, such as holiday arrangements, competitor ticket prices on the same route, and airport traffic control policies.

[0074] The aforementioned historical idle result data refers to the record data of idle items within the historical sales period. Taking air tickets as an example, it specifically includes detailed data of seats that have been booked but not boarded in historical flights. For example, on May 10th of a certain year, the number of reserved seats for the A-B flight was 150, and the actual number of boarded seats was 132. Among them, the specific information of the 18 seats that were not boarded (the booking channel, booking duration, cancellation and change records, seat class, etc. of the passengers who did not board), the percentage of seats that were not boarded, and the classification statistics of the reasons for not boarding, etc.

[0075] Specifically, historical sales information includes data that directly reflects the sales process, such as historical pre-orders, historical actual usage, historical cancellations and changes, cancellation and change rates, historical sales price ranges, historical pre-order percentages for different time periods, and historical inventory depletion rates. Information that affects the historical sales volume of the target product includes external and internal data that indirectly affects the sales scale, such as historical market demand, historical competitors' sales prices and inventory strategies, historical time-related factors such as holidays / seasons / working days, historical regional market consumption capacity data, and historical policy control information (such as travel restrictions and industry compliance adjustments).

[0076] For example, taking air tickets as an example, historical sales information specifically includes: the average number of reservations for the flight, the actual number of passengers boarding, the cancellation and change rate, and the selling price of the ticket; information affecting the sales volume specifically includes: the search volume of routes during holidays, the selling price of other airlines on the same route and / or the corresponding after-sales service, the impact of holidays, and travel restrictions caused by the whole vehicle, etc.

[0077] For example, in the same period of 2023, the average number of reserved seats for flights on the City A-City B route was 142, the actual number of boarding seats was 125, the cancellation and change rate was 12%, the economy class ticket price range was 800-1500 yuan, the proportion of bookings made 7 days before departure was 40%, and the inventory consumption rate 24 hours before departure was 5 seats per hour. Specific information affecting the number of seats sold included: the average daily search volume for this route during the 2023 Spring Festival holiday increased by 200% compared to weekdays; the average economy class ticket price of other airlines on the same route was 950 yuan and free change services were offered; the booking volume for this route during the summer vacation in July-August 2023 increased by 180% compared to other periods; data related to the per capita disposable income of Beijing, the departure city, in Q3 2023; and travel restrictions imposed by a certain airport due to epidemic prevention and control measures in October 2023.

[0078] The aforementioned historical sales period refers to the continuous or segmented sales period from the first historical sales time of the sales object to the period before the certainty of waste in this prediction, and must include at least the latest historical data of the most recent complete sales period up to the time of this prediction; taking air tickets as an example, if the sales period for a flight is from September 15, 2024 to October 20, 2024 at 14:00, then the historical sales period can be selected from the sales periods of flights on the same route and at the same time during October 2023 to September 2024, that is, the period from the first sale date to the last sale date for each historical flight, and must include all historical data of flights on the same route at 16:00 on September 20, 2024.

[0079] S304: Based on the deterministic waste probability and first information, predict the oversold quantity of the sales object to obtain the predicted oversold quantity of the sales object.

[0080] The embodiments provided in this application extract features related to the probability of deterministic waste, use a probability prediction model to output the probability of deterministic waste, and then combine the first information to predict the oversold quantity; this avoids the bias of static prediction, greatly improves the accuracy of waste probability prediction, and makes the oversold quantity accurately match the idle risk, thereby minimizing resource idle waste.

[0081] like Figure 4 As shown, Figure 4 The following is a flowchart illustrating another overselling management method provided in this application embodiment. The specific steps of the overselling management method provided in this application embodiment are as follows: S401: Obtain the first information corresponding to the sales object.

[0082] S402: Based on the first piece of information, predict the probability of certainty waste of the target for sale.

[0083] S403: Process the deterministic waste probability and the first information to obtain the sales status characteristics of the sales object.

[0084] Among them, the sales status feature is used to describe the sales status of the sales object. Taking air tickets as an example, it specifically includes: current occupancy rate and remaining available seats in terms of inventory; booking growth rate in the past 24 hours and real-time search volume of routes in terms of demand; probability of deterministic waste; and unit sales revenue corresponding to different cabin classes.

[0085] The aforementioned data processing refers to the unified preprocessing of discrete and continuous data in the deterministic waste probability and the first information, including data cleaning, standardization and quantification, and feature fusion, to form a structured sales status feature vector.

[0086] The aforementioned feature fusion refers to establishing connections between core dimension data after data cleaning, standardization, and quantification according to preset business logic, and integrating them with feature weights into a single structured feature vector. This ensures that the feature vector can not only comprehensively cover key influencing factors but also compactly represent the core sales status of the sales object.

[0087] For example, the preset business logic is inventory-demand-risk-return.

[0088] The aforementioned feature weights refer to the quantitative coefficients assigned to each core dimension of data after data cleaning, standardization, and quantification during the feature fusion process. These coefficients reflect the importance of the data in representing the sales status and are primarily used to highlight key influencing factors, suppress redundant data interference, and ensure the accuracy of the feature vector representation.

[0089] Specifically, the weights are determined based on historical data statistical analysis: dimensions that have a greater impact on the accuracy of sales status prediction are assigned higher weights; auxiliary dimensions that contribute less to the prediction are assigned lower weights. Dynamic adjustments are also supported, allowing for real-time optimization of weight allocation based on changes in business scenarios, such as holiday demand fluctuations and feedback from historical prediction accuracy. This ensures that the feature fusion results always align with the core needs of overbooking decisions. For example, in a scenario where the sales target is airline tickets, the probability of certain waste has the greatest impact on overbooking decisions, with a weight of 35%; occupancy rate is second, with a weight of 30%; the 24-hour booking growth rate has a weight of 20%; and unit sales revenue has a weight of 15%. If it is during peak travel periods such as Spring Festival or National Day, the booking growth rate weight of the demand dimension can be dynamically increased to 25%.

[0090] The aforementioned sales status refers to the comprehensive status of the current inventory utilization, market demand, booking progress, and idle risk level of the sales object. For example, the current occupancy rate of an economy class flight, the 16% growth rate of bookings in the past 18 hours, the 70% of bookings made 5 days in advance, and the 15% probability of certain waste, all combine to form a sales status of saturated inventory, stable demand, and controllable risk. In this case, the overselling decision can be made to appropriately increase the scale of overselling.

[0091] The aforementioned inventory utilization refers to indicators reflecting inventory utilization efficiency, such as the ratio of pre-ordered quantity to the rated quantity of the sold items, the remaining available inventory size, inventory turnover efficiency, and the inventory consumption rate in the same period of history.

[0092] The aforementioned market demand refers to indicators reflecting market attention and purchasing intentions, such as the real-time pre-order growth rate of the product, market search volume, frequency of user inquiries, click volume for price comparisons of similar products, and concentration of demand in different regions.

[0093] The aforementioned booking progress refers to indicators reflecting the pace of booking progress, such as the current booking volume as a percentage of the total expected booking volume, the distribution of the interval between booking time and usage time, and the booking ratio of different booking channels.

[0094] The aforementioned idle risk level refers to a graded risk level based on the probability of deterministic waste, such as low risk: probability < 10%, medium-low risk: 10% ≤ probability < 20%, medium risk: 20% ≤ probability < 30%, and high risk: probability ≥ 30%. This is used to intuitively quantify the degree of possibility that the sold object will be idle due to the user not actually using it.

[0095] S404: Based on the sales status characteristics, predict the oversold quantity of the sales object to obtain the candidate oversold quantity range.

[0096] In some embodiments, the upper limit of the candidate oversold quantity range is obtained based on the product of the deterministic waste probability in the sales status characteristics and the rated quantity of the sales object; the value between zero and the upper limit value is determined as the candidate oversold quantity range.

[0097] The rated quantity of the sold item refers to the maximum number of the sold item that can be used. For example, when the sold item is an airline ticket, the rated quantity is the number of physical seats on the corresponding flight.

[0098] Optionally, if the upper limit of the above-mentioned candidate oversold quantity range is greater than the compliance threshold, the compliance threshold shall be used as the upper limit of the candidate oversold quantity range.

[0099] The aforementioned compliance threshold refers to the threshold set by legal or industry rules to constrain the upper limit of overbooking. For example, when the sold item is an airline ticket, the compliance threshold must comply with the requirements of civil aviation industry regulations and is usually set at 5% of the rated number of seats on the corresponding flight. Some regions or airlines may make minor adjustments based on route type and flight density, but it must not exceed the legal limit. If the rated number of seats on a flight is 50, its compliance threshold is 7 seats. If the candidate overbooking limit calculated based on the deterministic waste probability and the rated number is 10 seats, then 7 seats will be used as the upper limit of the candidate overbooking range to ensure that overbooking behavior complies with legal compliance requirements and avoids the risk of violations.

[0100] S405: Select candidate oversold quantities that meet the selection criteria from the candidate oversold quantity range as the predicted oversold quantity.

[0101] The above screening criteria refer to the dual constraints of maximizing profits and controlling risks, specifically including: the expected total cost of overselling conflicts ≤ the preset cost threshold, and the overselling risk level ≤ the preset risk level.

[0102] For example, if the candidate overbooking quantity for a certain flight is 0-7 seats, and the candidate overbooking quantity is 5 seats, the expected total cost of the overbooking conflict is 5×(1-20%)×1200=4800 yuan, which is lower than the preset cost threshold of 5000 yuan. The overbooking risk level is level 2, which is lower than the preset risk level 3. Therefore, 5 seats meet the screening criteria. If the candidate overbooking quantity is 7 seats, the expected total cost is 7×(1-20%)×1200=6720 yuan, which exceeds the preset cost threshold, or the overbooking risk level reaches level 4, which is higher than the preset risk level. Therefore, 7 seats do not meet the screening criteria and need to be eliminated.

[0103] The aforementioned expected total cost refers to all costs incurred due to overselling causing users to be unable to use the sold items normally, based on the number of candidate oversold items, the probability of certain waste, and historical operating data. This includes both direct economic costs and indirect implicit costs.

[0104] The aforementioned preset cost threshold refers to the maximum acceptable total cost limit for overbooking conflicts set by enterprises based on their own operating budget, profit targets, risk tolerance, and industry cost levels. It serves as the cost floor for constraining overbooking decisions. For example, when selling air tickets, the unit conflict cost of business class tickets is higher, so the preset cost threshold can be set at 30,000 yuan per flight; the unit conflict cost of economy class tickets is lower, so the preset cost threshold can be set at 15,000 yuan per flight. If the historical average overbooking conflict cost of a certain route is consistently lower than 80% of the threshold, it can be dynamically increased by 10%-15%. If the historical cost frequently exceeds the threshold, it can be reduced to 12,000 yuan.

[0105] For example, the preset cost threshold for high-priced items can be appropriately increased, while that for low-priced items can be appropriately decreased. Furthermore, dynamic adjustments based on historical cost data are supported. When selling air tickets, the preset cost threshold for international long-haul economy class is higher than that for domestic short-haul economy class because the cost of alternative services for international routes is higher. During holidays, the preset cost threshold for air tickets can be increased by 20% compared to weekdays because the risk of user complaints and compensation costs are higher during holidays, and more cost buffer space needs to be reserved.

[0106] The above-mentioned overselling risk level refers to a graded risk level that is comprehensively evaluated based on multiple dimensions of indicators such as the number of candidate oversold items, overselling compliance threshold, historical overselling conflict rate, probability of user complaints, and compliance penalty risk. It is used to intuitively quantify the degree of operational, compliance, and brand risks that overselling behavior may cause.

[0107] The aforementioned preset risk levels refer to the highest acceptable overbooking risk level set by the enterprise based on legal compliance requirements, brand positioning, user experience standards, and industry risk benchmarks. These levels represent the red line for overbooking decisions. For example, when selling airline tickets, the preset risk levels are 1-5, with level 5 being the highest. Therefore: 2 candidate overbooked seats, historical overbooking conflict rate of 0.8%, user complaint probability of 0.3%, and no compliance penalty record are all rated as Level 1; 6 candidate overbooked seats, historical conflict rate of 4.5%, complaint probability of 2.8%, and one minor compliance warning record are all rated as Level 3; 7 candidate overbooked seats, historical conflict rate of 8%, complaint probability of 5%, and compliance penalty record are all rated as Level 5.

[0108] For example, sales scenarios involving public safety typically have a lower risk level than ordinary consumer scenarios, ensuring that risks are controllable and meet core business requirements.

[0109] In some embodiments, at least two candidate oversold quantities are determined from the candidate oversold quantity range.

[0110] For example, discrete candidate values ​​are selected within the candidate overbooking quantity range according to a preset step size: if the upper limit of the candidate overbooking quantity range is ≤50, the step size is 1; if the upper limit is >50, the step size is 2%-5% (rounded up) of the upper limit. Taking air tickets as an example, if the candidate overbooking upper limit calculated based on the deterministic waste probability and the rated quantity for a certain flight is 28, then the candidate overbooking quantity is selected as 0, 1, 2...28 with a step size of 1; if the candidate overbooking upper limit for another flight is 65, then the step size is 2%-5% (rounded up) of the upper limit. The discrete candidate values ​​selected according to this step size include 0, 2, 4...65 (the specific value is determined according to the step size calculation result).

[0111] As one feasible approach, the above-mentioned method of selecting candidate oversold quantities that meet the screening criteria from the range of candidate oversold quantities as the predicted oversold quantity includes the following steps: Step 1: Calculate the revenue value corresponding to each candidate oversold quantity in the candidate oversold quantity list.

[0112] Among them, the revenue value refers to the revenue generated from the over-sale of the sold item.

[0113] Specifically, revenue = additional sales revenue from oversales - total expected cost of oversales conflict; additional sales revenue from oversales = number of candidate oversales × unit sales revenue of the target seller; total expected cost of oversales conflict = number of candidate oversales × (1 - probability of deterministic waste) × unit cost of oversales conflict.

[0114] The aforementioned total expected cost of overbooking conflict refers to the sum of all conflict-related costs that must be borne when some users are unable to use the sold items as booked due to overbooking, determined based on the number of candidate overbookings and a 1-certainty waste probability. It essentially reflects the quantified cost of the risk of overbooking. Taking air tickets as an example, it refers to all related costs that the airline must pay when overbooking causes some booked passengers to be unable to board on time. This includes direct compensation to passengers who do not board, the cost of rebooking subsequent flights, direct expenses such as temporary accommodation and catering subsidies, as well as the cost of brand reputation loss caused by passenger complaints and negative public opinion. It comprehensively quantifies the actual impact of overbooking conflict on operations.

[0115] Optionally, the aforementioned unit conflict cost is a weighted average of cash compensation, alternative service costs, and brand reputation loss costs. The weights are determined based on historical data statistics. Taking air tickets as an example, cash compensation corresponds to direct economic expenditures such as cash subsidies and mileage compensation for passengers who did not board the plane; alternative service costs correspond to seat purchase costs for passengers who changed their tickets to other flights, temporary accommodation and transportation subsidies provided, etc.; and brand reputation loss costs correspond to the quantified amount of indirect losses such as user complaints, decreased repurchase intentions, and negative public opinion caused by overbooking conflicts.

[0116] It should be understood that the weighting should be based on the frequency, impact, and quantifiability of various costs in historical overselling conflict cases: cash compensation occurs most frequently and is the most quantifiable, typically accounting for 40%-50% of the weight; alternative service costs are next, accounting for 30%-40%; and brand reputation loss costs, although indirect, affect long-term revenue, accounting for 10%-20%. Furthermore, the weighting should be dynamically iterated based on business scenarios and user group characteristics.

[0117] Step II: Select candidate oversold quantities from the revenue values ​​corresponding to each candidate oversold quantity, and use them as the predicted oversold quantity.

[0118] In some embodiments, the candidate oversold quantity with the highest return value that meets the screening criteria is selected first; if there are multiple candidate oversold quantities with the same return value, the value closest to the deterministic waste probability × the rated quantity is selected as the predicted oversold quantity to ensure that the oversold scale is accurately matched with the idle risk; if the return value of all candidate oversold quantities is negative, 0 is selected as the predicted oversold quantity to avoid losses.

[0119] The embodiments provided in this application determine the candidate oversold quantity by determining the deterministic waste probability and the rated quantity of the sales object, and then obtain the predicted oversold quantity based on the revenue value, so that the oversold quantity is accurately matched with the idle risk and compliance requirements, taking into account both short-term benefits and long-term operational efficiency, and effectively reducing the conflict between resource idleness and oversold.

[0120] like Figure 5 As shown, Figure 5 The following is a flowchart illustrating another overselling management method provided in this application embodiment. The specific steps of the overselling management method provided in this application embodiment are as follows: S501: Obtain the first information corresponding to the sales target.

[0121] S502: Based on the first piece of information, predict the probability of certainty waste of the target for sale.

[0122] S503: Based on the deterministic waste probability and first information, predict the oversold quantity of the sales object to obtain the predicted oversold quantity of the sales object.

[0123] S504: When the number of items acquired for sale exceeds the rated number of items for sale, overselling management shall be implemented for users who have acquired items for sale but have not used them.

[0124] In some embodiments, the inventory of the saleable item in the reservation system is adjusted based on the predicted oversold quantity.

[0125] The total number of sales targets corresponding to the above inventory is the sum of the basic inventory of the sales targets and the predicted overbooking quantity. Taking air tickets as an example, if a flight has 150 physical seats in economy class and the predicted overbooking quantity is 6 seats, then the real-time available inventory limit of the booking system is set to 156 seats after adjustment. When users query through the airline's official website, APP, third-party platforms and other channels, the number of available economy class seats displayed will be updated to 156 seats until the actual booking quantity reaches this limit, so as to achieve accurate linkage between the overbooking scale and the inventory display.

[0126] Specifically, the execution module of the oversold management system is connected to the inventory management interface of the reservation system. The sum of the calculated basic inventory and the predicted oversold quantity is set as the real-time available inventory limit. At the same time, the inventory display logic of the reservation system is updated synchronously to ensure that the available reservation quantity seen by the user is consistent with the adjusted inventory. The inventory adjustment supports scheduled updates or triggered updates.

[0127] As one feasible approach, the aforementioned overselling management to users who have acquired but not used the saleable items includes the following steps: Step 1: Send compensation invitations to users who have acquired the sale object but have not used it.

[0128] The aforementioned users who have acquired the sale object but have not used it refer to users who have obtained the right to use the sale object through legal means such as reservation or purchase, and who, as of the time the compensation invitation was sent, have not actually used the sale object, nor have they voluntarily terminated or given up their right to use it.

[0129] For example, when selling airline tickets, a passenger who has successfully booked and paid for a flight but has not processed a refund or rebooking within 24 hours of the flight's departure and has not arrived at the airport to complete the check-in process falls into this category. Similarly, a user who has booked a hotel room (with a check-in date of 3 days) and has not canceled the reservation or checked in 1 day before check-in also falls into this category. Furthermore, when selling concert tickets, a user who purchases tickets for a performance but has not refunded them before the performance begins and has not attended the show also meets this user definition.

[0130] Sending compensation invitations to users who have acquired the sale item but have not used it includes the following steps: (1) Obtain the associated information of users who have acquired the sale object but have not used the sale object.

[0131] The aforementioned related information is used to characterize relevant data that contributes to the revenue of the sold items, specifically including: booking characteristic data, user attribute data, usage / trip characteristic data, and historical behavior data; specifically, the related information of users who have acquired the sold items but have not used them is extracted in batches from the database of the overbooking management system, and deduplication and completion processing is performed to ensure data integrity.

[0132] Booking characteristics data include: booking time, booking channel, payment amount, and refund / cancellation policy type.

[0133] User attribute data includes: membership level, historical purchase frequency, and cumulative purchase amount.

[0134] Usage / trip characteristic data includes: expected usage time, whether there are related service bookings, and feasibility of changing the usage location / trip.

[0135] Historical behavioral data includes: historical refund and change rates, historical records of accepting oversale adjustments, etc.

[0136] For example, when the item being sold is airline tickets, the specific example of the associated information is as follows: Booking characteristic data: booking time, booking channel, full-price ticket payment amount, ticket refund and change rules; User attribute data: airline membership level, number of historical flights and amount of historical flight consumption; Usage / trip characteristic data: flight departure time, whether there are related service bookings, trip location / feasibility of trip changes; Historical behavior data: historical ticket refund and change rate, previous receipt of overbooking and rebooking compensation, etc.

[0137] (2) Calculate elastic ratings for users who have acquired the sales object but have not used the sales object based on the associated information.

[0138] Among them, the resilience score is used to characterize the degree to which users accept and adapt to the adjustments.

[0139] The aforementioned acceptance of adjustment means that the user voluntarily gives up the right to use the originally reserved item and instead chooses an alternative or directly gives up the right to use it.

[0140] In some embodiments, a weighted summation algorithm can be used to calculate the elasticity score, with a preset weight for each related information dimension. This weight can be dynamically adjusted according to the business scenario. Specifically, the related information of each dimension is first standardized into a quantitative value in the range of [0,1]. For example, historical adjustment records: 0 if there are any, 1 if there are none; feasibility of trip changes: 0 if high, 1 if low. Then, a weighted sum is calculated according to the preset weight to obtain the elasticity score, which ranges from [0,1]. The lower the score, the higher the degree of adaptation to the adjustment.

[0141] For example, the weights of the aforementioned preset related information dimensions are as follows: historical adjustment records weight 0.3, itinerary change feasibility weight 0.25, membership level weight 0.2, booking time weight 0.15, and refund / change rule type weight 0.1.

[0142] (3) Based on the elasticity score, sort the users who have obtained the sales object but have not used the sales object, and generate a priority compensation invitation queue.

[0143] The aforementioned priority compensation invitation queue refers to a list of invitation recipients sorted according to the user's acceptance of adjustment and adaptation. It is used to guide the invitation order for oversold compensation, maximizing the invitation success rate and reducing compensation costs.

[0144] Specifically, users who have acquired sales objects but have not used them are sorted in ascending order according to their elasticity scores, and an initial compensation invitation queue is generated.

[0145] Optionally, for users with the same elasticity score, a secondary sorting can be performed based on the rules of membership level from high to low or booking time from late to early, to optimize user experience and balance revenue.

[0146] For example, in the context of air travel, airline membership levels can be divided into: Regular Member, Silver Member (flight mileage ≥ 20,000 km or spending ≥ 10,000 RMB in the past 12 months), Gold Member (flight mileage ≥ 50,000 km or spending ≥ 30,000 RMB in the past 12 months), Platinum Member (flight mileage ≥ 100,000 km or spending ≥ 80,000 RMB in the past 12 months), and Diamond Member (flight mileage ≥ 200,000 km or spending ≥ 150,000 RMB in the past 12 months). Similarly, in the context of hotel accommodation, where hotel rooms are the primary purchase, hotel membership levels can be divided into: Regular Member (available upon registration), Silver Member (stays ≥ 5 nights or spending ≥ 2,000 RMB in the past 12 months), Gold Member (stays ≥ 15 nights or spending ≥ 8,000 RMB in the past 12 months), Platinum Member (stays ≥ 30 nights or spending ≥ 20,000 RMB in the past 12 months), and Black Gold Member (stays ≥ 60 nights or spending ≥ 50,000 RMB in the past 12 months).

[0147] (4) Initiate compensation invitations to users in the order of the initial compensation invitation queue.

[0148] In some embodiments, the user's response to the initial compensation offer is obtained, and the response includes at least: the user's acceptance status of the initial compensation offer and the requested compensation amount.

[0149] Based on the user's acceptance status of the initial compensation invitation and the requested compensation amount, the initial compensation invitation queue is rearranged to obtain the priority compensation invitation queue.

[0150] The above acceptance status includes: Accept, Disaccept, and Pending.

[0151] Specifically, users in the initial compensation invitation queue are first arranged according to their acceptance status, with users in the acceptance status at the top of the queue, followed by users in the pending status, and finally users in the non-acceptance status.

[0152] Furthermore, based on the amount of compensation requested by the users, the queue after the first arrangement is arranged in descending order to obtain the priority compensation invitation queue.

[0153] Based on this, compensation invitations are sent to users in the order of the priority compensation invitation queue.

[0154] The aforementioned compensation offer includes at least one alternative and corresponding compensation rights.

[0155] Specifically, alternative solutions are generated according to the type of the target audience: for example, in the travel scenario, they can provide rebooking of subsequent flights on the same route and cabin upgrades; in the hotel scenario, they can provide extended stays and room upgrades for the same level of rooms; and in the cultural and tourism scenario, they can provide spare seats for the same event and ticket exchanges for other events.

[0156] Optionally, compensation benefits include one or more combinations of cash subsidies, platform points, no-threshold coupons, and free value-added services; invitations are sent simultaneously through multiple channels such as application push, SMS, and email, with a one-click response entry (accept / reject / inquiry).

[0157] It should be understood that the user response status is tracked in real time during the invitation process. If the previous user refuses the invitation, the invitation to the next user is automatically triggered until the number of users who actually confirm the use is less than or equal to the rated number of the sales target. At the same time, a manual intervention interface is reserved. When the automatic invitation fails to reach the target, customer service personnel can make telephone invitations based on the queue to ensure the effectiveness of the oversold response.

[0158] For example, when selling airline tickets, a flight with a capacity of 150 seats has 6 seats short due to overbooking. The initial compensation invitation queue includes 20 passengers who have booked but not yet boarded. After obtaining user responses, 8 users explicitly "accept" the invitation, of which user A requests 800 yuan, user B requests 500 yuan, user C requests 600 yuan, and user D requests 700 yuan. 5 users are in a "pending" state, and 7 users "do not accept" the invitation.

[0159] First, users are ranked according to their acceptance status: the 8 users in the acceptance status are ranked first, followed by the 5 users in the pending status, and finally the 7 users who do not accept. Then, users in the acceptance status are ranked according to the amount of compensation requested, from largest to smallest, forming a priority compensation invitation queue: User A (800 yuan) → User D (700 yuan) → User C (600 yuan) → User B (500 yuan) → other users in the acceptance status → users in the pending status → users who do not accept.

[0160] The compensation offer for the flight includes alternative options: (1) rebooking to a subsequent flight on the same route at 16:00 on the same day; (2) upgrading to business class on a flight at 10:00 the next day; (3) full refund and a free rebooking voucher; the compensation benefits package is "cash subsidy (300-800 yuan, adapted according to user requirements), 5000 platform points, and one experience in the airport VIP lounge". The offer is sent simultaneously via the airline's APP push, SMS and email. The SMS includes "one-click acceptance" and "contact customer service" links to facilitate quick response from users.

[0161] When the invitation is executed, the system first sends an invitation to user A, who declines within 15 minutes. The system then automatically triggers an invitation for user D, who accepts within 30 minutes. The system immediately rebooks the user on the 16:00 flight, simultaneously sending a confirmation SMS and a VIP lounge access code, and reclaims the user's original 14:00 flight seat. Invitations continue sequentially until 6 seats are successfully reclaimed, reducing the number of confirmed boarding users to 150. If the automatic invitation only reclaims 4 seats, the remaining 2 seats trigger manual intervention. Customer service contacts the users ranked higher in the "pending" status and adjusts compensation benefits via phone, such as adding a 200 RMB cash subsidy. Ultimately, all seats are reclaimed, ensuring normal flight operation and protecting user rights.

[0162] Step II: Based on the user's response to the compensation invitation, identify the target user from the users who sent the compensation invitation.

[0163] The target user refers to a user who, upon receiving a compensation offer, voluntarily accepts the offer within a preset response time limit, explicitly agrees to relinquish the right to use the original product, and acknowledges the alternative or compensation offer in the offer.

[0164] The aforementioned preset response time limit refers to the maximum time interval set by the overselling management system based on the urgency of the sales object's usage time, the efficiency requirements for overselling processing, and user response habits. It allows users to decide whether to accept a compensation invitation after receiving it. Its core purpose is to balance the user's decision-making cycle with the timeliness of resolving overselling issues.

[0165] The specific settings are based on factors such as the proximity of the target user's usage time, the urgency level of overselling conflicts, and the user's historical average response time. Dynamic adjustments are also supported. For example, the response time can be extended by 1-2 hours for elderly users and high-level members to avoid users missing the opportunity to respond due to the short time limit. This ensures that the response time meets both operational processing needs and the user experience.

[0166] Optionally, the specific screening logic can combine factors such as user resilience rating and response timeliness, prioritizing users with high adaptability and minimal impact on service order as target users, to ensure efficient and orderly handling of overselling.

[0167] Step 3: Reclaim the sales object owned by the target user.

[0168] Specifically, after identifying the target user, the system automatically unbinds the user's usage rights from the original sales object through the overbooking management system, reservation system, and service management system, and updates the inventory status of the sales object. At the same time, the system sends a rights recovery confirmation notification to the target user, clearly informing them that the original reservation has been cancelled, the time of issuance of compensation rights, and the method of use, ensuring the user's right to know and rights protection. The entire recovery process is recorded and an operation record is generated for verification, taking into account both process compliance and user experience.

[0169] The solution provided in this application provides a precise response to overselling scenarios when the quantity acquired by the seller exceeds the quota. This is achieved by sending compensation invitations to users who have reserved but not used the products, screening target users, and reclaiming their rights. The solution is based on voluntary acceptance, avoiding user conflicts and brand losses caused by forced reclamation, and protecting user rights. It also efficiently revitalizes idle resources, quickly resolves overselling pressure, balances operational order and user experience, and improves the compliance of overselling management.

[0170] Figure 6 This application provides a schematic diagram of the layered architecture of an overselling management system, as shown in the embodiments below. Figure 6 The layered architecture of the overselling management system shown includes: a data acquisition layer, a data processing and analysis layer, an overselling decision-making layer, and a compensation execution layer.

[0171] The aforementioned data acquisition layer is the data source support layer of the overselling management system, responsible for acquiring all the data required for overselling management, including: internal data interface module, external data interface module, and real-time data stream processing module.

[0172] The internal data interface module is used to connect with internal business systems and collect core business data such as sales information and user attribute data of the sales objects.

[0173] The external data interface module is used to collect external information that affects sales volume through methods such as web crawling and third-party application programming interfaces (APIs).

[0174] The real-time data stream processing module is used to clean, convert, and deduplicate the collected raw data in real time to ensure the timeliness and integrity of the data, and then pass the preprocessed data to the next level.

[0175] The aforementioned data processing and analysis layer is used for in-depth processing and analysis based on the output of the data acquisition layer, providing support for upper-level decision-making. It includes: data cleaning and feature engineering, overbooking prediction module, and passenger assessment module.

[0176] Data cleaning and feature engineering are used to further extract features and standardize the preprocessed data passed from the data acquisition layer, transforming the raw data into structured feature data that can be recognized by subsequent models.

[0177] The oversold prediction module integrates a deterministic waste prediction model and an oversold quantity prediction model. Based on the input feature data, it outputs core analysis results such as the deterministic waste probability of the sold item and the predicted oversold quantity.

[0178] The passenger assessment module is used to construct the feature dimensions required for user elasticity scoring based on users' booking characteristics, historical behavior and other related information, and to provide an assessment basis for user screening in subsequent overbooking response.

[0179] The aforementioned overselling decision-making layer is used to generate specific instructions for overselling management based on the results of the data processing and analysis layer. It includes: an overselling monitoring and triggering engine, an intelligent decision-making engine, and a compensation scheme management module.

[0180] The overbooking monitoring and triggering engine is used to monitor indicators such as reservation progress and overbooking quantity in real time. When the actual number of reservations exceeds the total basic inventory of the item being sold, the overbooking response process is triggered. The intelligent decision engine combines the results of the oversold prediction module with the oversold compliance threshold to generate optimal oversold strategies, such as inventory adjustment rules and compensation scheme matching logic. The compensation scheme management module is used to store and manage various oversold compensation schemes and match appropriate compensation content based on user evaluation results.

[0181] The aforementioned compensation execution layer is responsible for implementing the instructions from the oversold decision-making layer and completing actual business operations. It includes: a multi-channel notification gateway, a dynamic oversold ratio calculation layer, and a guarantee and settlement module.

[0182] A multi-channel notification gateway is used to push voluntary adjustment invitations to users who have reserved and are waiting to use the product and have not terminated their reservations, as well as users who have obtained the product but have not used it, through channels such as applications, SMS, and email, and to simultaneously track the user's response status.

[0183] Dynamic overbooking ratio calculation is used to dynamically adjust the overbooking ratio based on real-time reservation data and market demand fluctuations, adapting to real-time changes in business scenarios.

[0184] The guarantee and settlement module is responsible for follow-up services after overbooking, such as rescheduling guarantees, distribution of compensation rights and settlement of related fees, and feeding back the execution results to the overbooking decision-making level to form a closed loop of the whole process.

[0185] Figure 7 A schematic diagram of the logical structure of an overselling management method provided in this application embodiment is shown below. Figure 7The logical structure of the overselling management method shown includes four stages: data input, pre-overselling decision-making, compensation strategy matching, and implementation. The specific explanations of each stage are as follows: The data input stage is the foundation for overselling management decisions, and includes the following four types of data: Internal operational data refers to the company's own sales data, such as total basic inventory, real-time reservations, and unit sales revenue; market environment data refers to external environmental information affecting sales demand, such as market demand intensity and holiday customer flow trends; competitive landscape data refers to information on competing products of similar sales targets, such as competitors' pricing strategies and service benefits; and real-time dynamic data refers to real-time changes in the business process, such as real-time user booking progress and temporary resource adjustments. All of this data is uniformly transmitted to the subsequent overbooking calculation stage.

[0186] The pre-overselling decision-making process includes two steps: dynamic overselling quantity calculation and overselling judgment and triggering. Dynamic oversold quantity calculation: Receive the above four types of data as input, combine deterministic waste probability, and calculate the dynamic oversold quantity adapted to the current scenario through the oversold quantity prediction model; Oversold judgment and triggering: Based on the calculated dynamic oversold quantity, compare the actual number of reservations with the basic inventory of the sold items. If the actual number of reservations exceeds the basic inventory, the subsequent oversold compensation process will be triggered.

[0187] The compensation strategy matching process includes the following steps: Calculate acceptance probability: For users who have pre-ordered items but are still in the pending use state and have not terminated their pre-orders, calculate the probability that the user will accept the oversale adjustment based on their associated information; Match the optimal solution: Based on the acceptance probability results, match the compensation solution with the highest suitability for the user from the compensation solution library; Output volunteer list: Combining the results of the first two steps, filter out users with high suitability and generate a volunteer invitation list for oversale compensation.

[0188] The implementation phase is responsible for translating the compensation strategy into practical action, including the following steps: Automatic compensation execution: Receive the output list of volunteers and initiate the automated process for overselling compensation; App / SMS push: Push matched compensation invitations to users on the volunteer list through multiple channels; One-click acceptance for passengers: Users can accept compensation invitations through the response entry point of the push channel, ultimately realizing a closed loop for the entire process of dealing with overbooking.

[0189] Please see Figure 8 , Figure 8This is a schematic diagram of an overselling management device provided in an embodiment of this application. The overselling management device provided in this embodiment includes: an acquisition module 801, a first prediction module 802, a second prediction module 803, and an overselling management module 804.

[0190] The acquisition module 801 is used to acquire the first information corresponding to the sales object, the first information including the sales information of the sales object and / or information affecting the sales quantity of the sales object.

[0191] The first prediction module 802 is used to predict the deterministic waste probability of a sale object based on the first information. The deterministic waste probability refers to the probability that a sale object is reserved but is left idle.

[0192] The second prediction module 803 is used to predict the oversold quantity of the sales object based on the deterministic waste probability and the first information, so as to obtain the predicted oversold quantity of the sales object.

[0193] The first prediction module 802 is also used to process the first information to obtain the first information feature corresponding to the first information; input the first information feature into the probability prediction model for prediction to obtain the deterministic waste probability.

[0194] The second prediction module 803 is also used to process the deterministic waste probability and the first information to obtain the sales status characteristics of the sales object. The sales status characteristics are used to describe the sales status of the sales object. Based on the sales status characteristics, the oversold quantity of the sales object is predicted to obtain the candidate oversold quantity range. The candidate oversold quantity that meets the screening conditions is selected from the candidate oversold quantity range as the predicted oversold quantity.

[0195] The second prediction module 803 is also used to calculate the revenue value corresponding to each candidate oversold quantity in the candidate oversold quantity. The revenue value refers to the revenue brought by the oversold quantity of the sold object; and to select the candidate oversold quantity from the revenue value corresponding to each candidate oversold quantity as the predicted oversold quantity.

[0196] The second prediction module 803 is also used to obtain the upper limit of the candidate oversold quantity range based on the product of the deterministic waste probability in the sales status characteristics and the rated quantity of the sales object, where the rated quantity of the sales object refers to the maximum number of sales objects that can be used; and to determine the value between zero and the upper limit value as the candidate oversold quantity range.

[0197] This application provides an overbooking management device, which further includes an overbooking management module 804.

[0198] The overselling management module 804 is used to manage overselling to users who have acquired but not used the sales objects when the number of acquired sales objects exceeds the rated number of sales objects.

[0199] The overselling management module 804 is also used to send compensation invitations to users who have acquired sales objects but have not used them; based on the users' response information to the compensation invitations, to identify target users from the users who sent the compensation invitations; and to reclaim the sales objects owned by the target users.

[0200] It should be noted that, Figure 8 The module division shown is illustrative and represents only one logical functional division; in actual implementation, other division methods are possible. For example, two or more functions can be integrated into a single processing module. These integrated modules can be implemented either in hardware or as software functional modules.

[0201] In exemplary embodiments, as described above, the electronic device may specifically be an electronic device with computing processing capabilities, such as a computer or service. In this case, embodiments of this application also provide an electronic device. Figure 9 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 8 As shown, the electronic device includes: a processor 10, a memory 20, a communication line 30, a communication interface 40, and an input / output interface 50.

[0202] The processor 10, memory 20, communication interface 40, and input / output interface 50 can be connected via communication line 30.

[0203] Processor 10 is used to execute instructions stored in memory 20 to implement the flight scheduling method provided in the above embodiments of this application. Processor 10 can be a CPU, a network processor (NP), a digital signal processor (DSP), a microprocessor, a microcontroller (MCU) / single-chip microcomputer / microcontroller, a programmable logic device (PLD), or any combination thereof. Processor 10 can also be any other device with processing capabilities, such as a circuit, device, or software module; this application embodiment does not limit this. In one example, processor 10 may include one or more CPUs, for example... Figure 8 CPU0 and CPU1 in the example. As an optional implementation, the electronic device may include multiple processors; for example, in addition to processor 10, it may also include processor 60. Figure 9 (The example shown is a dashed line).

[0204] The memory 20 is used to store instructions. For example, the instructions may be computer programs. Optionally, the memory 20 may be a read-only memory (ROM) or other types of static storage devices that can store static information and / or instructions; it may also be a random access memory (RAM) or other types of dynamic storage devices that can store information and / or instructions; it may also be an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM), or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media, or other magnetic storage devices, etc. The embodiments of this application do not limit this.

[0205] It should be noted that the memory 20 can exist independently of the processor 10 or it can be integrated with the processor 10. The memory 20 can be located inside or outside the electronic device, and this application embodiment does not impose any restrictions on this.

[0206] Communication line 30 is used to transmit information between the components included in the electronic device.

[0207] Communication interface 40 is used to communicate with other devices or other communication networks. These other communication networks can be Ethernet, radio access network (RAN), wireless local area network (WLAN), etc. Communication interface 40 can be a module, circuit, transceiver, or any device capable of enabling communication.

[0208] Input / output interface 50 is used to enable human-computer interaction between users and electronic devices. For example, it enables action interaction or information exchange between users and electronic devices.

[0209] For example, the input / output interface 50 can be a mouse, keyboard, display screen, or touch screen. Action or information interaction between the user and the electronic device can be achieved through a mouse, keyboard, display screen, or touch screen.

[0210] It should be noted that, Figure 9 The structures shown do not constitute a limitation on electronic devices, except... Figure 8 In addition to the components shown, electronic devices may include more or fewer components than illustrated, or combinations of certain components, or different component arrangements.

[0211] In an exemplary embodiment, this application also provides a readable storage medium including software instructions that, when run on an electronic device, cause the electronic device to perform any of the methods provided in the above embodiments.

[0212] In an exemplary embodiment, this application also provides a computer program product containing computer execution instructions, which, when run on an electronic device, causes the electronic device to perform any of the methods provided in the above embodiments.

[0213] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented using software programs, implementation can be, in whole or in part, in the form of a computer program product. This computer program product includes one or more computer-executable instructions. When these computer-executable instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of this application is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer-executable instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer-executable instructions can be transmitted from one website, computer, server, or data center to another via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device containing one or more servers, data centers, etc., that can be integrated with the medium. The available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid-state disks, SSDs).

[0214] Although this application has been described herein in conjunction with various embodiments, those skilled in the art, by reviewing the accompanying drawings, disclosure, and appended claims, will understand and implement other variations of the disclosed embodiments in carrying out the claimed application. In the claims, the word "comprising" does not exclude other components or S, and "a" or "an" does not exclude a plurality. A single processor or other unit can implement several functions listed in the claims. While different dependent claims may recite certain measures, this does not mean that these measures cannot be combined to produce good results.

[0215] Although this application has been described in conjunction with specific features and embodiments, it is obvious that various modifications and combinations can be made thereto without departing from the spirit and scope of this application. Accordingly, this specification and drawings are merely exemplary illustrations of this application as defined by the appended claims, and are considered to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from the spirit and scope of this application. Thus, if such modifications and modifications of this application fall within the scope of the claims of this application and their equivalents, this application is also intended to include such modifications and modifications.

[0216] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. An overselling management method, characterized in that, The method includes: Obtain first information corresponding to the sales object, the first information including the sales information of the sales object and / or information affecting the sales quantity of the sales object; Based on the first information, predict the deterministic waste probability of the sale object, where the deterministic waste probability refers to the probability that the sale object is reserved but is left idle; Based on the deterministic waste probability and the first information, the oversold quantity of the sales object is predicted to obtain the predicted oversold quantity of the sales object.

2. The overselling management method according to claim 1, characterized in that, The step of predicting the oversold quantity of the sales object based on the deterministic waste probability and the first information, to obtain the predicted oversold quantity of the sales object, includes: Data processing is performed on the deterministic waste probability and the first information to obtain the sales status characteristics of the sales object, which are used to describe the sales status of the sales object; Based on the sales status characteristics, the oversold quantity of the sales object is predicted to obtain a candidate oversold quantity range; The candidate oversold quantity that meets the screening criteria is selected from the candidate oversold quantity range as the predicted oversold quantity.

3. The overselling management method according to claim 2, characterized in that, The step of selecting candidate oversold quantities that meet the selection criteria from the candidate oversold quantity range as the predicted oversold quantity includes: Calculate the revenue value corresponding to each candidate oversold quantity in the candidate oversold quantity, where the revenue value refers to the revenue brought by the oversold quantity of the sales object; The candidate oversold quantity is selected from the revenue value corresponding to each candidate oversold quantity, and is used as the predicted oversold quantity.

4. The overselling management method according to claim 2, characterized in that, Based on the sales status characteristics, the oversold quantity of the sales object is predicted to obtain a candidate oversold quantity range, including: Based on the product of the deterministic waste probability in the sales status characteristics and the rated quantity of the sales object, the upper limit of the candidate oversold quantity range is obtained, where the rated quantity of the sales object refers to the maximum number of the sales object that can be used. The range of values ​​between zero and the upper limit is determined as the candidate oversold quantity range.

5. The overselling management method according to claim 1, characterized in that, The step of predicting the probability of certainty of waste for the target of sale based on the first information includes: The first information is processed to obtain the first information feature corresponding to the first information. The first information feature is input into the probability prediction model for prediction to obtain the deterministic waste probability.

6. The overselling management method according to any one of claims 1 to 5, characterized in that, The method further includes: If the number of items acquired by the seller exceeds the rated quantity of the seller, overselling management is implemented for users who have acquired the seller but have not used the seller.

7. The overselling management method according to claim 6, characterized in that, The overselling management to users who have acquired the sale object but have not used it includes: Send compensation invitations to users who have acquired the sale object but have not used it; Based on the user's response to the compensation invitation, the target user is identified from the users who sent the compensation invitation; The sales object owned by the target user shall be reclaimed.

8. An electronic device, characterized in that, include: Processor and memory; The memory stores instructions that the processor can execute; When the processor is configured to execute the instructions, it causes the electronic device to implement the method as described in any one of claims 1 to 7.

9. A computer-readable storage medium, characterized in that, The storage medium stores a computer program or instructions, which, when executed by a processing device, implement the method of any one of claims 1 to 7.

10. A computer program product, comprising a computer program or instructions, characterized in that, When the computer program or instructions are executed on the processing device, the method of any one of claims 1 to 7 is implemented.

Images