Method for managing hotel inventory on hotel sales platform

Abstract

The invention relates to a method for managing hotel inventory on a hotel sales platform. The method comprises the steps that under the condition that no stock exists in a room type sold by a merchant on a hotel sales platform, the stock of a stock reference source of the room type is checked; under the condition that the inventory reference source has the inventory, estimating the probability of transaction of the house type after additional inventory; and determining whether to add the inventory to the house type or not based on the probability that the house type deals after the inventory is added. By using the method, the inventory can be accurately added, the selling rate of the hotel is improved, and extra room night quantity is brought.

Description

Technical Field

[0001] The present disclosure relates to the field of tourism products, and in particular to a method for managing hotel inventory on a hotel sales platform. Background Art

[0002] At present, there are many hotel merchants selling rooms on hotel sales platforms (such as travel service platforms that include hotel services for sale, or platforms that provide sales channel services specifically for hotel merchants, etc.). For hotels, in order to pursue the maximization of their room sales, hotel rooms are often sold through multiple sales channels (such as multiple hotel sales platforms or multiple merchants). There is a part of hotel merchants that do not interconnect the hotel's own hotel management system (which can manage room inventory, for example) with the backstage of the platform because of reasons such as privacy protection, that is, the actual inventory situation of the hotel is unknown to the platform. And some hotel merchants are not timely enough for the maintenance of the inventory on the platform, which may cause users who are interested in the hotel on the platform to be unable to buy the hotel's room even if the hotel actually has inventory, because there is no room to buy on the platform.

[0003] Therefore, for hotel merchants whose inventory cannot be truly grasped by hotel sales platforms, it is hoped to provide a method to assist merchants in adding inventory, which can increase the hotel's saleability and bring in additional room nights. Summary of the invention

[0004] A technical problem to be solved by the present disclosure is to provide a method for managing hotel inventory, which can accurately add inventory.

[0005] According to a first aspect of the present disclosure, a method for managing hotel inventory on a hotel sales platform is provided, comprising: when a room type sold by a merchant on the hotel sales platform is out of stock, checking the inventory of an inventory reference source of the room type; when the inventory reference source has inventory, estimating the probability of a transaction for the room type after additional inventory is added; and determining whether to add inventory for the room type based on the probability of a transaction for the room type after additional inventory is added.

[0006] Optionally, the probability of selling the room type after the additional inventory includes the order probability of the room type after the additional inventory, and / or the probability of the merchant accepting the order after the order is placed for the additional inventory, and when the order probability is greater than a first threshold and / or the probability of the merchant accepting the order is greater than a second threshold, it is determined that the room type is additionally stocked.

[0007] Optionally, the probability of selling the room type after the additional inventory includes the order probability of the room type after the additional inventory and the probability of the merchant accepting the order after the order is placed for the additional inventory, and the step of estimating the probability of selling the room type after the additional inventory includes: estimating the order probability of the room type after the additional inventory; and when the order probability is greater than a first threshold, estimating the probability of the merchant accepting the order after the order is placed for the additional inventory; and when the probability of the merchant accepting the order is greater than a second threshold, determining to add inventory for the room type.

[0008] Optionally, the step of checking the inventory of the inventory reference source of the room type includes: determining multiple candidate inventory reference sources for the room type; estimating the accuracy probability of the multiple candidate inventory reference sources as inventory reference sources respectively; and selecting a candidate inventory reference source whose accuracy probability exceeds a third threshold as the inventory reference source for the room type, and checking the inventory of the inventory reference source.

[0009] Optionally, the step of estimating the accurate probabilities of the multiple candidate inventory reference sources being respectively used as inventory reference sources includes: estimating the accurate probabilities of the inventory reference source types to which the multiple candidate inventory reference sources respectively belong as the accurate probabilities of the multiple candidate inventory reference sources respectively.

[0010] Optionally, the order probability is estimated using a trained first machine learning model.

[0011] Optionally, the probability of the merchant accepting the order is estimated using a trained second machine learning model.

[0012] Optionally, the inventory reference source type includes at least one of similar room types of similar hotels sold on the hotel sales platform, the same room types of the same hotel sold on other hotel sales platforms, and the same room types of the same hotel sold by other merchants on the hotel sales platform.

[0013] Optionally, the accurate probability of the inventory reference source type is estimated using the same trained third machine learning model, or the accurate probability of each of the multiple inventory reference source types is estimated using a corresponding one of the multiple trained third machine learning models.

[0014] Optionally, the input features of the first machine learning model include features representing the inventory status of an inventory reference source or multiple candidate inventory reference sources of the room type, features representing the historical order rate or historical transaction rate of the room type, features representing specific time information or specific time period information, and / or features representing the order rate or transaction rate of the room type at a specific time or specific time period, and the output of the first machine learning model is the order probability of the room type after additional inventory.

[0015] Optionally, the input features of the second machine learning model include features representing the inventory status of the inventory reference source of the room type or multiple candidate inventory reference sources, features representing the historical activity of the merchant on the hotel sales platform, features representing the historical order acceptance rate or historical order rejection rate of the room type or the merchant, features representing specific time information or specific time period information, features representing the activity of the merchant on the hotel sales platform at a specific time or specific time period, and / or features representing the order acceptance rate or rejection rate of the room type or the merchant at a specific time or specific time period, and the output of the second machine learning model is the probability of the merchant accepting the order.

[0016] Optionally, when only one third machine learning model is used to estimate the accuracy probability of each of the multiple inventory reference source types, the one third machine learning model is run multiple times to respectively estimate the accuracy probability of each of the multiple inventory reference source types, and the input features of the one third machine learning model include features representing the inventory reference source type estimated at the time, features representing the historical accuracy of the inventory reference source type estimated at the time, features representing specific time information or specific time period information, and / or features representing the accuracy of the inventory reference source type estimated at the time at a specific time or a specific time period, and the output of the one third machine learning model is the accuracy probability of the inventory reference source type estimated at the time.

[0017] Optionally, when using multiple third machine learning models to respectively estimate the accuracy probability of each of the multiple inventory reference source types, the multiple third machine learning models are trained for the multiple inventory reference source types respectively, and the input features of the multiple third machine learning models respectively include features representing the historical accuracy of the corresponding inventory reference source type, features representing specific time information or specific time period information, and / or features representing the accuracy of the corresponding inventory reference source type at a specific time or a specific time period, and the outputs of the multiple third machine learning models are respectively the accuracy probabilities of the corresponding inventory reference source types.

[0018] Optionally, the method further includes, after adding inventory to the room type and before the added inventory has been sold, performing an additional inventory check task on the room type to determine whether to cancel the inventory added for the room type, wherein the additional inventory check task includes: re-executing the step of checking the inventory of an inventory reference source of the room type, estimating the probability of selling the room type after adding inventory, and / or determining whether to add inventory to the room type, and in the event that the inventory reference source has no inventory or it is determined not to add inventory to the room type, determining to cancel the additional inventory for the room type.

[0019] Optionally, in a case where the room types for additional inventory include multiple room types in one or more hotels and the additional inventory inspection tasks are performed on the multiple room types respectively, an operations optimization model is constructed and solved to determine the inspection schedule of the additional inventory inspection tasks for the multiple room types within a first predetermined period in the future, wherein the objective of the operations optimization model is to maximize the total value of the additional inventory inspection tasks for the multiple room types to be inspected within the first predetermined period in the future, the value of the additional inventory inspection tasks for each room type in each inspection is determined based on the order probability of the corresponding room type within a second predetermined period in the future after the execution time of each inspection, and the constraints of the operations optimization model include the minimum number of times the additional inventory inspection tasks for each room type are to be inspected, the minimum number of intervals between inspections for the additional inventory inspection tasks for each room type, and / or an upper limit on the total number of times all the additional inventory inspection tasks to be inspected.

[0020] Optionally, the method further includes: before adding inventory for the room type, obtaining authorization from the merchant to add inventory.

[0021] Optionally, the method further includes: after adding inventory to the room type, prompting the merchant to indicate that the inventory has been added and presenting information about the added inventory.

[0022] Optionally, the method further includes: after adding inventory to the room type, providing the merchant with an entry or operation button for canceling the added inventory.

[0023] Optionally, the method also includes: after additional inventory is added to the room type and the additional inventory is ordered, prompting the merchant to indicate that an order has been generated for the additional inventory, allowing the merchant to reject the order, and evaluating the merchant's operations on the order for the additional inventory differently from the operations on the order for the inventory set by the merchant itself.

[0024] Optionally, the method also includes: after adding inventory for the room type, receiving order information of the user for the added inventory via a client of the hotel sales platform for users and transmitting the order information to a server of the hotel sales platform; and when the merchant accesses the merchant backend website of the hotel sales platform, based on the order information obtained by the server, prompting the merchant via the merchant backend website to indicate that an order has been generated for the added inventory, and receiving information about whether the merchant accepts the order.

[0025] According to a second aspect of the present disclosure, a computing device is provided, comprising: a processor; and a memory on which executable codes are stored, and when the executable codes are executed by the processor, the processor executes the method described in the first aspect above.

[0026] According to a third aspect of the present disclosure, a computer program product is provided, comprising an executable code, and when the executable code is executed by a processor of an electronic device, the processor is caused to execute the method as described in the first aspect above.

[0027] According to a fourth aspect of the present disclosure, a non-transitory machine-readable storage medium is provided, on which executable code is stored. When the executable code is executed by a processor of an electronic device, the processor executes the method described in the first aspect above.

[0028] Therefore, by using the method according to the present disclosure, it is possible to accurately add inventory, improve the hotel's saleability, and bring in additional room nights. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The above and other objects, features and advantages of the present disclosure will become more apparent through a more detailed description of exemplary embodiments of the present disclosure in conjunction with the accompanying drawings, wherein like reference numerals generally represent like components in the exemplary embodiments of the present disclosure.

[0030] Figure 1 A schematic flow chart of a method for managing hotel inventory on a hotel sales platform according to at least one embodiment of the present disclosure is shown.

[0031] Figure 2 A schematic flow chart of a method for managing hotel inventory on a hotel sales platform according to at least one embodiment of the present disclosure is shown.

[0032] Figure 3 An exemplary operation interface for a hotel merchant to perform inventory management on a travel service platform according to at least one embodiment of the present disclosure is shown.

[0033] Figure 4 A schematic diagram of the structure of a computing device according to at least one embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0034] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to make the present disclosure more thorough and complete, and to fully convey the scope of the present disclosure to those skilled in the art.

[0035] First, some explanations of terms involved in this disclosure are given:

[0036] Hotel sales platform: refers to a platform that allows multiple hotel merchants to settle in and provide hotel room sales services to users of the platform. It can be, for example, a travel service platform that includes the sale of hotel services, or a platform that specifically provides sales channel services for hotel merchants, etc. In this disclosure, unless otherwise specified, "user" generally refers to a consumer of hotel services who browses, inquires, books or purchases hotel services on the hotel sales platform, etc.;

[0037] Hotel Management System: A software system used to manage the daily operations of a hotel, which covers multiple aspects such as room management, reservation management, front desk management, bill management, inventory management and financial management, aiming to improve the hotel's work efficiency and service quality;

[0038] Backstage of hotel sales platform / travel service platform: the backstage system of hotel merchants opening stores on the hotel sales platform / travel service platform, which has similar functions to hotel management system, mainly for hotel price / inventory / order maintenance, etc. In addition, the backstage can be implemented through a website, which can also be called the hotel backstage website;

[0039] Room type: refers to a certain type of room in the hotel, such as a king-size bed room, a twin bed room, etc. Rooms of the same room type usually have the same or similar selling prices;

[0040] Inventory: in the solution of the present disclosure, it usually refers to the inventory of a room type sold by a hotel merchant on a hotel sales platform, which reflects the number of rooms available for sale of the room type. Inventory means that there are rooms available for sale of the room type, and no inventory means that there are no rooms available for sale of the room type or the room type has been sold out;

[0041] Room type sold out: refers to the number of rooms of a certain room type set by the hotel merchant on a hotel sales platform (including travel service platform) has been sold out, which can also be called the room type is out of stock;

[0042] Room type inventory reference source: refers to any source that can be used as a reference or reference object for the inventory of this room type, which can reflect the inventory of this room type to a certain extent;

[0043] Adding inventory: refers to increasing inventory, so that the sold-out rooms are available again;

[0044] Similar hotels: Hotels that have a certain degree of similarity with the target hotel can be called similar hotels. For example, if the similarity between them and the target hotel in one or some characteristics is the highest or exceeds the threshold, such as close geographical location, similar selling prices, and similar hotel grades (such as grades rated or defined according to certain rules), they can be considered similar hotels;

[0045] Similar room types: Room types that have a certain degree of similarity with the target room type can be called similar room types. For example, the similarity between the room type and the target room type in one or some characteristics is the highest or exceeds the threshold. For example, the room types are the same or similar (such as both are king-size rooms or belong to a certain type of king-size room) and the selling prices are similar.

[0046] Transaction: refers to the transaction completed when the buyer and seller reach an agreement. A complete transaction process includes two parts: the user places an order and the merchant accepts the order.

[0047] Machine learning model: It uses computers to learn model parameters from training data and can predict and analyze data after training;

[0048] Deep learning model: refers to the deep neural network machine learning model;

[0049] Inspection: refers to a scheduled task that is executed at regular intervals and polls multiple tasks each time, such as checking a specific state or condition to determine whether an operation needs to be performed or a corresponding measure needs to be taken;

[0050] Operational optimization: In mathematics and computer science, it is a problem of finding the best solution from all feasible solutions, such as the knapsack problem, which can be in the following form:

[0051]

[0052] subject to g i (x)≤0, i=1, ..., m

[0053] h j (x) = 0, j = 1, ..., p

[0054] Operations research optimization model: a mathematical tool used to describe and solve the problem of how to find the best solution under given conditions; it usually includes one or more objective functions (such as cost, benefit, time, etc.), as well as constraints on these objective functions. Various methods can be used to solve the operations research optimization model to find the best solution that meets the constraints, so as to achieve goals such as maximizing benefits, minimizing costs, and optimizing resource allocation.

[0055] As mentioned above, the present disclosure has discovered that, due to problems such as inventory management on hotel sales platforms, it is possible that even if a hotel room is actually available, users on the platform who are interested in it cannot purchase a room at the hotel because there are no rooms available for purchase on the platform.

[0056] To address this issue, the platform can manage the hotel inventory on the platform in the following ways:

[0057] ·Use simple rules to control room tracking tasks: For example, use predetermined rules to exclude high-risk merchants, and for other merchants, if the room types they sell are sold out, they are forced to add inventory; this method is simple, but it is easy to bring more false inventory and there is a greater risk of risk control. If the order is rejected after the inventory is added, it will affect the user experience and damage the merchant's service indicators; that is, in many cases, the hotel actually has no inventory, but the platform still opens the inventory for users to place orders. As a result, the merchant has to reject the order and the user cannot complete the transaction, leading to complaints from users and merchants.

[0058] Therefore, the present disclosure proposes a new method for managing hotel inventory, which determines whether to add inventory by referring to the inventory of the inventory reference source of the sold-out room type and further estimating the probability of the transaction after adding inventory. It can accurately add inventory, improve the hotel's saleability, bring additional room nights, and at the same time avoid merchants rejecting orders as much as possible, thereby avoiding complaints from merchants and users.

[0059] In addition, for hotel merchants whose inventory cannot be truly mastered by the hotel sales platform, such as hotel merchants whose hotel management system is not interconnected with the platform, the hotel sales platform can assist merchants to automatically add inventory and minimize the impact on merchants through the solution disclosed in this disclosure. For example, an agreement can be signed with the merchant in advance to obtain the merchant's authorization to agree to add inventory. If a room type of the merchant is sold out, if it is judged through the solution disclosed in this disclosure that the room type actually has inventory and has the value of adding inventory (i.e., a possible transaction), then the room type will be added to the inventory, for example, the inventory of the room type will be added by one, and after the additional inventory, the merchant can be prompted in various ways, allowing it to cancel the additional inventory and the orders generated by it without any negative comments on the merchant. This can effectively assist merchants in managing inventory, improve the hotel's saleability, bring extra room nights, without adversely affecting merchants, and avoid complaints from merchants and platform users.

[0060] At least some embodiments of the solutions disclosed herein will be described in detail below with reference to the accompanying drawings.

[0061] Figure 1 A schematic flow chart of a method for managing hotel inventory on a hotel sales platform according to at least one embodiment of the present disclosure is shown.

[0062] In some cases, the hotel sales platform may include a server, a client for users, and a merchant backend website for merchants to access. For example, the client for users may be an application corresponding to the hotel sales platform, which may be installed on a mobile smart device such as a mobile phone or a terminal device such as a computer, through which users may perform various operations related to purchasing hotel services; the merchant backend website for merchants to access may be a merchant backend website corresponding to the hotel sales platform, which may be accessed using various browsers on a mobile smart device such as a mobile phone or a terminal device such as a computer, through which merchants may maintain / manage / operate the hotel services sold on the hotel sales platform, etc.; the server may be a software application that provides backend services for the application and / or merchant backend website of the hotel sales platform, which may be installed on the corresponding server device.

[0063] Therefore, in some cases, each step in the various methods of the embodiments of the present disclosure described in detail below can be implemented by the service end of the hotel sales platform, the client for use by the user, or the merchant backend website for the merchant to access, as needed, or can be completed by at least two of the client for use by the user, the merchant backend website for the merchant to access, and the service end in collaboration. For example, various steps of providing / presenting information to users / merchants, steps of receiving information from users / merchants, or steps of interacting with users / merchants in the scheme of the present disclosure can be implemented by the client for use by the user or the merchant backend website for the merchant to access, and various other operations / steps can be performed by the client for use by the user, the merchant backend website for the merchant to access, and / or the service end as needed, and the present disclosure has no limitation on this.

[0064] like Figure 1 As shown, in step S110, when the room type sold by the merchant on the hotel sales platform is out of stock, the inventory of the room type is checked against the inventory of the reference source.

[0065] As mentioned above, in some cases, the inventory of the room types sold on the platform does not reflect their actual inventory situation. Sometimes the room type actually still has inventory but the platform shows that it is sold out. The actual inventory situation of these sold-out room types can be reflected in some ways, such as through the inventory reference source of the room type. If the reference source still has inventory, then the hotel's sold-out room type may still have inventory; if the reference source has no inventory, then the hotel's sold-out room type may no longer have inventory.

[0066] In some cases, multiple inventory reference sources can be set for the room type. For example, if any of the inventory reference sources has inventory, it can be considered that the inventory reference source of the room type has inventory, so that subsequent operations can be performed.

[0067] In some cases, because resources are limited, trade-offs need to be made when setting multiple inventory reference sources. For example, various methods can be used to select the most appropriate reference source for the sold-out room type and use its inventory as the basis.

[0068] For example, in some embodiments, the operation of checking the inventory of the inventory reference source of the room type in the above-mentioned step S110 may include: determining multiple candidate inventory reference sources for the room type; estimating the accuracy probability of the multiple candidate inventory reference sources as inventory reference sources respectively; and selecting a candidate inventory reference source whose accuracy probability exceeds a third threshold as the inventory reference source for the room type, and checking the inventory of the inventory reference source.

[0069] If the accuracy probabilities of multiple candidate inventory reference sources exceed the third threshold, then: for example, all candidate inventory reference sources exceeding the third threshold can be selected as the inventory reference sources and their inventories can be checked, and if any of the inventory reference sources has inventory, it can be considered that the inventory reference source of the room type has inventory, so as to perform subsequent operations; or, for example, the candidate inventory reference source with the highest accuracy probability can be selected as the inventory reference source and its inventory can be checked; etc. The present disclosure does not limit the above-mentioned method of selecting candidate inventory reference sources.

[0070] Various methods can be used to determine the above-mentioned multiple candidate inventory reference sources. For example, multiple candidate inventory reference sources can be selected for the room type according to multiple inventory reference source types. The inventory reference source type may include at least one of the following types: similar room types of similar hotels sold on the same hotel sales platform, the same room types of the same hotel sold on other hotel sales platforms, and the same room types of the same hotel sold by other merchants on the same hotel sales platform. One or more candidate inventory reference sources can be selected for each type of multiple inventory reference source types. Various methods can be used to predetermine the candidate inventory reference sources of all room types of all hotels on the hotel sales platform, store the information of all candidate inventory reference sources in a database, and directly query the candidate inventory reference source of the currently sold room type from the database when necessary. In addition, various methods can be used to determine similar room types of similar hotels for the hotel room type. For example, the similarity between the room types of other hotels sold on the same hotel sales platform and the sold room types of the hotel can be determined by any trained machine learning model, and the room types of the previous one or the previous N other hotels with the highest similarity are selected as the candidate inventory reference sources of the sold room types of the hotel. The machine learning model can determine the above similarity from multiple aspects such as the geographical location of the hotel, the overall selling price of the hotel, the grade of the hotel (such as the grade assessed or delineated according to certain rules), the type of room type, the selling price of the room type, etc. In addition, the inventory data of the same room type of the same hotel sold on other hotel sales platforms can be obtained by manually asking the hotel or asking the hotel to fill out a questionnaire, or can be obtained by receiving data provided by other hotel sales platforms with inventory data sharing partnerships. For example, the service end of the hotel sales platform can regularly obtain the inventory data of all hotel room types from other hotel sales platforms with inventory data sharing partnerships, and then find the corresponding inventory data from the selected inventory reference source type when the same room type of the same hotel sold on other hotel sales platforms; or, the service end of the hotel sales platform can receive the corresponding hotel room type inventory data from other hotel sales platforms with inventory data sharing partnerships after determining that the type of the selected inventory reference source is the same room type of the same hotel sold on other hotel sales platforms.In addition, for example, when the type of inventory reference source selected is similar room types of similar hotels sold on the same hotel sales platform or the same room types of the same hotel sold by other merchants on the same hotel sales platform, the server of the hotel sales platform can directly obtain the corresponding inventory data from the database that stores the inventory data of all hotel room types sold by all merchants on its own platform. The database can be a local database, that is, located on the same device as the server, or it can be a remote database, that is, located on a different device from the server, such as located on other data center devices or cloud storage devices. The server can obtain the corresponding inventory data from the remote database through various communication means.

[0071] Therefore, a wider coverage inventory reference source can be used, and a more accurate inventory reference source can be selected as the inventory reference of the sold-out room type. In addition, the inventory reference disclosed in the present invention can be judged in real time, with higher timeliness and easier control.

[0072] In some embodiments, the accurate probabilities of the inventory reference source types to which the multiple candidate inventory reference sources each belong may be estimated as the accurate probabilities of the multiple candidate inventory reference sources each.

[0073] For example, the accurate probability of the inventory reference source type may be estimated using the same trained third machine learning model, or the accurate probabilities of the multiple inventory reference source types may be estimated using corresponding ones of the multiple trained third machine learning models. For example, the same third machine learning model may be trained for multiple inventory reference source types, or the corresponding third machine learning model may be trained for each of the multiple inventory reference source types.

[0074] For example, for the third machine learning model, modeling can be performed with the goal of whether the reference source type is accurate. For example, it can be considered that the reference source type is accurate (the accuracy probability can be set to 1) in the following two cases: the reference source belonging to the reference source type has inventory and the room type generates a completed order after opening the inventory; the reference source belonging to the reference source type has no inventory and the merchant rejects the order after opening the inventory of the room type. During the model training process, the data of these two cases can be used as positive samples (the accuracy probability can be set to 1), and negative samples (the accuracy probability can be set to 0) can be obtained, for example, by random sampling. The model can use various related features (such as various statistical features, time features, and / or cross-features of statistical features and time, etc.) as input features, and the present disclosure does not limit this. In addition, the type, structure, etc. of the third machine learning model are also not limited. For example, a simple machine learning model can be used, or a deep learning model can be used.

[0075] For example, in some embodiments, when using multiple third machine learning models to estimate the accuracy probability of each of multiple inventory reference source types, the multiple third machine learning models are trained for the multiple inventory reference source types, and the input features of the multiple third machine learning models include features representing the historical accuracy of the corresponding inventory reference source type, features representing specific time information or specific period information, and / or features representing the accuracy of the corresponding inventory reference source type at a specific time or a specific period, and the outputs of the multiple third machine learning models are the accuracy probabilities of the corresponding inventory reference source types. The historical accuracy rate is, for example, the proportion of the reference source type being accurate in all past records, and the accuracy rate at a specific time or a specific period is, for example, the proportion of the reference source type being accurate in all records at a specific time or a specific period. It should be understood that the accuracy rate is not the same indicator as the aforementioned accuracy probability. The aforementioned accuracy probability indicates the possibility of the occurrence of the situation where the reference source type is accurate. For example, "0" indicates that the situation is impossible to occur (such as the aforementioned negative sample), and "1" indicates that the situation is bound to occur (such as the aforementioned positive sample).

[0076] For example, assuming there are n reference source types, n third machine learning models f(x) are trained. If it is judged that the accuracy probability is greater than the third threshold α, the candidate inventory reference source belonging to the reference source type can be selected (not necessarily selected), that is, if f(x)>α, the candidate inventory reference source belonging to the reference source type can be selected, and the candidate inventory reference source of the reference source type not greater than the third threshold α will not be selected.

[0077] For example, in some embodiments, when only one third machine learning model is used to estimate the accuracy probability of each of the multiple inventory reference source types, the one third machine learning model is run multiple times to estimate the accuracy probability of each of the multiple inventory reference source types, so its input features also include features representing the inventory reference source type estimated at that time. For example, the input features of the one third machine learning model include features representing the inventory reference source type estimated at that time, features representing the historical accuracy rate of the inventory reference source type estimated at that time, features representing specific time information or specific time period information, and / or features representing the accuracy rate of the inventory reference source type estimated at that time at a specific time or specific time period, and the output of the one third machine learning model is the accuracy probability of the inventory reference source type estimated at that time.

[0078] Then, after checking the inventory of the inventory reference source in the aforementioned step S110, in step S120, when it is determined that the inventory reference source has inventory, the probability of the room type being sold after the additional inventory is added is estimated, and then, in step S130, based on the probability of the room type being sold after the additional inventory, it is determined whether to add inventory to the room type.

[0079] If you simply add an inventory to the sold-out room type after the hotel room type is sold out, the probability of selling the additional inventory may be low, causing trouble to the merchant and wasting resources. Therefore, the present disclosure estimates the probability of the room type being sold after the additional inventory is added, and determines whether to add inventory based on this.

[0080] A complete transaction includes two processes: the user placing an order and the merchant accepting the order. Therefore, in some embodiments, the probability of the room type being sold after the additional inventory is added may include the probability of placing an order for the room type after the additional inventory is added, and / or the probability of the merchant accepting the order after the order is placed for the additional inventory. In the case where the order probability is greater than the first threshold and / or the probability of the merchant accepting the order is greater than the second threshold, it may be determined in the subsequent step S130 that the room type is to be additionally stocked. That is, in some cases, only the probability of the user placing an order may be considered, or only the probability of the merchant accepting the order may be considered. In other cases, both the probability of the user placing an order and the probability of the merchant accepting the order may be considered.

[0081] In addition, in some cases, the probability of the room type being sold after the additional inventory is added can be considered in two stages, that is, the probability of the user placing an order is first estimated, and the probability of the merchant accepting the order is estimated when the probability exceeds the first predetermined threshold, and the room type is determined to be additionally stocked when the probability of accepting the order is greater than the second predetermined threshold. This can provide a backup for the previous inventory reference operation and the user order probability estimation operation, further improve the accuracy of the additional inventory, and prevent overselling; in addition, if the merchant is not very active or has a low willingness to cooperate with the platform, it may lead to the rejection of the order due to the delay in accepting the order, and this situation can be minimized by estimating the merchant's order acceptance probability.

[0082] Therefore, for example, in some embodiments, the probability of selling the room type after the additional inventory is added may include the probability of placing an order for the room type after the additional inventory and the probability of the merchant accepting the order after the order is placed for the additional inventory, and the operation of estimating the probability of selling the room type after the additional inventory in step S120 may include: estimating the probability of placing an order for the room type after the additional inventory; and when the order probability is greater than a first threshold, estimating the probability of the merchant accepting the order after the order is placed for the additional inventory; and when the probability of the merchant accepting the order is greater than a second threshold, determining to add inventory to the room type.

[0083] Various methods can be used to estimate the above-mentioned order probability and / or the probability of the merchant accepting the order. For example, the trained first machine learning model can be used to estimate the order probability. For example, the trained second machine learning model can be used to estimate the probability of the merchant accepting the order.

[0084] For example, for the first machine learning model, modeling can be performed with the goal of whether a room type will generate an order after additional inventory is added. The positive samples of the model during the training process can be: the sold-out room type generates an order after additional inventory is added (the order probability can be set to 1), and the negative samples can be: the sold-out room type does not generate an order after additional inventory (the order probability can be set to 0). The model can use various related features (such as various statistical features, time features, and / or cross-features of statistical features and time, etc.) as input features, and the present disclosure does not limit this. In addition, the type, structure, etc. of the first machine learning model are also not limited. For example, a simple machine learning model can be used, or a deep learning model can be used.

[0085] For example, in some embodiments, the input features of the first machine learning model may include features representing the inventory status of the inventory reference source or multiple candidate inventory reference sources of the room type (for example, "1" can be used to represent inventory, and "0" can be used to represent no inventory), features representing the historical order rate or historical transaction rate of the room type, features representing specific time information or specific time period information, and / or features representing the order rate or transaction rate of the room type at a specific time or specific time period, and the output of the first machine learning model is the order probability of the room type after the additional inventory. The historical order rate or historical transaction rate is, for example, the proportion of order situations or final transaction situations in all records of additional inventory in the past, and the order rate or transaction rate at a specific time or a specific time period is, for example, the proportion of order situations or final transaction situations in all records of additional inventory at a specific time or a specific time period. It should also be understood that the order rate here is not the same indicator as the aforementioned order probability. The aforementioned order probability indicates the possibility of placing an order after the additional inventory this time.

[0086] For example, assuming that the first machine learning model learned after training is g(x), if it is judged that the order probability is greater than the first threshold β, that is, g(x)>β, then the subsequent process continues, otherwise it terminates.

[0087] For example, for the second machine learning model, modeling can be performed with the goal of whether the merchant will accept the order after the order is placed for the additional inventory. The positive samples of the model during the training process can be: the order generated by the additional inventory of the sold-out room type is accepted by the merchant (the probability of accepting the order can be set to 1), and the negative samples can be: the order generated by the additional inventory of the sold-out room type is rejected by the merchant (the probability of accepting the order can be set to 0). The model can use various related features (such as various statistical features, time features, and / or cross-features of statistical features and time, etc.) as input features, and the present disclosure does not limit this. In addition, the type, structure, etc. of the second machine learning model are also not limited. For example, a simple machine learning model can be used, or a deep learning model can be used.

[0088] For example, in some embodiments, the input features of the second machine learning model may include features representing the inventory status of an inventory reference source of the room type or multiple candidate inventory reference sources (for example, inventory can be represented by "1" and no inventory can be represented by "0"), features representing the historical activity of the merchant on the hotel sales platform, features representing the historical order acceptance rate or historical order rejection rate of the room type or the merchant, features representing specific time information or specific time period information, features representing the activity of the merchant on the hotel sales platform at a specific time or specific time period, and / or features representing the order acceptance rate or rejection rate of the room type or the merchant at a specific time or specific time period, and the output of the second machine learning model is the probability of the merchant accepting the order.

[0089] The historical activity of the merchant on the hotel sales platform can, for example, represent the frequency or activity of the merchant's operations on the background of the hotel sales platform in all the past time. It can be evaluated from one or more aspects such as the merchant's interactive behavior in the background (such as participating in activities, etc.), the number of times the inventory / price is operated, the number / frequency of logins, and the online time. The historical order acceptance rate or historical order rejection rate is, for example, the proportion of order acceptance or order rejection in all the records of adding inventory and generating orders in the past. The order acceptance rate or order rejection rate at a specific time or a specific period of time is, for example, the proportion of order acceptance or order rejection in all the records of adding inventory and generating orders at a specific time or a specific period of time. It should also be understood that the order acceptance rate here is not the same indicator as the aforementioned order acceptance probability. The aforementioned order acceptance probability indicates the possibility of the merchant accepting the order after the inventory is added and the order is generated.

[0090] For example, assuming that the second machine learning model learned after training is z(x), if it is judged that its order acceptance probability is greater than the second threshold γ, that is, z(x)>γ, then continue the subsequent process, otherwise terminate.

[0091] The types or structures of the aforementioned machine learning models may be the same or different. In addition, for example, in the case where the relevant historical data required by the aforementioned machine learning models is lacking or insufficient, various methods may be used to cold start them.

[0092] In addition, in some cases, the aforementioned trained machine learning models can be deployed on the server side of the hotel sales platform, and part or all of the data required for their respective input features (such as various historical data about each room type / merchant, etc.) can be stored in the database corresponding to the server side and retrieved from the database for use when needed.

[0093] Through the aforementioned machine learning models, it is possible to more accurately estimate the actual inventory status of sold-out rooms and the probability of a transaction after additional inventory is added, thereby being able to more accurately judge the situation of additional inventory and effectively reduce problems such as order rejections and merchant customer complaints.

[0094] In addition, after determining to add inventory as described above, an additional inventory can be added to the sold-out room type, for example, but market demand changes in real time, and the various conditions mentioned above may no longer be met after a period of time, that is, no additional inventory should be added. Therefore, the method disclosed in the present invention can also add an inspection task after the additional inventory is added and before the additional inventory is sold, re-evaluate the room type, cancel the additional inventory when the aforementioned requirements are not met, and change the inventory to a sold-out state again. For example, the service end of the hotel sales platform can perform the additional inventory inspection task and related operations, such as the subsequent construction of an operations research optimization model and the corresponding determination of inspection arrangements.

[0095] Therefore, in some embodiments, after adding inventory to the room type and before the added inventory has been sold, the method according to the present disclosure may further include performing an additional inventory check task on the room type to determine whether to cancel the additional inventory for the room type; wherein the additional inventory check task includes: re-executing the step of checking the inventory of the inventory reference source of the room type, the step of estimating the probability of the room type being sold after the additional inventory is added, and / or the step of determining whether to add inventory to the room type, and in the case where the inventory reference source has no inventory or it is determined not to add inventory to the room type, determining to cancel the additional inventory for the room type. The additional inventory check task is to combine the aforementioned Figure 1 The described operations are executed again to re-determine whether to add inventory to the room type, and if any of the aforementioned conditions for adding inventory are not met, determine to cancel the previously added inventory.

[0096] If there are multiple room types that have been added to the inventory but have not been sold, you can make inspections on these multiple room types, that is, you can periodically poll and inspect these multiple room types. However, in some cases, due to limited resources, if the same inspection frequency is used for all room types, it will cause task accumulation and resource waste. Therefore, you can construct and solve an operations optimization model to find the most optimized inspection arrangement for each room type under limited resources.

[0097] Therefore, in some embodiments, when the room types for additional inventory include multiple room types in one or more hotels and the additional inventory inspection tasks are performed on the multiple room types respectively, the inspection schedule of the additional inventory inspection tasks for the multiple room types in a first predetermined period in the future can be determined by constructing an operations optimization model and solving the operations optimization model, wherein the goal of the operations optimization model is to maximize the total value of the additional inventory inspection tasks of the multiple room types to be inspected in the first predetermined period in the future (such as a day), the value of the additional inventory inspection tasks of each room type in the multiple room types in each inspection is determined according to the order probability of the corresponding room type in a second predetermined period in the future (such as within one hour) after the execution time of each inspection, and the constraints of the operations optimization model include the minimum number of times the additional inventory inspection tasks of each room type are to be inspected, the minimum number of intervals between inspections of the additional inventory inspection tasks of each room type, and / or the upper limit of the total number of times all the additional inventory inspection tasks are to be inspected.

[0098] For example, the operations research optimization model can be constructed as follows:

[0099] ·parameter:

[0100]

[0101] Decision variables:

[0102]

[0103] Objective function:

[0104]

[0105] ·constraint:

[0106]

[0107] After defining the parameters, decision variables, constraints and objective function of the operations research optimization model as above, the operations research optimization solver can be called to solve it, so as to obtain the inspection schedule for the additional inventory inspection tasks for these N room types in the coming day, that is, which inspection tasks should be performed in each inspection in the coming day.

[0108] Therefore, it is possible to perform personalized inspection frequency control on additional inventory inspection tasks, with high resource utilization.

[0109] In addition, in some cases, the platform's automatic operation of adding inventory can fully inform the merchant and minimize the negative impact on the merchant. For example, the merchant's authorization or reminder can be obtained in various ways before or after the execution of the method disclosed in the present invention, and any operation performed by the merchant on the platform's automatically added inventory will not affect the merchant's evaluation or rights on the platform, etc. For example, when a merchant receives an order for the added inventory, it can refuse the order, which is different from the rejection of the general inventory (i.e., the inventory set by the merchant himself), and will not be counted in the platform's evaluation system for the merchant. For example, on the merchant backend website of the hotel sales platform for merchants to access, the merchant is usually presented with an operation interface for managing the hotel services sold on the hotel sales platform, and the merchant's authorization or reminder can be obtained by presenting corresponding information on the operation interface.

[0110] Therefore, in some embodiments, the method according to the present disclosure may further include: before adding inventory to the room type, obtaining authorization from the merchant to add inventory. For example, a corresponding authorization request may be presented to the merchant on a corresponding operation interface of a merchant backend website of the hotel sales platform for the merchant to access, and receiving information that the merchant agrees to the authorization request means obtaining authorization from the merchant to add inventory.

[0111] In some cases, you can ask the merchant for authorization every time before adding inventory for the room type, and only add inventory for the room type after obtaining the merchant's authorization to add inventory. In other cases, you can obtain the merchant's general authorization, that is, the merchant agrees to the platform's authorization to automatically add inventory when there is no inventory. After obtaining this authorization, the platform can automatically add inventory without requesting the merchant's authorization again.

[0112] In some embodiments, the method according to the present disclosure may further include: after adding inventory to the room type, prompting the merchant to indicate that the inventory has been added and presenting information about the added inventory.

[0113] For example, corresponding prompt information can be presented to the merchant on the corresponding operation interface / backend interface of the merchant backend website of the hotel sales platform for the merchant to access (such as the hotel management interface and / or inventory management interface provided for the merchant) to remind the merchant that a certain inventory is to be added; or a message can be sent to the merchant through various channels to provide a corresponding prompt.

[0114] Figure 3 FIG. 1 shows an exemplary operation interface for a hotel merchant to manage inventory on a travel service platform according to at least one embodiment of the present disclosure. Figure 3 As shown in the red dotted box in the figure, the prompt text "Automatic Room Adjustment" is displayed on the interface to remind the merchant that the platform has automatically added inventory.

[0115] In some embodiments, the method according to the present disclosure may further include: after adding inventory to the room type, providing the merchant with an entry or operation button to cancel the added inventory. For example, the corresponding entry or operation button may be presented to the merchant on a corresponding operation interface of a merchant backend website of the hotel sales platform for the merchant to access.

[0116] In some embodiments, the method according to the present disclosure may further include: after adding inventory to the room type and the added inventory is ordered, prompting the merchant to indicate that an order has been generated for the added inventory, allowing the merchant to reject the order, and evaluating the merchant's operation on the order for the added inventory differently from the operation on the order for the inventory set by the merchant itself. For example, when the merchant rejects the order for the added inventory, the platform will treat the merchant's evaluation method differently from general orders, which will not affect the merchant's rating.

[0117] For example, the method according to the present disclosure may also include: after adding inventory for the room type, receiving order information of the user for the added inventory via a client of the hotel sales platform for users and transmitting the order information to a server of the hotel sales platform; and when the merchant accesses the merchant backend website of the hotel sales platform, based on the order information obtained by the server, prompting the merchant via the merchant backend website to indicate that an order has been generated for the added inventory, and receiving information about whether the merchant accepts the order.

[0118] It should be understood that the aforementioned embodiments of the present disclosure or the operations / steps in the embodiments may be implemented in combination in various ways as needed, and the present disclosure is not limited thereto.

[0119] Combine the following Figure 2 The complete implementation process of the method according to some embodiments of the present disclosure is described exemplarily. It should be understood that Figure 2 The processes in the foregoing are merely examples and are not intended to limit the present disclosure. Other solutions of the present disclosure may be modified in various ways based on the teachings given herein. According to the ideas of the present disclosure described below, those skilled in the art may easily implement other solutions of the present disclosure, which are all included in the protection scope of the present disclosure.

[0120] like Figure 2 As shown, the automatic inventory replenishment solution according to the present disclosure can be mainly divided into five key parts:

[0121] 1. Hotel room type sold out event detection: a real-time detection task can be set. When a certain room type of a certain hotel of a certain merchant is sold out, as shown in step S210, a hotel room type sold out event is generated. This type of event will trigger a subsequent automatic inventory replenishment task, that is, enter step S220;

[0122] 2. Room type inventory reference source selection:

[0123] In step S220, as described above, Figure 1 As described above, the trained third machine learning model can be used to select the most appropriate inventory reference source for the sold-out room type;

[0124] Then, in step S230, the inventory status of the selected inventory reference source is checked. If there is inventory, the subsequent process continues, that is, entering step S240. If there is no inventory, the process is terminated, that is, no additional inventory is added;

[0125] 3. Order probability estimation:

[0126] In step S240, as previously described, Figure 1 As described above, the trained first machine learning model can be used to estimate the probability of placing an order for a room type after adding inventory. If the probability of placing an order exceeds the first threshold, the subsequent process continues, i.e., enters step S250. If it does not exceed the first threshold, the process terminates, i.e., no inventory will be added.

[0127] 4. Estimation of the probability of accepting an order:

[0128] In step S250, as described above, Figure 1 As described above, the trained second machine learning model can be used to estimate the probability of the merchant accepting the order after the order for the added inventory is placed. If the probability of accepting the order exceeds the second threshold, the subsequent process continues, that is, entering step S260. If it does not exceed the second threshold, the process is terminated, that is, the inventory will not be added;

[0129] Then, in step S260, additional inventory is added for the room type, for example, one inventory is added;

[0130] 5. Inspection and judgment after adding inventory:

[0131] After the additional inventory is added in step S260 and the additional inventory has not been sold, the sold-out room types of the additional inventory can be inspected regularly, and the additional inventory can be cancelled when the previous requirements are not met; in the case where there are currently multiple sold-out room types for the additional inventory, as mentioned above, an operations optimization model can be constructed and solved to arrange the inspections of the multiple room types, as shown in step S270; after the inspection plan for each additional inventory inspection task is arranged in step S270, Figure 2 As shown by the dotted line in the figure, the additional inventory check task is performed according to the arrangement, that is, the previous judgment operation process from step S220 to step S250 is re-executed, and the additional inventory is cancelled if it is judged that additional inventory is not needed. If it is still judged that additional inventory is needed, no operation is performed, that is, the operation of step S260 is not repeated.

[0132] The specific implementation methods of the above operations can be combined with the above Figure 1 The various methods described above, or other suitable methods.

[0133] The above solution disclosed in the present invention selects the inventory reference source -> estimates the probability of placing an order -> estimates the probability of merchant accepting an order -> solves the inspection frequency through the operations optimization model, decomposes the goals of each node, does not make special requirements for the model, and constructs a relatively complete system.

[0134] In addition, an online experiment was conducted on the method of adding inventory according to the present invention, and the effect was significant, which led to an increase in order volume and a decrease in the merchant's rejection rate.

[0135] Figure 4 A schematic diagram of the structure of a computing device that can be used to implement the above-mentioned method for managing hotel inventory on a hotel sales platform according to at least one embodiment of the present disclosure is shown.

[0136] See also Figure 4 , the computing device 400 includes a memory 410 and a processor 420 .

[0137] Processor 420 may be a multi-core processor or may include multiple processors. In some embodiments, processor 420 may include a general-purpose main processor and one or more special coprocessors, such as a graphics processing unit (GPU), a digital signal processor (DSP), etc. In some embodiments, processor 420 may be implemented using a customized circuit, such as an application-specific integrated circuit (ASIC) or a field programmable gate array (FPGA).

[0138] The memory 410 may include various types of storage units, such as system memory, read-only memory (ROM), and permanent storage devices. Among them, ROM can store static data or instructions required by the processor 420 or other modules of the computer. The permanent storage device may be a readable and writable storage device. The permanent storage device may be a non-volatile storage device that does not lose the stored instructions and data even after the computer is powered off. In some embodiments, the permanent storage device uses a large-capacity storage device (such as a magnetic or optical disk, flash memory) as a permanent storage device. In some other embodiments, the permanent storage device may be a removable storage device (such as a floppy disk, optical drive). The system memory may be a readable and writable storage device or a volatile readable and writable storage device, such as a dynamic random access memory. The system memory may store some or all instructions and data required by the processor at run time. In addition, the memory 410 may include any combination of computer-readable storage media, including various types of semiconductor memory chips (DRAM, SRAM, SDRAM, flash memory, programmable read-only memory), and disks and / or optical disks may also be used. In some embodiments, the memory 410 may include a readable and / or writable removable storage device, such as a laser disc (CD), a read-only digital versatile disc (such as a DVD-ROM, a double-layer DVD-ROM), a read-only Blu-ray disc, an ultra-density optical disc, a flash memory card (such as an SD card, a mini SD card, a Micro-SD card, etc.), a magnetic floppy disk, etc. The computer-readable storage medium does not include carrier waves and transient electronic signals transmitted wirelessly or wired.

[0139] The memory 410 stores executable codes, and when the executable codes are processed by the processor 420 , the processor 420 can execute the above-mentioned method for managing hotel inventory on the hotel sales platform.

[0140] It should be noted that the user / merchant information (including but not limited to user / merchant device information, user / merchant personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user / merchant or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and provide corresponding operation entrances for users / merchants to choose to authorize or refuse.

[0141] In addition, the method according to the present disclosure may also be implemented as a computer program or a computer program product, which includes computer program code instructions for executing the above steps defined in the above method of the present disclosure.

[0142] Alternatively, the present disclosure may also be implemented as a non-temporary machine-readable storage medium (or computer-readable storage medium, or machine-readable storage medium) on which executable code (or computer program, or computer instruction code) is stored. When the executable code (or computer program, or computer instruction code) is executed by a processor of an electronic device (or computing device, server, etc.), the processor executes the various steps of the above-mentioned method according to the present disclosure.

[0143] Those skilled in the art will further appreciate that the various illustrative logical blocks, modules, circuits, and algorithm steps described in connection with the disclosure herein may be implemented as electronic hardware, computer software, or a combination of both.

[0144] The flow charts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the system and method according to multiple embodiments of the present disclosure. In this regard, each box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, a program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flow chart, and the combination of the boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0145] The embodiments of the present disclosure have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The selection of terms used herein is intended to best explain the principles of the embodiments, practical applications, or improvements to the technology in the market, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein.

Claims

1. A method for managing hotel inventory on a hotel sales platform, comprising: When the room type sold by the merchant on the hotel sales platform is out of stock, check the inventory of the room type with reference to the inventory source; If the inventory reference source has inventory, estimate the probability of the house type being sold after the inventory is added; as well as Based on the probability of the room type being sold after the additional inventory is added, it is determined whether to add additional inventory for the room type.

2. The method according to claim 1, wherein: The probability of the room type being sold after the additional inventory is added includes the probability of placing an order for the room type after the additional inventory is added, and / or the probability of the merchant accepting the order after the order is placed for the additional inventory, and when the order probability is greater than a first threshold and / or the probability of the merchant accepting the order is greater than a second threshold, it is determined that the room type is to be added to the inventory; or, The probability of the room type being sold after the additional inventory is added includes the probability of placing an order for the room type after the additional inventory is added and the probability of the merchant accepting the order after the order is placed for the additional inventory, and the step of estimating the probability of the room type being sold after the additional inventory is added includes: estimating the probability of ordering the room type after additional inventory is added; and When the order placement probability is greater than a first threshold, estimating the probability that the merchant will accept the order after placing an order for the added inventory; and When the probability that the merchant accepts the order is greater than a second threshold, it is determined to add inventory to the room type.

3. The method according to claim 1, wherein: The steps of viewing the inventory of the room type with reference to the inventory source include: Determining multiple candidate inventory reference sources for the room type; estimating the accuracy probability of each of the plurality of candidate inventory reference sources being used as an inventory reference source; and A candidate inventory reference source whose accuracy probability exceeds a third threshold is selected as the inventory reference source of the room type, and the inventory of the inventory reference source is checked.

4. The method according to claim 3, wherein: The step of estimating the accuracy probability of the plurality of candidate inventory reference sources being used as inventory reference sources comprises: The accurate probabilities of the inventory reference source types to which the multiple candidate inventory reference sources respectively belong are estimated as the accurate probabilities of the multiple candidate inventory reference sources respectively.

5. The method according to claim 2 or 4, wherein: The order placement probability is estimated using the trained first machine learning model; and / or, The probability of the merchant accepting the order is estimated using the trained second machine learning model; and / or, The inventory reference source type includes at least one of similar room types in similar hotels sold on the hotel sales platform, the same room types in the same hotel sold on other hotel sales platforms, and the same room types in the same hotel sold on the hotel sales platform by other merchants; and / or, The accurate probability of the inventory reference source type is estimated using the same trained third machine learning model, or the accurate probability of each of the multiple inventory reference source types is estimated using a corresponding one of the multiple trained third machine learning models.

6. The method according to claim 5, wherein: The input features of the first machine learning model include features representing the inventory status of the inventory reference source or multiple candidate inventory reference sources of the room type, features representing the historical order rate or historical transaction rate of the room type, features representing specific time information or specific period information, and / or features representing the order rate or transaction rate of the room type at a specific time or specific period, and the output of the first machine learning model is the order probability of the room type after the additional inventory is added; and / or, The input features of the second machine learning model include features representing the inventory status of the inventory reference source of the room type or multiple candidate inventory reference sources, features representing the historical activity of the merchant on the hotel sales platform, features representing the historical order acceptance rate or historical order rejection rate of the room type or the merchant, features representing specific time information or specific time period information, features representing the activity of the merchant on the hotel sales platform at a specific time or a specific time period, and / or features representing the order acceptance rate or order rejection rate of the room type or the merchant at a specific time or a specific time period, and the output of the second machine learning model is the probability of the merchant accepting the order; and / or, In the case where only one third machine learning model is used to estimate the accuracy probability of each of the multiple inventory reference source types, the one third machine learning model is run multiple times to respectively estimate the accuracy probability of each of the multiple inventory reference source types, and the input features of the one third machine learning model include features representing the inventory reference source type estimated at the time, features representing the historical accuracy rate of the inventory reference source type estimated at the time, features representing specific time information or specific time period information, and / or features representing the accuracy rate of the inventory reference source type estimated at the time at a specific time or a specific time period, and the output of the one third machine learning model is the accuracy probability of the inventory reference source type estimated at the time, When using multiple third machine learning models to respectively estimate the accuracy probability of each of the multiple inventory reference source types, the multiple third machine learning models are trained for the multiple inventory reference source types respectively, and the input features of the multiple third machine learning models respectively include features representing the historical accuracy of the corresponding inventory reference source type, features representing specific time information or specific time period information, and / or features representing the accuracy of the corresponding inventory reference source type at a specific time or a specific time period, and the outputs of the multiple third machine learning models are respectively the accuracy probabilities of the corresponding inventory reference source types.

7. The method according to claim 1, after adding inventory to the room type and before the added inventory is sold, further comprising: Perform an additional inventory check task on the room type to determine whether to cancel the additional inventory for the room type. The additional inventory check task includes: re-executing the step of checking the inventory of the inventory reference source of the room type, estimating the probability of the room type being sold after the additional inventory, and / or determining whether to add additional inventory for the room type, and in the case that the inventory reference source has no inventory or it is determined not to add additional inventory for the room type, determining to cancel the additional inventory for the room type.

8. The method according to claim 7, wherein: In a case where the room types to be additionally stocked include multiple room types in one or more hotels and the additional stock inspection tasks are performed on the multiple room types respectively, an inspection arrangement for the additional stock inspection tasks for the multiple room types within a first predetermined period in the future is determined by constructing an operations research optimization model and solving the operations research optimization model. Among them, the goal of the operations optimization model is to maximize the total value of the additional inventory inspection tasks of the multiple room types to be inspected in a first predetermined time period in the future, the value of the additional inventory inspection task of each room type in the multiple room types in each inspection is determined according to the order probability of the corresponding room type in a second predetermined time period in the future after the execution time of each inspection, and the constraints of the operations optimization model include the minimum number of times the additional inventory inspection task of each room type must be inspected, the minimum number of intervals between inspections of the additional inventory inspection task of each room type, and / or an upper limit on the total number of times all additional inventory inspection tasks are to be inspected.

9. The method according to claim 1, further comprising: Before adding inventory to the room type, obtain authorization from the merchant to add inventory; and / or, After adding inventory to the room type, prompting the merchant to indicate that the inventory has been added and presenting information about the added inventory; and / or, After adding inventory to the room type, providing the merchant with an entry or operation button to cancel the added inventory; and / or, After the additional inventory is added to the room type and the additional inventory is ordered, the merchant is prompted to indicate that an order has been generated for the additional inventory, the merchant is allowed to reject the order, and the merchant's operation on the order for the additional inventory is evaluated differently from the operation on the order for the inventory set by the merchant itself; and / or, After adding inventory to the room type, receiving order information of the user for the added inventory through a client of the hotel sales platform for use by the user and transmitting the order information to a server of the hotel sales platform; and When the merchant visits the merchant backend website of the hotel sales platform, based on the order information obtained by the server, the merchant is prompted via the merchant backend website to indicate that an order has been generated for the additional inventory, and information about whether the merchant accepts the order is received.

10. A computer program product, comprising executable codes, which, when executed by a processor of an electronic device, cause the processor to perform the method according to any one of claims 1 to 9.

Images