WO2018036307A1

WO2018036307A1 - Information processing method utilized in pushing information order, allocation method, device, and data storage medium

Info

Publication number: WO2018036307A1
Application number: PCT/CN2017/093510
Authority: WO
Inventors: 徐澜; 陈戈; 金伟; 江志; 赵鹏昊; 张弘; 刘磊; 黄伟; 黄东波; 姜磊; 黄浩; 朱思宇; 谷俊青; 游乐; 魏望; 洪福兴; 陈怡然; 李世强
Original assignee: 腾讯科技（深圳）有限公司
Priority date: 2016-08-23
Filing date: 2017-07-19
Publication date: 2018-03-01
Also published as: CN107767164A; CN107767164B

Abstract

An information processing method utilized in pushing information order comprises: acquiring user data history (301); determining, according to the user data history, and with respect to each of preconfigured dimensions or dimension combinations, corresponding exposures, and determining a set of exposure proportions, wherein each of the exposure proportions corresponds to a frequency (302); and establishing, according to each of the dimensions or the dimension combinations, and the corresponding exposures and the set of the exposure proportions, an inventory model describing each supply node and a corresponding inventory thereof and a set of inventory proportions (303). Also provided are an allocation method, an exposure determination method, and a corresponding device.

Description

Information processing method, distribution method, device and storage medium for pushing information order

This application claims the priority of the Chinese Patent Application filed on August 23, 2016, the Chinese Patent Office, the application number is 201610708108.5, and the invention name is "information processing method, distribution method and device for pushing information orders", the entire contents of which are incorporated by reference. Combined in this application.

Technical field

The present application relates to the field of the Internet, and in particular, to an information processing method, a distribution method, and corresponding devices and storage media for pushing information orders.

background

With the development of Internet technology, more and more data (including text, pictures, audio, video, etc.) will be pushed to various users through the Internet. For example, when a user browses a webpage using a mobile device, a PC, or the like, the user receives various data pushed by the network side, such as an advertisement in a picture or video format, public welfare information, news, and the like. In this way, the user can know the news, the content of interest, and the like in time. Such data can be referred to as push information or push media content, and the like.

Technical content

The present application provides an information processing method for pushing an information order, comprising: acquiring historical user data; wherein the historical user data includes exposure data that has occurred by each user; and according to the historical user data, for each dimension predetermined Or a combination of dimensions, determining a corresponding number of exposures, and determining a proportion of a set of exposures, wherein each exposure quantity corresponds to a frequency; according to each dimension or combination of dimensions and the corresponding number of exposures and Calculating a set of exposure quantity, establishing an inventory model, the inventory model describing each supply node and its corresponding inventory quantity and a set of inventory proportions; wherein one supply node corresponds to one dimension or one dimension combination, the supply node The corresponding inventory quantity is the exposure quantity corresponding to the dimension or the combination of the dimensions, and the corresponding set of inventory proportions of the supply node includes the ratio of the set of exposure quantity corresponding to the dimension or the combination of the dimensions.

The application also provides a service allocation method for pushing information orders, comprising: extracting information of each order; extracting the above inventory model; generating an allocation plan according to the information of each order and the inventory model, the distribution plan includes The service ratio for each order, where the service ratio for each order is determined as follows:

Using the inventory model, determining an inventory ratio corresponding to the frequency limit of the order corresponding to all the supply nodes corresponding to the order of the order; and the inventory amount corresponding to each supply node corresponding to the orientation of the order, Determining the service ratio of the order, the current estimated inventory remaining amount, the determined one inventory share, and the booked amount of the order;

Thereafter, the allocation plan is provided to the push server to cause the push server to perform a menu process according to the service ratio in the allocation plan.

The application also provides an inquiry method for pushing information orders. When receiving a request for a new order from the demand side, the method includes: extracting information of each existing order; extracting the above inventory model; The new order and the orientation of each existing order determine the order of allocation of the new order and the existing orders; in the order of allocation, the existing orders before the new order are sequentially processed as follows:

Using the inventory model, determining an inventory ratio corresponding to the frequency limit conditions of the existing order corresponding to all supply nodes corresponding to the orientation of the current order; and corresponding to each supply node corresponding to the orientation of the current order Stock quantity, current estimated stock remaining amount, determined one stock ratio, and current order booking amount, determined The service ratio of the current order and update the current estimated stock remaining amount;

Using the inventory model, determining an inventory ratio corresponding to the frequency limit of the new order corresponding to all the supply nodes corresponding to the orientation of the new order; and corresponding inventory corresponding to each supply node corresponding to the orientation of the new order Quantity, the current estimated stock remaining amount, and the one inventory ratio, determining an initial value of the reservation amount upper limit, and setting an initial value of the lower limit of the reservation amount to a preset value; in the order of the allocation, starting from the new order The order is processed as follows until the lower limit of the reservation amount is equal to the upper limit of the reservation amount:

Using the inventory model, determining an inventory ratio corresponding to the frequency limit of the current order corresponding to all supply nodes corresponding to the orientation of the current order; and the inventory amount corresponding to each supply node according to the orientation of the current order, The current estimated inventory remaining amount, the one inventory share ratio, and the current bookable quantity detection value, attempting to determine the service ratio of the current order, updating the current estimated stock remaining amount, and up-regulating according to whether the current order service ratio can be determined. The lower limit of the reservation quantity or the value of the upper limit of the reservation quantity is adjusted, and the bookable quantity detection value is adjusted according to the lower limit of the reservation quantity and the upper limit of the reservation quantity; wherein the initial value of the reservation quantity upper limit is taken as Determining an initial value of the recordable quantity detection value;

Thereafter, the final value of the upper limit of the reservation amount is taken as the maximum amount of the bookable amount of the new order.

Correspondingly, the present application provides an information processing apparatus for pushing an information order, including:

One or more memories;

One or more processors; among them,

The one or more memories storing one or more instruction modules configured to be executed by the one or more processors; wherein

The one or more instruction modules include:

a data extraction module, which acquires historical user data; wherein the historical user data includes exposure data that has occurred by each user; and a calculation module, according to the historical user data, the needle Determining a corresponding exposure quantity for each predetermined dimension or combination of dimensions, and determining a proportion of exposure quantity, wherein each exposure quantity corresponds to a frequency; a model building module, according to each dimension or dimension combination Corresponding to the ratio of the exposure quantity to the set of exposure quantity, an inventory model is established, and the inventory model describes each supply node and its corresponding inventory quantity and a group of inventory proportions; wherein one supply node corresponds to one dimension Or a combination of dimensions, the inventory quantity corresponding to the supply node is the exposure quantity corresponding to the dimension or the combination of the dimensions, and the set of inventory proportions corresponding to the supply node includes the set of exposures corresponding to the dimension or the combination of the dimensions The proportion is proportional.

Correspondingly, the present application provides a service distribution device for pushing information orders, including:

One or more memories;

One or more processors; among them,

The one or more instruction modules include:

a data extraction module, extracting information of each order, extracting the above inventory model; and assigning a planning module, generating an allocation plan according to the information of each order and the inventory model, where the distribution plan includes a service ratio of each order, wherein each The service ratio of an order is determined by using the inventory model to determine an inventory ratio corresponding to the frequency limit of the order corresponding to all the supply nodes corresponding to the order of the order; and according to the orientation of the order Determining the service ratio of the order corresponding to the inventory quantity, the current estimated inventory remaining quantity, the determined one inventory share ratio, and the order quantity of the order; determining an output ratio of the order; The allocation plan is given to the push server to cause the push server to perform a menu process according to the service ratio in the allocation plan.

Correspondingly, the present application provides an inquiry device for pushing information orders, when demanding When the party receives a request for a new order, the device includes:

One or more memories;

One or more processors; among them,

The one or more instruction modules include:

a data extraction module, extracting information of each existing order, extracting the above inventory model; and assigning a planning module, determining an order of assigning the new order and the existing orders according to the new order and the orientation of each existing order; In the order of allocation, the following orders are processed in sequence for each existing order before the new order:

Using the inventory model to determine an inventory ratio corresponding to the frequency limit of the existing order corresponding to all supply nodes that match the orientation of the current order;

Determine the service ratio of the current order and update the current order based on the inventory amount corresponding to each supply node corresponding to the orientation of the current order, the current estimated inventory remaining amount, the determined one inventory share, and the current order subscription amount. Estimated stock remaining amount;

The apparatus further includes: a query boundary module, using the inventory model to determine an inventory ratio corresponding to a frequency limit of the new order corresponding to all supply nodes corresponding to the orientation of the new order; according to the orientation with the new order The corresponding inventory quantity corresponding to each supply node, the current estimated inventory remaining quantity, and the one inventory proportion, determine the initial value of the reservation quantity upper limit, and set the initial value of the reservation quantity lower limit to a preset value; the reservation quantity detection module, In the order of allocation, the following processing is performed on each order in order from the new order until the lower limit of the reservation amount is equal to the upper limit of the reservation amount: using the inventory model, all supply nodes corresponding to the orientation of the current order are determined Corresponding inventory ratio that meets the frequency limit of the current order; and the inventory amount corresponding to each supply node corresponding to the orientation of the current order, the current estimated inventory remaining amount, the one inventory ratio, and the current The amount of the detection value can be reserved, an attempt is made to determine the service ratio of the current order, the current estimated inventory remaining amount is updated, and the lower limit of the reservation amount or the lower limit of the reservation amount is adjusted according to whether the service ratio of the current order can be determined. And adjusting, according to the lower limit of the reservation quantity and the upper limit of the reservation quantity, the bookable quantity detection value; wherein the reservation quantity upper limit initial value is used as an initial value of the bookable quantity detection value; the reservation quantity detection module The final value of the reservation amount upper limit is taken as the maximum bookable amount of the new order.

The present application also proposes a non-transitory computer readable storage medium storing computer readable instructions that enable at least one processor to perform the above method.

BRIEF DESCRIPTION OF THE DRAWINGS

In order to more clearly illustrate the technical solutions in the present application, the drawings used in the description of the examples will be briefly described below. It is obvious that the drawings in the following description are only some examples of the present application, which are common in the art. For the skilled person, other drawings can be obtained from these drawings without any creative work. among them,

1 is a system architecture diagram related to an example of the present application;

Figure 2 shows the relationship between the supply node and the demand node;

3 is a flow chart of an information processing method for an order in an example of the present application;

4 is a flow chart of a service allocation method for an order in an example of the present application;

5 is a flow chart of an inquiry method of an order in an example of the present application;

6 is a schematic diagram of an inventory model establishing apparatus in an example of the present application;

7 is a schematic diagram of a service allocation apparatus in an example of the present application;

8 is a schematic diagram of an interrogation device in an example of the present application; and

FIG. 9 is a schematic structural diagram of a computing device in an example of the present application.

Detailed ways

For the sake of brevity and clarity of the description, the aspects of the present application are set forth below by describing several representative examples. However, not all embodiments are shown herein. A large number of the details in the examples are only used to help understand the solution of the present application, and the technical solutions of the present application may not be limited to these details. In order to avoid unnecessarily obscuring the aspects of the present application, some embodiments are not described in detail, but only the framework is given. Hereinafter, "including" means "including but not limited to", and "according to" means "at least according to ..., but not limited to only based on". The word "comprising" in the specification and claims is intended to mean, to the extent that,

In the Internet-based information push technology, the online push information display service (such as the display of online advertisements) can be realized by the order guaranteed by the flow (ie, the exposure quantity). Among them, the media party (Publisher, also known as the supplier) responsible for displaying the push information guarantees a predetermined amount of exposure (ie, inventory) that satisfies the ordering predicates to the demanding party (such as the advertiser Advertiser). Here, the orientation describes the target demographic of the push information corresponding to the order, and the impression of the push information to the user is referred to as an exposure.

In the traffic-guaranteed information push scenario, there are usually two phases involved: First, the sales phase, the demand side purchases a predetermined amount of exposure for an order from the media (ie, purchases a predetermined flow or exposure amount for an order); Second, in the service stage, based on the current inventory quantity, the media party pushes the push information corresponding to each order to the user who meets the target audience characteristics, wherein the order describes the target audience characteristics agreed by the media party and the demand side. Specifically, in the sales stage, it is necessary to determine whether the predetermined quantity of the new order to be added currently can be satisfied based on the current inventory estimation result; in the service stage, it is necessary to select an order to be exposed based on the current inventory estimation result and determine Prepare for each exposure How much exposure is allocated by the light order. Therefore, the inventory forecast will have certain impact on these two stages. Inaccurate inventory estimation results and imperfect inventory models may cause the exposure flow of some orders to be unguaranteed or some orders to be overexposed (ie, actual exposure). The amount exceeds the predetermined amount of the order).

For example, the core issue in the CPM (Cost Per Impressions) contract advertising system is how to allocate an available ad exposure, that is, a set of orders with contractual and audience targeting constraints. Under the estimated inventory, each order is allocated advertising stock according to its audience targeting constraints and demand, so that all orders are not lacking or the overall deficiency is minimal. This allocation scheme can be used to:

1. Guided delivery (ie the second phase above): Provide a basic allocation plan, guide the delivery engine to deliver and adjust the distribution plan according to the real-time feedback data.

2. Auxiliary Inquiries (ie, the first phase above): Provides residual exposure information for various audience targeting constraints to help advertisers optimize their advertising plans.

FIG. 1 is a structural diagram of a system involved in an example of the present application. As shown in FIG. 1, the system includes at least an inventory allocation module 101, a data platform 102, a push server 103, and an order management module 104. The inventory allocation module 101 may include an inventory estimation module 111, a sales distribution module 121, and a service distribution module 131. The data platform 102 may include an offline data module 112 and a real-time data module 122.

In the Internet, each user uses a client to access some websites, such as browsing a webpage or watching online videos, etc., and the push server 103 (for example, an advertisement service area for placing Internet advertisements) can collect which pages are currently accessed by which users. (URL), which in turn determines which users need to push information and what information to push. The push server 103 generates a respective exposure request for the user currently accessing the network and returns it to the client of the corresponding user, and transmits it to the data platform 102 for recording as log data, such as the real-time data module 122 transmitted thereto. Real-time data module 122 will pass these exposure requests to the offline data module 112 as historical exposure data, and will also update the current subscription amount (or demand amount) of each order based on these exposure requests and pass the latest push amount of each order to the inventory allocation. Module 101, such as service distribution module 131, is passed to it.

The inventory allocation module 101 (such as the inventory estimation module 111 therein) may estimate the inventory amount based on historical exposure data acquired from the data platform 102 (such as the offline data module 112 therein) to obtain an estimated inventory amount, and based on the pre- The inventory model and historical exposure data are used to establish an inventory model. The inventory allocation module 101 (such as the service distribution module 131 therein) may extract information for each order from the order management module 104, extract an inventory model from the inventory estimation module 111, and determine an allocation plan based on the inventory model and information for each order.

The inventory allocation module 101 (such as the service distribution module 131 therein) can update the subscription of each order when the latest push amount of each order (ie, the amount of exposure, such as the amount of played of an advertisement) is received from the real-time data module 122. Quantity, which in turn can update the allocation plan. The push server 103 can extract information from the order management module 104 that it is directed to a plurality of orders that match those users for the currently received exposure requests of the plurality of users. The service ratios corresponding to the respective orders are determined according to the distribution plan acquired from the inventory allocation module 101, and the push server 103 can perform menu processing on the orders according to the service ratio in the distribution plan, and can provide information push services based on the menu results (for example, advertisements) The server can push the corresponding advertisement to each user according to the menu result, or determine the unselected order without having to push the advertisement to the user).

The inventory allocation module 101 (such as the sales distribution module 121 therein), when receiving a request for a new order from the demand side (such as an advertiser), extracts information of each existing order from the order management module 104, from the inventory The estimation module 111 extracts the inventory model, and determines the most bookable quantity of the new order based on the information of each existing order and the inventory model. Great value. In an example, when the demander wants to newly add an order, the order management module 104 can send a query request to the stock allocation module 101, and then the demander can know the maximum amount of bookable quantity that can be set for the order, and the stock allocation. 121 also records the maximum amount of bookable amount for the order. When the demander officially joins the order, the inventory allocation module 101 automatically determines whether the order quantity (or the booked exposure amount or demand amount) of the order is acceptable.

In some examples, the inventory model described above describes each supply node and its corresponding inventory quantity, where a supply node (also referred to as a flow unit) corresponds to a dimension or a combination of dimensions ( Combination of dimensions), the inventory quantity corresponding to the supply node is the exposure quantity corresponding to the dimension or the combination of the dimensions, and various information of the supply node is determined according to historical exposure data (can be obtained by statistics or estimated by an algorithm). In addition, an order can also be characterized by a demand node, which corresponds to the ordering of the order and the demand quantity (ie, the order quantity of the order).

Figure 2 shows the relationship between the supply node and the demand node. Among them, there are 6 supply nodes, they have their own dimension/dimension combination and inventory quantity, and there are N demand nodes, which have their own orientation and demand. For example, the dimension combination of supply node 1 is {Beijing, Sports}, which represents the user who visits the sports channel from Beijing, and the corresponding inventory amount is 8M, representing the number of user visits with this dimension combination {Beijing, Sports} (ie The number of exposure opportunities, also known as the number of exposures, based on historical exposure data is 8M. The orientation of the demand node 1 is {sports}, which represents the user who accesses the sports channel, and the corresponding subscription amount is 15M, which means that the number of times the corresponding order is required to be exposed to the user who accesses the sports channel is 15M. As shown in FIG. 2, for any pair of supply nodes and demand nodes having a connection, the dimension/dimension combination of the supply node is consistent with the orientation of the demand node, and it may be considered to expose the corresponding node corresponding to the user corresponding to the supply node. Order. Regarding how to allocate the inventory quantity of each supply node to each demand node, and to meet the orientation constraint and the reservation quantity requirement of the demand node, the following will be elaborated.

The present application proposes a series of technical solutions aimed at optimizing the processing of the inventory model, service allocation and sales distribution to ensure the exposure flow of the order (ie, the booked amount) and avoid overexposure, thereby improving the performance of the information push system.

3 is a flow chart of an information processing method for an order in an example of the present application. The method is applicable to the inventory allocation module 101 of FIG. 1, and specifically to the inventory estimation module 111. As shown in FIG. 3, the method may include the following steps:

Step 301: Acquire historical user data; wherein the historical user data includes exposure data that has occurred by each user.

Here, historical user data may be acquired from the data platform 102, such as obtaining historical user data from the offline data module 112 therein, wherein the exposure data includes exposure opportunity data that has occurred, and the exposure opportunity for a certain user access is a pointer to this. The secondary user accesses the generated exposure request, but after the service allocation process, the actual exposure is not necessarily generated for the exposure request.

In some examples, the exposure data described above may include page exposure data and media content exposure data (eg, advertisement exposure), page exposure refers to user access to the page, and multiple media content exposure may occur during a user's page visit. .

Step 302: Determine, according to the acquired historical user data, a corresponding exposure quantity for each predetermined dimension or combination of dimensions, and determine a set of exposure quantity ratio, wherein each exposure quantity ratio corresponds to one frequency.

Here, the ratio of the number of exposures corresponding to one frequency may be the ratio of the number of exposures of the frequency to the total amount of exposure of the corresponding dimension or the combination of dimensions, or may be the number of exposures that meet the frequency limit when the frequency is used as the frequency limit. Dimension or dimension combination The ratio of the total amount of exposure.

In some examples, for each dimension or combination of dimensions, the foregoing determining a set of exposure quantity ratios may include: determining, for each dimension or combination of dimensions, an exposure frequency of each user corresponding to the dimension or combination of dimensions, and The exposure quantity ratio is determined separately for each of the determined exposure frequencies to obtain a set of exposure quantity ratios corresponding to the dimension or the combination of dimensions. For example, based on the historical user data, the frequency of access (may be referred to as frequency) of each user under the dimension or combination of dimensions within a predetermined period may be determined. For example, for the combination of dimensions {Beijing, Sports}, it can be determined that within five days, user A from Beijing visited the sports channel once, and user B from Beijing visited the sports channel three times. This can determine the frequency of visits that have occurred under the five-day dimension combination {Beijing, Sports}, including: five days, three times, five days, five times, five times. Each access frequency can correspond to an exposure quantity. For example, there are 100 users who can be determined once every five days (that is, there are 100 exposures), 50 users who have 3 times three times (that is, there are 50 exposures), etc. . The ratio of the number of exposures corresponding to each access frequency to the total number of exposures in the dimension or combination of dimensions is the ratio of the number of exposures for that access frequency. Further, according to the ratio of the number of exposures of each access frequency, the ratio of the number of exposures that meet the frequency limit of each access frequency as the frequency limit to the total number of exposures can be calculated.

Step 303: Establish an inventory model according to each dimension or combination of dimensions and its corresponding exposure quantity and a set of exposure quantity, the inventory model describes each supply node and its corresponding inventory quantity and a group of inventory proportion .

Wherein, one supply node corresponds to one dimension or one dimension combination, and the inventory quantity corresponding to the supply node is the exposure quantity corresponding to the dimension or the dimension combination, and the corresponding group of inventory proportions of the supply node includes the dimension or the combination of the dimensions The proportion of a set of exposures. Here, each supply node and its corresponding set of inventory ratios may also be referred to as a frequency model of inventory (for example, a frequency limit scale model, Frequency Capping) Ratio Model), the frequency model can be considered as part of the inventory model, that is, the inventory model established in the above example is an inventory model that considers the frequency, or an inventory model that includes the frequency model.

Based on the above examples, it can be seen that the established inventory model not only takes into account the inventory of each supply node, but also takes into account the inventory ratio corresponding to each frequency, so that it can be used in subsequent service allocation or sales distribution using this inventory model. Based on the inventory ratio of each frequency, the distribution result is more accurate and reasonable, which can better ensure the order exposure of the order and avoid over-exposure of the order, thereby improving the information push effect and improving system performance.

In step 302, the above process of determining the exposure frequency of each user corresponding to a certain dimension or combination of dimensions and separately counting the proportion of exposures for each determined frequency of exposure may be performed in several ways:

method one:

1. Determine each user and its page access frequency corresponding to the dimension or dimension combination within a predetermined time period.

2. Determine the total number of page visits corresponding to the dimension or dimension combination within the predetermined time period.

3. For each page access frequency, determine the number of users corresponding to the page access frequency to the first proportion of the total number of page visits, and use the first ratio to calculate the number of page visits that satisfy the page access frequency limit as the total number of page visits. The second ratio, and the second ratio as the proportion of exposure.

The frequency model established in this way can be referred to as a static frequency model.

In some instances, only one media content exposure (CV, Content View) is allowed in a user's page visit (PV), for example, only one ad exposure (AV, Advertisement View) is allowed in one PV, ie the same one Ads occupy up to one exposure in the same PV. In this way, the PV ratio of each frequency can be determined, that is, the ratio of the number of PVs satisfying x days and y times to the total number of PVs. It can be seen that the first method can solve the problem of exposure limitation of the same media content in the same PV, and the calculation complexity of the statistical PV ratio is lower, and the data accuracy can be ensured.

Specifically, the page access frequency can be expressed as (n, k), where n is a time parameter and k is a number of times. For example, given continuous n-day historical exposure data, the total number of PVs occurring during that time period is expressed as V _PV (n), and the number of users contributing to k PVs (which can be expressed as UV) is expressed as V _UV (n, k) . Then the ratio of the UV contributing to the PV to the total PV number can be expressed as P _UV (n, k) (ie the first ratio described above) and can be calculated by the following formula (1):

If the frequency control requirement (ie, frequency limit) of an order is expressed as (x, y), where x is the time parameter and y is the number of times, such as the frequency limit is x days y times, the PV inventory that can satisfy the frequency control requirement The ratio is expressed as Q _PV (x, y) (ie, the aforementioned second ratio, indicating the proportion of exposures of the user whose number of page visits is less than or equal to y times within a specified time period x days), and can be expressed by the following formula (2) ) Calculated:

The frequency model (that is, the set of stock ratios corresponding to each supply node) can be described as the following formula (3):

Q _PV ={Q _PV (x,y)|x≥1, y≥1} (3)

Method 2:

1. Determine exposure data for each user in the set of users corresponding to the dimension or combination of dimensions.

2, for each user's exposure data, estimated to consume at each remaining inventory ratio The number of exposures, and the exposure data of the user is updated based on the estimated number of exposures consumed.

3. Determine the exposure frequency of each user according to the updated exposure data of each user.

4. For each exposure frequency, determine, according to the updated exposure data of each user, the proportion of exposures corresponding to the exposure frequency under each remaining inventory ratio.

The exposure data V _PV (u) of each user u in the user set U corresponding to the dimension or the combination of dimensions may be represented as a set: V _PV (u)={V ¹ _PV (u), V ² _PV (u ), ..., V ^k _PV (u)}, wherein the number of exposures occurring on the i-th page access is expressed as V ⁱ _PV (u).

In the above step 2, according to each V ⁱ _PV (u) in the exposure data of each user u, the probability p _PV (x, r) of x exposures is calculated from the historical data for each remaining inventory ratio r, The V ⁱ _PV (u) is updated according to p _PV (x, r).

In the above step 4, the foregoing method 1 can be used to count the proportion of the exposure amount. Specifically, the frequency of the exposure determined in the foregoing step 3 is the page access frequency of each user in the specified time period, and the step 4 includes: determining the total number of page visits of each user in the specified time period; determining the frequency of access for each page. The number of users corresponding to the frequency of access to the page in each remaining inventory ratio accounts for the first proportion of the total number of visits to the page, and the number of page visits satisfying the access frequency limit of the page for each remaining inventory ratio is calculated by using the first ratio. The second ratio of the total number of page visits, and the second ratio under each remaining inventory ratio is taken as the proportion of the exposure amount corresponding to the page access frequency under each remaining inventory ratio.

In some examples, in the second method above, calculating the probability p _PV (x) of x exposures at each remaining inventory ratio and calculating the number of exposures of the V ⁱ _PV (u) deducted according to p _PV (x) The expectation, and updating the V ⁱ _PV (u) according to the calculated expected number of deducted exposures, may specifically include the following processing:

First, the initialization process is performed:

1) Define the number of discretized intervals of the remaining stock ratio, and the discretized remaining stock ratio can be expressed as L={0,1/l,2/l,...,(l-1)/l,1} .

2) Define the consumption probability p*(n,m) of PVs with different CV (for example, AV) numbers at different remaining inventory ratios, that is, the CV consumption probability when the remaining inventory ratio is m for PVs with CV number n. The study found that the following two methods can achieve good results:

First, to imitate the theoretical model without frequency limitation, assuming that CV is consumed in an equal probability manner, that is, p*(n, m) is a constant.

2. When the different remaining inventory ratios are sampled, the PVs of different CV numbers are used in the push log (such as the advertisement log), and a regression algorithm is established to learn.

Based on the above initialization process, the specific calculation steps are as follows:

1) Let the remaining stock ratio r = 100%.

2) When r<1, the loop executes the following process to consume the expected ratio equal to

Stock:

a) For any u∈U, do the following:

I.for any

The probability of the CV of the PV being deducted is

Then the probability that the PV is deducted by x CVs p _PV (x) is calculated according to the following formula (4):

Ii. The distribution of the probability can be expressed as the following formula (5):

Iii. Generate a random number r between 0-1, and the number of CVs that the current PV should deduct is expressed as

v=y,r>Q _PV (y-1) and r≤Q _PV (y)

Iv. update the current number based on the number of CVs that should be deducted

E.g:

b) Calculate the exposure number ratio Q _PV (x, y) corresponding to each frequency under the current remaining inventory ratio according to the foregoing manner, and let q _PV (x, y, r) = Q _PV (x, y).

c)

The q _PV (x, y, r) obtained in the above mode 2 is a frequency model considering the influence of the remaining stock ratio. This frequency model can be referred to as a dynamic frequency model (such as a dynamic frequency-restricted scale model).

In the second mode, the frequency variation in the case of inventory consumption can be simulated, that is, the inventory structure is restored according to the historical log data, the inventory consumption is simulated, and the proportion of exposures corresponding to each frequency under different inventory remaining ratios is learned, so that the service distribution is performed subsequently. When the distribution is distributed, the number of exposures of each frequency after the inventory is estimated according to the inventory model is more accurate, which can further optimize the distribution result, and can more accurately allocate the order exposure of the order, and better guarantee the order. Book exposure and avoid over-exposure of orders, which improves information push and improves system performance.

The inventors also found in the study that the proportion of exposures of each frequency is directly calculated from each supply node i, and is applied to the corresponding supply node set Γ(j) corresponding to the order orientation to generate an error. Because usually Γ(j) contains multiple supply nodes, the direct summation of the number of exposures to each supply node will cause a problem of overlapping traffic, resulting in a larger proportion of exposures corresponding to the calculated frequency. For example, for the frequency 7 times 3 times, the exposures of the dimensions {movie} and {television} account for 20%, but when the order is directed to "movie + TV series", there must be some users watching movies and TV shows. Therefore, for order-oriented "movie + TV series", the actual number of exposures corresponding to the frequency of 7 times 3 days will be less than 20%. The precise method is to take all the possible Γ(j) from history Learn the respective frequency models in the data. However, there are tens of thousands of supply nodes in the actual system, and the combination of Γ(j) will be quite large, and the calculation of the frequency model in turn is very large.

In some examples, for the above problem, the following methods are used to obtain the above predetermined dimensions or combinations of dimensions:

1. Extract multiple orientations of historical orders from historical user data.

2. Perform clustering calculation on multiple orientations to obtain at least one orientation group.

3. Combining the at least one orientation group as the predetermined dimension and/or dimension.

With this method, the exposure data across the order-oriented constraints can be counted, and the problem of overlapping UVs of multiple flow units can be solved, which can improve the accuracy of statistics, reduce the calculation amount, and improve the system efficiency.

The above clustering calculation method can be specifically as follows:

First, define the orientation similarity of any two orders a and b as calculated by the following formula (6):

Where V(x) represents the number of page visits (which can be considered as the number of exposures) determined for the supply node x,

Represents a collection of supply nodes that match the orientation of order a.

Secondly, using the above similarity calculation method, the similarity between any two orders can be calculated. According to the calculated similarities and arbitrary clustering algorithms, all the orientations that have appeared in history can be divided into multiple orientation groups. .

In some examples, for each orientation group, an orientation is determined as an orientation representation of the orientation group, such as taking a union of orientations within the group or selecting one of the orientations as an orientation representation, and combining the orientation representation as a dimension or dimension combination, According to historical data The frequency model represented by the orientation is to combine the orientation representation as a dimension or a dimension and learn the corresponding frequency model according to the foregoing method.

Finally, in the case of sales distribution and service allocation, for any order, the orientation group most similar to the orientation of the order is found according to the above similarity formula (6), and the frequency model of the orientation group is used as the frequency model of the order. Here, as described above, the frequency model corresponding to each dimension/dimension combination is established, and on this basis, the frequency model corresponding to the orientation of each historical order can be determined, wherein the frequency model corresponding to the orientation of an order is The frequency model corresponding to the dimension or dimension combination that is aligned.

The above solution solves the problem of user overlap of different traffic units, and also greatly reduces the algorithm complexity and computational overhead by losing the accuracy of a small number of orders. The data shows that dozens of orientations can accurately serve all orders in history; at the same time, the more similar the orientation, the more similar the frequency structure. Therefore, the impact of this scheme on system accuracy is very small.

In some instances, the establishment of the frequency model described above can be calculated offline and the results stored in a cache. The mitigation results can be directly used when online, and there is no additional time overhead for online calculation. In other examples, the above frequency model can be calculated online, making the settlement result more real-time and more accurate.

Using the solution of establishing the inventory model (including the frequency model) provided by the above example, the average order quantity of the order in the sales distribution can be increased by 20%, and the weighted deviation of distribution and delivery in the service allocation is reduced by 35%, which greatly improves The accuracy of the order for pushing information during the sale and delivery process.

In addition, the present application also proposes an allocation method based on the above-mentioned consideration of the inventory model, including the service allocation method of the order and the sales distribution method of the order.

4 is a flow chart of a service allocation method for an order in an example of the present application. The method is applicable to the inventory allocation module 101 of FIG. 1, and specifically to the service allocation module 131. As shown in FIG. 4, the method includes the following steps:

Step 401: Extract information of each order from the order management module 104.

Step 402: Extract the inventory model established by the foregoing method from the inventory estimation module 111.

Step 403: Generate an allocation plan according to the information of each order and the inventory model, and the allocation plan includes a service rate of each order.

Among them, for each order, the following processing is performed to determine the service ratio:

1. Using the inventory model, determining an inventory ratio corresponding to the frequency limit of the order corresponding to all the supply nodes corresponding to the orientation of the order;

2. Determine the service ratio of the order based on the inventory amount corresponding to each supply node corresponding to the orientation of the order, the current estimated stock remaining amount, the determined one inventory share, and the order quantity of the order.

Step 404: Providing the above-mentioned allocation plan to the push server 103, so that the push server 103 performs the menu processing according to the service ratio in the allocation plan.

Wherein, when an exposure request for a plurality of users is received from the push server 103, information of a plurality of orders directed to the users is extracted from the order management module 104, and then determined according to the distribution plan obtained from the stock allocation module 101. The service ratios corresponding to each of these orders are extracted to make a menu for these orders. The push server 103 can provide an information push service based on the menu result, for example, according to the menu result, pushing information to the corresponding user (such as placing an advertisement) or determining that the menu fails does not have to push information to any user.

In some instances, when the order is fetched, one is to perform directional filtering, that is, to extract each order that matches the users, and the second is to perform frequency filtering, that is, to extract the frequency-constrained requirements from the orders that are aligned with the users. The order, that is to say, for each of the orders, determines whether the order has reached the frequency limit number of times for the corresponding user, and then extracts the order that has not reached the frequency limit number.

In the above example, when the service allocation is performed, the service ratio of each order can be calculated based on the inventory ratio of the frequency restriction condition extracted from the inventory model, so that the service ratio is more accurate, thereby making the exposure received in real time. The requested menu results are more accurate and more able to meet the information push requirements, which can better guarantee the order flow and prevent overexposure.

In some examples, when generating an allocation plan, the inventory quantity of each order's supply node that matches its orientation may be further determined based on the inventory model, and each order is sorted by inventory quantity to determine an allocation order. Then, in this order of allocation, the service ratio of each order is determined in turn.

In some instances, for order j, the service ratio is expressed as α _j , and each order may be sorted in a certain order of allocation, and then the service ratio of each order is sequentially calculated in this order. There are two order of this assignment:

1. The inventory quantity of the supply node corresponding to the orientation of the order is arranged in ascending order, wherein the inventory quantity of the matching supply node corresponding to the order j (regardless of the influence of the frequency model) can be expressed as S _j (S _j = ∑ _{i ∈Γ(j)} s _i ), if considering the influence of the frequency model proposed in the present application, this stock quantity can be expressed as S' _j (S' _j = ∑ _{i ∈Γ (j)} s _i f _{Γ (j)} ).

2. The order between the order quantity and the stock quantity of the supply node corresponding to the above is arranged in descending order, wherein the order quantity (or demand quantity) of the order j is expressed as d _j , and the ratio is expressed as d _j /S _j . With this order of assignment, orders with the same orientation but with different booking quantities can be prioritized.

In some examples, the set of inventory ratios corresponding to each of the supply nodes described by the inventory model includes: an inventory share corresponding to each frequency limit under each remaining inventory ratio. At this time, the service ratio of each order j in the above step 403 is determined as follows.

Wherein, the estimated remaining inventory quantity of each supply node i is defined as r _i , and the specific processing flow for determining the service ratio α _j for each order j is as follows:

1. Initialize the estimated remaining inventory of all supply nodes i as r _i = s _i , where s _i is the inventory of supply node i in the inventory model.

2. Perform the following processing for each order j in the predetermined order of allocation:

a) Calculate the inventory remaining rate p by the following formula (7):

b) extracting, from the previously established inventory model (including the frequency model), each supply node i (ie, each dimension and/or combination of dimensions) that matches the orientation of the order j, meeting the frequency constraints of the order One share of inventory

Wherein, given an order j and an inventory remaining ratio p, an inventory ratio f(Γ(j), n _j , m _j , p) corresponding to the frequency restriction condition (n _j , m _j ) of the order j can be determined, It can be simplified as f ^* (j, p). In some examples, if the orientation of the order has been clustered to generate a plurality of orientation groups, the orientation group Γ'(j) most similar to the orientation of the order j can be determined, and based on this Γ'(j) To calculate f ^* (j, p) = f(Γ'(j), n _j , m _j , p).

Among them, for orders without frequency restrictions,

c) Determine the service ratio α _{j of} the order j considering the above inventory ratio according to the following formula (8):

If it is not possible to determine α _j that satisfies the formula (8), α _j =1.

d) for each supply node i (i∈Γ(j)) corresponding to the order j, update its remaining inventory r _i ,

In the above example, the inventory ratio of each frequency under different inventory residual ratios is considered in the inventory model, so that when the service is allocated, the exposure quantity of each frequency after the inventory is estimated according to the inventory model is more accurate, and then It can further optimize the distribution result, and can more accurately allocate the order exposure of the order, better guarantee the order exposure of the order and avoid over-exposure of the order, thereby improving the information push effect and improving the system performance.

The above process of determining the ratio of each order service can be calculated while offline. After receiving the online exposure request, information of a plurality of orders corresponding to the exposure request is extracted, and the service ratio of the orders is determined according to the generated distribution plan, so that the push server 103 can perform online based on the service ratio of the orders. Menu processing. Here, for an order, the service ratio of the order for the online menu determined according to the distribution plan may be the same as or slightly smaller than the service ratio of the order included in the distribution plan. That is to say, the service ratio in the allocation plan generated in step 403 represents the service ratio value that the order may select during the actual online menu selection process, but the actual online menu does not necessarily use the service ratio in the distribution plan, which may be based on actual conditions. Adjustment.

In some examples, given an exposure request i, J={c ₁ ,...,c _|J| } is a set of orders that match the exposure request, wherein the orders are filtered by directed matching, frequency control, etc. And obtained, and arranged in the order of allocation. If the sum of the service ratios of these orders is greater than 1,

Then find a maximum l to make

Recalculate the service ratio of the l+1th order in the above allocation order

Thus, α' _l+1 <α _l+1 . After that, an order j(j∈[1,l]) is selected from the new order set J'={c ₁ ,...,c _l }.

FIG. 5 is a flow chart of an inquiry method of an order in an example of the present application. This polling method can The sales distribution process that belongs to the order. The method is applicable to the inventory allocation module 101 of FIG. 1, and specifically to the sales distribution module 121. When a request for a new order (such as a maximum orderable query request) is received from the demanding party, the query processing of the maximum orderable amount for this new order is required, which is part of the sales allocation. The demand side refers to the party who wants to purchase an order (for example, the advertiser who wants to purchase an advertising order), and needs to make a bookable quantity inquiry before determining the purchase to determine the maximum amount of the orderable quantity of the order. An inquiry request is sent to the inventory allocation module 101 by the order management module 103, which carries information of the new order to be purchased.

As shown in Figure 5, the method is divided into three phases, including the following steps:

(Phase 1: Inventory attributable to each order before the new order is assigned)

Step 501: Extract the information of each existing order from the order management module 103, and extract the inventory model established by the foregoing method from the inventory estimation module 111.

Step 502: Determine the order of allocation of the new order and the existing orders according to the new order and the orientation of each existing order.

In some examples, the process of determining an allocation order may include determining an inventory quantity of each order's supply node that matches its orientation based on the inventory model, and ordering each order by inventory quantity to determine an allocation order. For example, according to the orientation of each order, the allocation order can be determined according to the first manner described above, that is, the inventory quantity S' _j of the supply node i corresponding to the orientation of the order _j (S' _j = ∑ _i∈Γ(j) s _i f _Γ(j) ) is sorted in ascending order.

Step 503: Perform the following processing for each existing order before the new order in the determined allocation order:

1) Using the inventory model, determine an inventory ratio corresponding to the frequency limit of the existing order corresponding to all the supply nodes that match the orientation of the existing order.

2) an inventory amount corresponding to each supply node corresponding to the orientation of the existing order, a current estimated inventory remaining amount, the determined one inventory share ratio, and the existing order The amount of reservations, determine the service ratio for the existing order and update the current estimated stock remaining.

(Phase 2: Define query boundaries)

Step 504: Using the inventory model, determine an inventory ratio corresponding to the frequency limit of the order corresponding to all the supply nodes that match the orientation of the new order.

Step 505: Determine an initial value of the reservation quantity upper limit according to the inventory quantity corresponding to each supply node corresponding to the orientation of the new order and an inventory ratio, and set the initial value of the reservation quantity lower limit to a preset value (for example, set to 0). .

(Phase 3: Maximum number of probe bookings)

Step 506: Perform the following processing on each order in order from the new order in the order of allocation until the lower limit of the reservation quantity is greater than or equal to the upper limit of the reservation quantity:

1) Using the inventory model, determine an inventory ratio corresponding to the frequency constraints of the current order for all supply nodes that match the orientation of the current order.

2) Attempt to determine the service ratio of the current order based on the inventory amount corresponding to each supply node corresponding to the orientation of the current order, the current estimated inventory remaining amount, the above-mentioned inventory occupancy ratio, and the current orderable quantity detection value, and update the current Estimating the remaining amount of the stock, adjusting the lower limit of the predetermined booking amount or lowering the upper limit of the predetermined amount according to whether the service ratio of the current order can be determined, and adjusting the minimum amount according to the predetermined amount of reservation and the upper limit of the predetermined amount The amount of reservation detection value. Wherein, the initial value of the reservation amount upper limit is taken as the initial value of the recordable amount detection value.

Step 507: The final value of the reservation amount upper limit is taken as the maximum bookable amount of the new order.

After that, the maximum amount of the reservation can be used to determine whether the reservation amount of the new order is acceptable.

Here, the maximum amount of the reservation can be used as the sales distribution module 121 as an output parameter. The number is fed back to the order management module 104, and the order management module 104 can determine whether the subscription amount of the new order can be allowed; or, the sales distribution module 121 directly determines whether the reservation amount of the new order is allowed according to the maximum amount of the reservation amount, The judgment result is fed back to the order management module 104 as an output parameter.

In the above example, when the demand side performs the bookable quantity inquiry (ie, the inquiry quantity) for a new order, the order of each order before the new order can be allocated based on the inventory proportion of the frequency restriction condition extracted from the inventory model. The inventory, define the query boundary, and detect the maximum amount of bookable quantity, so that the maximum amount of the bookable quantity obtained by the detection is more accurate, and it can better meet the requirements of the order sale distribution, and can better limit the reservation quantity in the order sale (ie, Order flow or order exposure) to prevent over-selling or under-selling of order traffic, thereby better ensuring order traffic in information push and improving the performance of information push services.

In some examples, a set of inventory ratios corresponding to all supply nodes that are consistent with a certain orientation described by the inventory model may include: an inventory ratio corresponding to each frequency limit under each remaining inventory ratio, at this time, in step 503 1) The inventory ratio corresponding to the current order determined in the point may include: the frequency of all the supply nodes in the inventory model that match the orientation of the current order j, and the frequency of the order j in the current estimated remaining inventory ratio p Restricted inventory ratio

Where Γ(j) is a set of supply nodes corresponding to the orientation of the order j, the frequency limit is expressed as (n _j, m _j ), n _j is a time parameter, and m _j is a number of times.

In some instances, the estimated remaining inventory for each supply node i is defined as r _i and the new order is order x.

The specific processing of step 503 is as follows:

2. In the order of allocation, for each order j before the order x, the following processing is performed:

a) Calculate the remaining stock ratio p according to formula (7):

b) extracting from the previously established inventory model (including the frequency model) all the supply nodes i (ie, the supply node set Γ(j) corresponding to the orientation of the order j) that match the orientation of the order j, and conform to the order An inventory share of the frequency limit

Regarding f ^* (j, p), the foregoing has been described in detail, and will not be described again here.

Among them, for orders without frequency restrictions,

c) Determine the service ratio α _{j of} the order j considering the above inventory ratio according to formula (8):

Step 504 specifically includes processing:

1. Calculate the remaining stock ratio p of the new order x according to formula (7):

2, using the inventory model to get a stock ratio of new orders x

In step 505, the lower limit of the reservation amount d _lower initial value is set to 0, and the initial value of the upper limit of the reservation amount d _upper can be determined by the following formula (9):

The specific processing of step 506 includes:

1. Save the current estimated remaining inventory r' _{i of} all supply nodes i, ie r' _i = r _i .

2. Set the initial value d _x =d _{upper of the} reservation amount detection value d _x .

3. Repeat the following steps until d _lower ≥d _upper :

a) For each new order x in the order of assignment and each order j in each subsequent order, perform the following processing:

i. Calculate the remaining stock ratio p of order j according to formula (7):

Ii. Use the inventory model to get an inventory share of order j

Iii. Determine the service ratio α _{j of} the order j considering the above inventory ratio according to formula (8):

If it is not possible to determine α _j that satisfies equation (8), then let d _upper =d _x and perform the following steps c)

Iv. for each supply node i (i∈Γ(j)) corresponding to the order j, update its remaining inventory r _i ,

b) Let d _lower = d _x , that is, raise the lower limit of the reservation amount.

c) order

The upcoming amount of detection value d _{x is} lowered to the mean of the upper and lower limits. At the same time, the current estimated remaining inventory r _{i of} all supply nodes i is restored to r' _i saved in step 1. Thereafter, the process returns to step a) to re-process the new order x and each of the subsequent orders j to perform the processing of i to iv in the above step a).

In the above example, the inventory ratio of each frequency under different inventory residual ratios is considered in the inventory model, so that when the order is inquired, the exposure quantity of each frequency after the inventory is estimated according to the inventory model is more accurate. The detection of the maximum amount of reservations is more accurate, more able to meet the requirements of order sales distribution, can better limit the amount of reservations in the order sale (ie order flow or order exposure), to prevent over-sale or sale of order traffic Insufficient, in order to better ensure the order flow in the information push, improve the performance of the information push service.

Based on the above examples, the present application also proposes an information processing apparatus for pushing information orders. The device can be located in an inventory allocation module 101, such as inventory estimation module 111 therein. As shown in FIG. 6, the information processing apparatus 600 includes:

The data extraction module 601 acquires historical user data; wherein the historical user data includes exposure data that has occurred by each user. For example, historical user data is obtained from data platform 102 (eg, offline data module 112).

The calculation module 602 determines, according to the historical user data acquired by the data extraction module 601, a corresponding exposure quantity for each predetermined dimension or combination of dimensions, and determines a proportion of exposure quantity, wherein each exposure quantity corresponds to one frequency.

The model establishing module 603 establishes an inventory model according to each dimension or combination of dimensions determined by the calculation module 602 and its corresponding exposure quantity and the set of exposure quantity, the inventory model describes each supply node and Corresponding stock quantity and a group of stock ratio.

Wherein, one supply node corresponds to one dimension or one dimension combination, and the inventory quantity corresponding to the supply node is the exposure quantity corresponding to the dimension or the dimension combination, and the corresponding group of inventory proportion of the supply node includes the dimension or the dimension The ratio of the set of exposures corresponding to the combination is proportioned.

In some examples, the calculation module 602 determines, for each dimension or combination of dimensions, the frequency of exposure of each user corresponding to the dimension or combination of dimensions, and determines the proportion of exposures for each of the determined exposure frequencies to obtain the dimension. Or a proportion of the number of exposures corresponding to a combination of dimensions.

In some examples, the calculating module 602 determines, according to the dimension or the combination of the users in the specified time period, the user and the page access frequency thereof; and determines the total number of page visits corresponding to the dimension or the dimension combination in the specified time period; Frequency of page visits, determining a first ratio of the number of users corresponding to the frequency of access to the page to the total number of times of accessing the page, and using the first ratio to calculate the number of page visits satisfying the access frequency limit of the page as the total number of page visits The second ratio is used as the ratio of the exposure amount.

In some examples, the calculation module 602 determines exposure data of each user in the user set corresponding to the dimension or combination of dimensions; for each user's exposure data, estimates the number of exposures consumed under each remaining inventory ratio, and Updating the exposure data of the user according to the estimated number of exposures consumed; determining an exposure frequency of each user according to the updated exposure data of each user; for each exposure frequency, according to each updated user The exposure data determines the proportion of exposures corresponding to the frequency of exposures at each remaining inventory ratio.

In some examples, the apparatus 600 further includes:

The clustering module 604 extracts multiple orientations of the historical order from the historical user data acquired by the data extraction module 601, performs clustering calculation on the multiple orientations to obtain at least one orientation group, and uses the at least one orientation group as a location Said predetermined dimensions and/or said dimensions Degree combination.

The apparatus 600 may be a computing device, wherein each module may be an instruction module, such that the apparatus 600 may include one or more memories and one or more processors; wherein the one or more memories store one or The one or more instruction modules are configured to be executed by the one or more processors; wherein the one or more instruction modules comprise: any one or any combination of the above modules 601-604.

Based on the above examples, the present application also proposes a service distribution device for pushing information orders. The device can be located in an inventory allocation module 101, such as the service distribution module 131 therein. As shown in FIG. 7, the service distribution device 700 includes:

The data extraction module 701 extracts information of each order and extracts the aforementioned inventory model. For example, the information of each order is extracted from the order management module 104, and the inventory model is extracted from the inventory estimation module 111.

The distribution planning module 702 generates an allocation plan according to the information of each order and the inventory model extracted by the data extraction module 701, the distribution plan includes a service ratio of each order, wherein the service ratio of each order is determined as follows :

Using the inventory model, determining an inventory ratio corresponding to the frequency limit of the order corresponding to all supply nodes corresponding to the orientation of the order; and

The service ratio of the order is determined based on the inventory amount corresponding to each supply node corresponding to the orientation of the order, the current estimated inventory remaining amount, the determined one inventory share ratio, and the booked quantity of the order.

The output module 703 provides an allocation plan generated by the distribution planning module 702 to the push server 103 to cause the push server 103 to perform a menu process according to the service ratio in the distribution plan.

In some examples, the allocation plan module 702, when generating the allocation plan, further The inventory quantity of each order corresponding to the orientation of each order is determined according to the inventory model, and each order is sorted according to the inventory quantity to determine an allocation order; then, in this allocation order, the service ratio of each order is determined in turn.

The apparatus 700 described above can be a computing device, wherein each module can be an instruction module, such that the apparatus 700 can include one or more memories and one or more processors; wherein the one or more memories store one or The one or more instruction modules are configured to be executed by the one or more processors; wherein the one or more instruction modules comprise: any one or any combination of the above modules 701-703.

Based on the above examples, the present application also proposes an inquiry device for pushing information orders, which may be located in the inventory allocation module 101 (such as the sales distribution module 121 therein). When a request for a new order is received from the demanding party, as shown in FIG. 8, the polling device 800 includes:

The data extraction module 801 extracts information of each existing order and extracts the above inventory model. For example, the information of each order is extracted from the order management module 104, and the inventory model is extracted from the inventory estimation module 111.

The distribution planning module 802 determines the order of allocation of the new order and the existing orders according to the new order and the orientation of each existing order; in the order of the allocation, the following orders are sequentially processed for each existing order before the new order :

Determine the service ratio of the current order and update the current order based on the inventory amount corresponding to each supply node corresponding to the orientation of the current order, the current estimated inventory remaining amount, the determined one inventory share, and the current order subscription amount. Estimated stock remaining.

a query boundary module 803, using the inventory model, to determine an inventory ratio corresponding to the frequency limit of the new order corresponding to all the supply nodes corresponding to the orientation of the new order; each according to the orientation of the new order The inventory quantity corresponding to the supply node and the one inventory ratio are determined, the initial value of the reservation quantity upper limit is determined, and the initial value of the reservation quantity lower limit is set to a preset value.

The reservation amount detecting module 804 performs the following processing on each order in order from the new order in the order of allocation until the lower limit of the reservation amount is equal to the upper limit of the reservation amount:

Using the inventory model to determine an inventory share of all supply nodes corresponding to the current order's orientation that meets the frequency limit of the current order;

Attempt to determine the service ratio of the current order and update the current pre-determination according to the inventory quantity corresponding to the orientation of the current order, the current estimated inventory remaining quantity, the one inventory share ratio, and the current bookable quantity detection value. Estimating the remaining amount of the inventory, adjusting the lower limit of the reservation amount or lowering the value of the upper limit of the reservation amount according to whether the service ratio of the current order can be determined, and adjusting the quantity according to the lower limit of the reservation quantity and the upper limit of the reservation quantity a reservation amount detection value; wherein the reservation amount upper limit initial value is used as an initial value of the bookable amount detection value.

Finally, the booking amount detection module 804 takes the final value of the reservation amount upper limit as the maximum amount of the new order's bookable amount. The reservation amount detection module 804 can feed back the maximum amount of the orderable quantity of the new order to the order management module 104 for reference when the demand side sets the reservation amount of the new order.

The apparatus 800 may be a computing device, each of which may be an instruction module, such that the apparatus 800 may include one or more memories and one or more processors; wherein the one or more memories store one or One or more instruction modules configured to be executed by the one or more processors; wherein the one or more instruction modules comprise: the above modules 801-804 Either or any combination.

The implementation principles of the above modules are detailed in the foregoing, and are not described here.

The present application also proposes an order distribution system for pushing information. The system can include the inventory allocation module 101 shown in FIG. The inventory allocation module 101 may include an inventory estimation module 111, a sales distribution module 121, and a service distribution module 131.

In some examples, inventory estimation module 111 includes inventory model establishing device 600 described above.

Further, in some examples, the service distribution module 131 includes the service distribution device 700 described above.

Further, in some examples, the sale distribution module 121 includes the query device 800 described above.

Each of the modules in the inventory allocation module 101 can implement the processing in the various examples described above through interaction with the data platform 102, the push server 103, and the order management module 104. The specific implementation principle has been described in the foregoing and will not be described here.

In addition, the devices and modules in the various examples of the present application may be integrated into one processing unit, or each module may exist physically separately, or two or more devices or modules may be integrated into one unit. The above integrated unit can be implemented in the form of hardware or in the form of a software functional unit.

In some examples, the various devices described above can be run in various computing devices and loaded into the memory of the computing device.

The present application proposes a computing device that can operate any one or any combination of the above. As shown in FIG. 9, the apparatus includes the computing device including one or more processors (CPUs) 902, a communication module 904, a memory 906, a user interface 910, and a communication bus 908 for interconnecting these components.

The processor 902 can receive and send data through the communication module 904 to implement network communication. Letter and / or local communication.

User interface 910 includes one or more output devices 912 that include one or more speakers and/or one or more visual displays. User interface 910 also includes one or more input devices 914 including, for example, a keyboard, a mouse, a voice command input unit or loudspeaker, a touch screen display, a touch sensitive tablet, a gesture capture camera or other input button or control, and the like.

The memory 906 can be a high speed random access memory such as DRAM, SRAM, DDR RAM, or other random access solid state storage device; or a non-volatile memory such as one or more magnetic disk storage devices, optical disk storage devices, flash memory devices, Or other non-volatile solid-state storage devices.

The memory 906 stores a set of instructions executable by the processor 902, including at least:

Operating system client 916 includes programs for processing various basic system services and for performing hardware related tasks.

Application 918 includes an application for implementing a process flow of any or any combination of the above examples.

In some examples, application 918 can include device 600 of FIG. 6, device 700 of FIG. 7, and/or device 800 of FIG. 8, each of

devices

600, 700, and/or 800 can store a machine executable instruction. The processor 902 can implement the functions of the various modules by executing machine executable instructions in the various modules in the memory 906.

It should be noted that not all the steps and modules in the foregoing processes and the various structural diagrams are necessary, and some steps or modules may be omitted according to actual needs. The order of execution of each step is not fixed and can be adjusted as needed. The division of each module is only for the convenience of description of the functional division. In actual implementation, one module can be implemented by multiple modules, and the functions of multiple modules can also be implemented by the same module. These modules can be located in the same device. It can also be located in different devices.

The hardware modules in each example may be implemented in a hardware manner or a hardware platform plus software. The above software includes machine readable instructions stored in a non-volatile storage medium. Therefore, each instance can also be embodied as a software product.

Accordingly, the present application also provides a storage medium in which is stored a data processing program for performing any of the above-described methods of the present application.

In various examples, the hardware may be implemented by specialized hardware or hardware that executes machine readable instructions. For example, the hardware can be a specially designed permanent circuit or logic device (such as a dedicated processor such as an FPGA or ASIC) for performing a particular operation. The hardware may also include programmable logic devices or circuits (such as including general purpose processors or other programmable processors) that are temporarily configured by software for performing particular operations.

The machine readable instructions corresponding to the modules of Figures 6-9 may cause an operating system or the like operating on a computer to perform some or all of the operations described herein. The non-transitory computer readable storage medium may be inserted into a memory provided in an expansion board within the computer or written to a memory provided in an expansion unit connected to the computer. The CPU or the like installed on the expansion board or the expansion unit can perform part and all of the actual operations according to the instructions.

The non-transitory computer readable storage medium includes a floppy disk, a hard disk, a magneto-optical disk, an optical disk (such as a CD-ROM, a CD-R, a CD-RW, a DVD-ROM, a DVD-RAM, a DVD-RW, a DVD+RW), and a magnetic tape. , non-volatile memory card and ROM. Alternatively, the program code can be downloaded from the server computer by the communication network.

In conclusion, the scope of the claims should not be limited to the embodiments in the examples described above, but the description should be construed as a whole and the broadest explanation.

Claims

An information processing method for pushing information orders, comprising:

Obtaining historical user data; wherein the historical user data includes exposure data that has occurred by each user;

Determining, according to the historical user data, a corresponding exposure quantity for each predetermined dimension or combination of dimensions, and determining a proportion of exposure quantity, wherein each exposure quantity ratio corresponds to one frequency;

Establishing an inventory model according to each dimension or combination of dimensions and its corresponding number of exposures and the proportion of the set of exposures, the inventory model describing each supply node and its corresponding inventory amount and a set of inventory ratio ;

Wherein, one supply node corresponds to one dimension or one dimension combination, and the inventory quantity corresponding to the supply node is the exposure quantity corresponding to the dimension or the dimension combination, and the corresponding group of inventory proportion of the supply node includes the dimension or the dimension The ratio of the set of exposures corresponding to the combination is proportioned.
The method of claim 1 wherein for each dimension or combination of dimensions, said determining a percentage of exposures comprises:

For each dimension or combination of dimensions, determining the frequency of exposure of each user corresponding to the dimension or combination of dimensions, and determining the proportion of exposures for each of the determined exposure frequencies to obtain a set of exposures corresponding to the dimension or combination of dimensions Proportion.
The method of claim 2, wherein

Determining an exposure frequency of each user corresponding to the dimension or the combination of the dimensions, and determining an exposure quantity ratio for each of the determined exposure frequencies, including:

Determining each user and its page for that dimension or combination of dimensions over a specified time period Frequency of visits;

Determining the total number of page visits for the dimension or combination of dimensions over a specified time period;

For each page access frequency, determining the first proportion of the number of users corresponding to the page access frequency to the total number of page visits, and using the first ratio to calculate the number of page visits that satisfy the page access frequency limit as the total number of page visits The second ratio is taken as the ratio of the exposure amount.
The method according to claim 3, wherein the page access frequency corresponding to the dimension or combination of dimensions is represented as (n, k), wherein n is a time parameter and k is a number of times;

The first ratio is calculated according to the following formula (1):

Wherein, P UV (n, k) is the first ratio, and V UV (n, k) is the number of users corresponding to the page frequency within a specified time period n, and V PV (n) is within the specified time period n The total number of page visits for this dimension or combination of dimensions.
The method according to claim 3, wherein the page access frequency corresponding to the dimension or combination of dimensions is represented as (x, y), wherein x is a time parameter and y is a number of times;

The second ratio is calculated according to the following formula (2):

Where Q PV (x, y) is the second ratio, and P UV (x, i) is the first ratio.
The method according to claim 2, wherein the determining the frequency of exposure of each user corresponding to the dimension or the combination of dimensions, and determining the proportion of the exposure amount for each of the determined exposure frequencies, comprises:

Determining exposure data of each user in the user set corresponding to the dimension or combination of dimensions;

Estimating the number of exposures consumed under each remaining inventory ratio for each user's exposure data, and updating the exposure data of the user according to the estimated number of exposures consumed;

Determining the frequency of exposure of each user according to the updated exposure data of each user; and

For each exposure frequency, the exposure amount corresponding to the exposure frequency is determined according to the updated exposure data of each user.
The method of claim 6, wherein the exposure data V PV (u) of each user u in the set of users comprises: V PV (u) = {V 1 PV (u), V 2 PV (u) ,..., V k PV (u)}, where the number of exposures occurring on the i-th page access is expressed as V i PV (u);

The estimating the number of exposures consumed under each remaining inventory ratio, and updating the exposure data of the user according to the estimated number of exposures, including:

According to each V i PV (u) in the V PV (u) of each user u, learn from the historical data the probability p PV (x, r) of x exposures for each remaining inventory ratio r, and according to p PV (x, r) updates the V i PV (u).
The method according to claim 6, wherein said determining an exposure frequency of each user comprises: determining a page access frequency of each user within a specified time period;

Determining, according to the updated exposure data of each user, an exposure quantity ratio corresponding to the exposure frequency under each remaining inventory ratio, including:

Determine the total number of page visits by each user during the specified time period;

For each page access frequency, determining a first ratio of the number of users corresponding to the page access frequency in each remaining inventory ratio to the total number of page visits, using the first The ratio calculates a second proportion of the number of page visits satisfying the page access frequency limit for each page of the remaining inventory ratio as a total number of page visits, and the second ratio under each remaining inventory ratio is taken as each remaining inventory ratio. The number of exposures corresponding to the frequency of page visits.
The method of claim 1 further comprising:

Extracting multiple orientations of historical orders from the historical user data;

Performing a clustering calculation on the plurality of orientations to obtain at least one orientation group; and

The at least one orientation group is combined as the predetermined dimension and/or the dimension.
The method of claim 9, wherein said combining said at least one orientation group as said predetermined said dimension and/or said dimension comprises:

For each orientation group, an orientation is determined as the orientation representation of the orientation group and the orientation representation is combined as a dimension or dimension.
A service allocation method for pushing information orders, including:

Extract information for each order;

Extracting the inventory model according to claim 1;

An allocation plan is generated based on the information of each order and the inventory model, the allocation plan including a service ratio of each order, wherein the service ratio of each order is determined as follows:

Using the inventory model, determining an inventory ratio corresponding to the frequency limit of the order corresponding to all supply nodes corresponding to the orientation of the order; and

According to the inventory amount corresponding to each supply node corresponding to the orientation of the order, when Determining the remaining amount of stock, the determined stock ratio and the booking amount of the order, determining the service ratio of the order;

The allocation plan is provided to a push server to cause the push server to perform a menu process according to the service ratio in the allocation plan.
The method of claim 11, wherein when the allocation plan is generated, the method further comprises:

Determining, according to the inventory model, an inventory quantity of a supply node of each order that matches its orientation, and sorting each order according to the inventory quantity to determine an allocation order;

Wherein, according to the allocation order, the service ratio of each order is sequentially determined.
The method according to claim 12, wherein the stock amount S j of the supply node in accordance with the orientation of the order j is determined by the following formula (3):

S j =Σ i∈Γ(j) s i f Γ(j) (3)

Where Γ(j) is the set of supply nodes that match the orientation of order j, s i is the inventory quantity of supply node i extracted from the inventory model, and f Γ(j) is all supplies in Γ(j) The proportion of inventory of nodes that meets the frequency limit of order j.
The method according to claim 11, wherein a set of inventory proportions corresponding to each of the supply nodes described by the inventory model comprises: an inventory ratio corresponding to each frequency limit under each remaining inventory ratio;

Among them, the service ratio of each order j is determined as follows:

Calculate the remaining stock ratio p by the following formula (4):

Where Γ(j) is the set of supply nodes that match the orientation of the order j, r i is the current estimated stock remaining amount of the supply node i, and s i is the stock quantity of the supply node i extracted from the inventory model ;

Extracting, from the inventory model, the inventory ratio of all supply nodes that match the orientation of the order j, which meets the frequency limit of the order under the calculated remaining inventory ratio p
Wherein, the frequency limit is expressed as (n j , m j ), n j is a time parameter, and m j is a number of times;

Calculate the service ratio α j of the order j according to the following formula (5):

Where d j is the predetermined amount of the order j; if j j cannot be obtained, α j =1;

The current estimated stock remaining amount r i of the supply node i in Γ(j) is updated according to the following formula (6):
An inquiry method for pushing an information order, wherein when a request for a new order is received from a demanding party, the method includes:

Extract information about each existing order;

Extracting the inventory model according to claim 1;

According to the new order and the orientation of each existing order, the new order and the existing order are determined. Single order of assignment;

In the order of allocation, the following orders are processed in sequence for each existing order before the new order:

Using the inventory model, determining an inventory ratio corresponding to the frequency limit conditions of the existing order corresponding to all supply nodes corresponding to the orientation of the current order; and

Determine the service ratio of the current order and update the current order based on the inventory amount corresponding to each supply node corresponding to the orientation of the current order, the current estimated inventory remaining amount, the determined one inventory share, and the current order subscription amount. Estimated stock remaining amount;

Using the inventory model, determining an inventory ratio corresponding to the frequency limit of the new order corresponding to all the supply nodes corresponding to the orientation of the new order;

The initial value of the reservation amount upper limit is determined according to the inventory quantity corresponding to each supply node corresponding to the orientation of the new order, the current estimated inventory remaining quantity, and the one inventory ratio, and the initial value of the reservation quantity lower limit is set as a preset. Numerical value

In the order of allocation, the following processing is performed on each order in order from the new order until the lower limit of the reservation amount is equal to the upper limit of the reservation amount:

Using the inventory model to determine an inventory share of all supply nodes corresponding to the current order's orientation that meets the frequency limit of the current order;

Attempt to determine the service ratio of the current order and update the current pre-determination according to the inventory quantity corresponding to the orientation of the current order, the current estimated inventory remaining quantity, the one inventory share ratio, and the current bookable quantity detection value. The estimated stock remaining amount, which is raised according to whether the service ratio of the current order can be determined Or adjusting the value of the upper limit of the reservation amount, and adjusting the bookable quantity detection value according to the lower limit of the reservation amount and the upper limit of the reservation amount;

Wherein the predetermined amount of the reservation amount upper limit is used as an initial value of the recordable amount detection value;

The final value of the upper limit of the reservation amount is taken as the maximum amount of the bookable amount of the new order.
The method of claim 15 wherein said determining the order in which the new order and the existing order are assigned comprises:

The inventory quantity of the supply node of each order corresponding to its orientation is determined according to the inventory model, and each order is sorted according to the inventory quantity to determine an allocation order.
The method according to claim 15, wherein the set of inventory ratios corresponding to each of the supply nodes described by the inventory model comprises: an inventory ratio corresponding to each frequency limit under each remaining inventory ratio;

The determined inventory ratio corresponding to the frequency limit of the current order corresponding to all the supply nodes corresponding to the orientation of the current order includes: all the supply nodes in the inventory model that match the orientation of the current order j, at the current The estimated remaining stock ratio p is in line with the inventory ratio of the frequency limit of the order j
Where Γ(j) is a set of supply nodes corresponding to the orientation of the order j, the frequency limit is expressed as (n j , m j ), n j is a time parameter, and m j is a number of times.
The method of claim 17, wherein the service ratio α j of the current order j is determined in the following manner:

Calculate the remaining stock ratio p by the following formula (7):

Where r i is the current estimated inventory remaining amount of the supply node i, and s i is the inventory amount of the supply node i extracted from the inventory model;

Extracting, from the inventory model, the inventory ratio of all supply nodes that match the orientation of the order j, which meets the frequency limit of the order j under the calculated remaining inventory ratio p

Calculate the service ratio α j of the order j according to the following formula (8):

Where d j is the booking amount of the order j; if α j cannot be obtained, α j =1;

The estimated stock remaining amount r i of the supply node i in Γ(j) is updated according to the following formula (9):

Wherein, the upper limit of the reservation amount d upper is determined by the following formula (10):

Wherein, before performing the processing on each order in order from the new order, the current estimated remaining inventory quantity r i ' of all supply nodes i in the inventory model is saved, so r i '=r i ; booking quantity sensing value of the initial value d d x x = d upper;

When the processing is executed for each order in order from the new order, for the current order j, if it is impossible to determine α j that satisfies the formula (8), the estimated stock remaining amount of the supply node i in Γ(j) is not updated. i , let d upper =d x , let
And restoring the current estimated remaining inventory r i of all supply nodes i in the inventory model to the saved r i '; otherwise, updating the estimated inventory remaining amount r i of the supply node i in Γ (j); If it is possible to determine α j that satisfies the formula (8) for the new order and all subsequent orders j, then the lower limit of the reservation amount d lower = d x , and then
And recalculating the current estimated remaining inventory r i of all supply nodes i in the inventory model to the saved r i '.
An information processing device for pushing an information order, comprising:

One or more memories;

One or more processors; among them,

The one or more memories storing one or more instruction modules configured to be executed by the one or more processors; wherein

The one or more instruction modules include:

a data extraction module, which acquires historical user data; wherein the historical user data includes exposure data that has occurred by each user;

The calculating module determines, according to the historical user data, a corresponding exposure quantity for each predetermined dimension or combination of dimensions, and determines a proportion of the exposure quantity, wherein each exposure quantity corresponds to a frequency;

a model building module, based on each dimension or combination of dimensions and its corresponding number of exposures and the proportion of the set of exposures, establishing an inventory model, the inventory model describing each supply node and its corresponding inventory amount and Group inventory ratio;

Wherein, one supply node corresponds to one dimension or one dimension combination, and the inventory quantity corresponding to the supply node is the exposure quantity corresponding to the dimension or the dimension combination, and the corresponding group of inventory proportion of the supply node includes the dimension or the dimension The ratio of the set of exposures corresponding to the combination is proportioned.
The apparatus according to claim 19, wherein the calculation module determines, for each dimension or combination of dimensions, an exposure frequency of each user corresponding to the dimension or combination of dimensions, and determines an exposure quantity for each determined frequency of each exposure. The ratio is taken to obtain the proportion of a set of exposures corresponding to the dimension or combination of dimensions.
The device according to claim 20, wherein the calculation module determines each user and its page access frequency corresponding to the dimension or dimension combination within a specified time period; and determines the dimension or dimension combination corresponding to the specified time period. The total number of page visits; for each page access frequency, determining the first ratio of the number of users corresponding to the page access frequency to the total number of page visits, and using the first ratio to calculate the number of page visits that satisfy the page access frequency limit The second proportion of the total number of page visits, and the second ratio is taken as the exposure amount.
The apparatus according to claim 20, wherein said calculating module determines exposure data of each user in the user set corresponding to the dimension or dimension combination; and estimates the remaining inventory ratio for each user's exposure data The number of exposures consumed is updated, and the exposure data of the user is updated according to the estimated number of exposures consumed; and the exposure frequency of each user is determined according to the updated exposure data of each user; for each exposure frequency, And determining, according to the updated exposure data of each user, an exposure quantity ratio corresponding to the exposure frequency under each remaining inventory ratio.
The apparatus of claim 22, further comprising:

a clustering module, extracting a plurality of orientations of the historical order from the historical user data; performing clustering calculation on the plurality of orientations to obtain at least one orientation group; An orientation group is combined as the predetermined dimension and/or the dimension.
A service distribution device for pushing information orders, comprising:

One or more memories;

One or more processors; among them,

The one or more memories storing one or more instruction modules configured to be executed by the one or more processors; wherein

The one or more instruction modules include:

a data extraction module, extracting information of each order, and extracting the inventory model according to claim 1;

An allocation plan module generates an allocation plan based on the information of each order and the inventory model, the allocation plan including a service ratio of each order, wherein the service ratio of each order is determined as follows:

Using the inventory model, determining an inventory ratio corresponding to the frequency limit of the order corresponding to all supply nodes corresponding to the orientation of the order; and

Determining the service ratio of the order according to the inventory amount corresponding to each supply node corresponding to the orientation of the order, the current estimated inventory remaining amount, the determined one inventory share ratio, and the reservation amount of the order;

And an output module, configured to provide the allocation plan to the push server, so that the push server performs a menu process according to the service ratio in the allocation plan.
The apparatus according to claim 24, wherein said distribution planning module further determines, according to said inventory model, an inventory amount of a supply node of each order corresponding to its orientation, and according to said inventory, when said distribution plan is generated A quantity is sorted to determine an order of allocation; wherein, in the order of the allocation, the services of each order are sequentially determined ratio.
An interrogation device for pushing an information order, wherein when receiving a request for a new order from a demanding party, the device comprises:

One or more memories;

One or more processors; among them,

The one or more memories storing one or more instruction modules configured to be executed by the one or more processors; wherein

The one or more instruction modules include:

a data extraction module, extracting information of each existing order, and extracting the inventory model according to claim 1;

The distribution planning module determines the order of allocation of the new order and the existing orders according to the new order and the orientation of each existing order; in the order of allocation, the existing orders before the new order are sequentially processed as follows:

Using the inventory model, determining an inventory ratio corresponding to the frequency limit of the existing order corresponding to all supply nodes corresponding to the orientation of the current order; and corresponding inventory corresponding to each supply node corresponding to the orientation of the current order Quantity, the current estimated stock remaining amount, the determined one inventory share, and the current order booking amount, determining the service ratio of the current order and updating the current estimated stock remaining amount;

Querying a boundary module, using the inventory model, determining an inventory ratio corresponding to a frequency limit of the new order corresponding to all supply nodes corresponding to the orientation of the new order; each supply according to the orientation of the new order Inventory corresponding to the node, current estimate The remaining quantity of the stock and the one inventory ratio, determining the initial value of the reservation quantity upper limit, and setting the initial value of the reservation quantity lower limit to a preset value;

The reservation quantity detecting module performs the following processing on each order in order from the new order in the order of allocation until the lower limit of the reservation quantity is equal to the upper limit of the reservation quantity:

Using the inventory model to determine an inventory share of all supply nodes corresponding to the current order's orientation that meets the frequency limit of the current order;

Attempt to determine the service ratio of the current order and update the current pre-determination according to the inventory quantity corresponding to the orientation of the current order, the current estimated inventory remaining quantity, the one inventory share ratio, and the current bookable quantity detection value. Estimating the remaining amount of the inventory, adjusting the lower limit of the reservation amount or lowering the value of the upper limit of the reservation amount according to whether the service ratio of the current order can be determined, and adjusting the quantity according to the lower limit of the reservation quantity and the upper limit of the reservation quantity a reservation amount detection value; wherein the reservation amount upper limit initial value is used as an initial value of the bookable amount detection value;

The reservation quantity detecting module uses the final value of the reservation amount upper limit as the maximum amount of the new order.
A non-transitory computer readable storage medium storing computer readable instructions for causing at least one processor to perform the method of any of claims 1-18.