WO2019061903A1

WO2019061903A1 - Information output method and device

Info

Publication number: WO2019061903A1
Application number: PCT/CN2017/118690
Authority: WO
Inventors: 饶佳佳; 徐明泉; 黄绍建; 咸珂; 陈进清; 杨秋源; 裴松年; 涂丽佳
Original assignee: 北京小度信息科技有限公司
Priority date: 2017-09-26
Filing date: 2017-12-26
Publication date: 2019-04-04
Also published as: CN107845016A; CN107845016B

Abstract

An information output method and device for improving estimate accuracy of order output duration. The method comprises: obtaining order data of a current order (201); extracting a feature vector of the current order from the order data of the current order (202), wherein the feature vector of the current order is used for describing the features of the current order; inputting the feature vector of the current order to a pre-trained order output duration estimate model to obtain the order output duration of the current order (203), wherein the order output duration estimate model is used for representing a correspondence of the feature vector and the order output duration; and outputting the order output duration of the current order (204).

Description

Information output method and device

Cross-reference to related applications

The present disclosure claims the priority of the Chinese Patent Application No. Hei. No. Hei. No. Hei.

Technical field

The present disclosure relates to the field of computer technologies, and in particular, to the field of Internet technologies, and in particular, to an information output method and apparatus.

Background technique

With the rapid development of the Internet and the mobile Internet, e-commerce applications have also developed rapidly. In particular, the take-away website, which changes the mode of traditional telephone order take-out service, can provide free, convenient, fast and independent information to help users find the right take-away service. Reasonable planning of the delivery route can improve the efficiency of order delivery, thereby increasing the delivery punctuality rate of the take-out order and helping to improve the user experience.

Estimating the order delivery time helps the delivery personnel to reasonably plan the delivery route, thereby improving order delivery efficiency.

However, the existing billing time estimation method usually takes the average billing time of the merchant as the billing time of all the orders of the merchant, and does not consider the difference between different orders, resulting in the estimated accuracy of the billing time. Lower.

Summary of the invention

It is an object of embodiments of the present disclosure to provide an improved information output method and apparatus to solve the technical problems mentioned in the background section above.

In a first aspect, an embodiment of the present disclosure provides an A1, an information output method, the method includes: acquiring order data of a current order; extracting a feature vector of the current order from the order data of the current order, where the feature vector of the current order Used to describe the characteristics of the current order; input the feature vector of the current order into the pre-trained billing time estimation model to obtain the billing duration of the current order, wherein the billing time prediction model is used to characterize the feature vector and the billing The corresponding relationship of the duration; output the time of the current order.

A2. The method according to A1, the method further comprising the step of training a single-time length estimation model, and the step of training the single-time length estimation model comprises: obtaining a feature vector of the sample order and a billing time of the sample order; The feature vector of the order is used as an input, and the billing time of the sample order is taken as an output, and the single time duration estimation model is trained.

A3. The method according to A2, taking the feature vector of the sample order as an input, and outputting the billing time of the sample order as an output, training to obtain a billing time estimation model, comprising: performing the following training step: eigenvector of the sample order Input to the deep neural network to obtain the estimated billing time of the sample order, and use the estimated billing time of the sample order and the billing time of the sample order to determine the prediction accuracy of the deep neural network, if the estimated accuracy is greater than the pre-predicted When the accuracy rate threshold is set, the deep neural network is used as the billing time prediction model; in response to determining that the estimated accuracy is not greater than the preset accuracy threshold, the parameters of the deep neural network are adjusted, and the training step is continued.

A4. The method as described in A2, obtaining the feature vector of the sample order and the billing time of the sample order, comprising: obtaining the order data of the first historical order and the billing time; and extracting the first from the order data of the first historical order. The feature vector of the historical order is used as the feature vector of the sample order, wherein the feature vector of the sample order is used to describe the feature of the first historical order; the billing category of the first historical order is obtained, wherein the billing category includes the first category and the first category In the second category, the first category is used to indicate that the order has not been placed after the delivery resource of the order reaches the merchant to which the order belongs, and the second category is used to indicate that the order has been placed before the delivery resource of the order reaches the merchant to which the order belongs; The billing category of the first historical order is the first category, and the billing time of the first historical order is used as the billing time of the sample order.

A5. The method as described in A4, obtaining the feature vector of the sample order and the billing time of the sample order, further comprising: if the billing category of the first historical order is the second category, acquiring the first corresponding to the first historical order The time of the single bill, the average billing time of the merchant to which the first historical order belongs, and the first time difference, wherein the first time difference is the time when the distribution resource of the first historical order reaches the merchant to which the first historical order belongs and the merchant to which the first historical order belongs The time difference of the time when the first historical order is received; the time of the billing of the sample order is generated based on the first billing time corresponding to the first historical order, the average billing time of the merchant to which the first historical order belongs, and the first time difference.

A6. The method according to A5, generating a billing duration of the sample order based on the first billing time corresponding to the first historical order, the average billing time of the merchant to which the first historical order belongs, and the first time difference, including: obtaining The first weight of the first billing time corresponding to the first historical order, the second weight of the average billing time of the merchant to which the first historical order belongs, and the third weight of the first time difference; based on the first historical order The billing time, the first weight, the average billing time of the merchant to which the first historical order belongs, the second weight, the first time difference, and the third weight, generate the billing time of the sample order.

A7. The method of A6, based on the first billing time length corresponding to the first historical order, the first weight, the average billing time of the merchant to which the first historical order belongs, the second weight, the first time difference, and the third weight And generating a sample order time, comprising: calculating a product of a first billing time corresponding to the first historical order and a first weight, a product of an average billing time of the merchant to which the first historical order belongs, and a second weight The sum of the product of a time difference and a third weight, and the resulting sum is taken as the time of the bill for the sample order.

A8. The method of A5, the first billing time corresponding to the first historical order is obtained by acquiring the billing time of the at least one second historical order of the merchant to which the first historical order belongs; from at least one The billing duration of the second historical order is selected from the billing duration of the second historical order; the average of the billing time of the selected second historical order is calculated, and is the first corresponding to the first historical order. A single time.

A9. The method according to A8, the time length of the preset number of second historical orders is selected from the billing time of the at least one second historical order, including: the duration of the billing of the at least one second historical order is in accordance with the duration Sort the length and length; select the preset duration of the second historical order from the side of the short duration.

In a second aspect, an embodiment of the present disclosure provides a B1, an information output device, where the device includes: an obtaining unit configured to acquire order data of a current order; and an extracting unit configured to extract current from order data of the current order The feature vector of the order, wherein the feature vector of the current order is used to describe the feature of the current order; the estimating unit is configured to input the feature vector of the current order into the pre-trained billing time estimation model to obtain the current order The single time length, wherein the billing time estimation model is used to represent the correspondence between the feature vector and the billing time; the output unit is configured to output the billing duration of the current order.

B2. The device of B1, further comprising a training unit, the training unit comprising: an acquisition subunit configured to acquire a feature vector of the sample order and a billing time of the sample order; and a training subunit configured to use the sample The feature vector of the order is used as an input, and the billing time of the sample order is taken as an output, and the single time duration estimation model is trained.

B3. The device as described in B2, the training subunit comprises: a training module, configured to perform the following training step: input the feature vector of the sample order into the deep neural network, obtain the estimated order duration of the sample order, and use the sample order Estimated delivery time and sample order time, determine the prediction accuracy of the deep neural network. If the prediction accuracy is greater than the preset accuracy threshold, the deep neural network is used as the estimation model of the release duration; The module is configured to adjust parameters of the deep neural network in response to determining that the estimated accuracy is not greater than the preset accuracy threshold, and continue to perform the training step.

B4. The device as described in B2, the obtaining subunit comprises: a first obtaining module configured to acquire order data and a billing time of the first historical order; and an extracting module configured to be used in order data from the first historical order Extracting a feature vector of the first historical order as a feature vector of the sample order, wherein the feature vector of the sample order is used to describe the feature of the first historical order; and the second obtaining module is configured to obtain the billing category of the first historical order, The billing category includes a first category and a second category. The first category is used to indicate that the order has not been placed after the delivery resource of the order reaches the merchant to which the order belongs, and the second category is used to represent the delivery resource arrival order in the order. The first generation module is configured to use the billing time of the first historical order as the billing time of the sample order if the order of the first order is the first category.

B5. The device of claim 4, the obtaining subunit further comprising: a third obtaining module, configured to acquire the first bill corresponding to the first historical order if the billing category of the first historical order is the second category The duration, the average billing time of the merchant to which the first historical order belongs, and the first time difference, wherein the first time difference is that the delivery resource of the first historical order reaches the merchant of the first historical order and the merchant of the first historical order receives the first The time difference of the time of the historical order; the second generating module is configured to generate the sample order based on the first billing time corresponding to the first historical order, the average billing time of the merchant to which the first historical order belongs, and the first time difference Single time.

B6. The device according to B5, the second generating module includes: an obtaining submodule configured to obtain a first weight of the first billing time corresponding to the first historical order, and an average bill of the merchant to which the first historical order belongs a second weight of the duration and a third weight of the first time difference; generating a sub-module configured to use the first billing duration, the first weight, and the average billing time of the merchant to which the first historical order belongs according to the first historical order The second weight, the first time difference, and the third weight generate a billing time for the sample order.

B7. The device according to B6, wherein the generating submodule is further configured to: calculate a product of a first billing time length corresponding to the first historical order and a first weight, and an average billing time and a first billing time of the merchant to which the first historical order belongs The sum of the product of the two weights and the product of the first time difference and the third weight, and the obtained sum is taken as the billing time of the sample order.

B8. The device of claim 5, the third obtaining module includes a first billing time acquisition sub-module, and a first billing time acquisition sub-module configured to: acquire at least one second historical order of the merchant to which the first historical order belongs The time of the billing time is selected; the billing time of the preset number of second historical orders is selected from the billing time of at least one second historical order; and the average value of the billing time of the selected second historical order is calculated and used as The first billing time corresponding to the first historical order.

B9. The device according to B8, wherein the first billing time acquisition submodule is further configured to: sort the time of the billing of the at least one second historical order according to the length and the length; and select the preset from the side of the short duration The number of times the second historical order is issued. In a third aspect, an embodiment of the present disclosure provides C1, a server, the server includes: one or more processors; a storage device for storing one or more programs; and one or more programs when one or more programs are The processor executes such that one or more processors implement the methods as described in any of the implementations of A1 through A9.

In a fourth aspect, an embodiment of the present disclosure provides a computer readable storage medium, where a computer program is stored thereon, and when the computer program is executed by the processor, the method described in any one of the embodiments A1 to A9 is implemented.

The information output method and apparatus provided by the embodiment of the present disclosure first acquires order data of the current order, so as to extract the feature vector of the current order from the order data of the current order; and then input the feature vector of the current order into the billing time estimate. The model, so as to get the current order's billing time; finally, the resulting current order's billing time output. The billing time estimation model, which can characterize the correspondence between the feature vector and the billing time, is used to estimate the single time duration, which improves the estimation accuracy of the billing time.

DRAWINGS

Other features, objects, and advantages of the present disclosure will become more apparent from the detailed description of the accompanying drawings.

1 is an exemplary system architecture diagram to which an embodiment of the present disclosure may be applied;

2 is a flow chart of one embodiment of an information output method in accordance with the present disclosure;

3 is a flow diagram of one embodiment of a method of training a single duration estimation model in accordance with the present disclosure;

4 is a flow chart of still another embodiment of a method of training a single duration estimation model in accordance with the present disclosure;

FIG. 5 is a schematic structural diagram of an embodiment of an information output device according to the present disclosure; FIG.

6 is a block diagram of a computer system suitable for use in implementing an electronic device of an embodiment of the present disclosure.

Detailed ways

The present disclosure will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is understood that the specific embodiments described herein are merely illustrative of the invention, rather than the invention. It is also to be noted that, for the convenience of description, only the parts related to the related invention are shown in the drawings.

It should be noted that the embodiments in the present disclosure and the features in the embodiments may be combined with each other without conflict. The present disclosure will be described in detail below with reference to the drawings and embodiments.

FIG. 1 illustrates an exemplary system architecture 100 in which an embodiment of an information output method or information output device of the present disclosure may be applied.

As shown in FIG. 1, system architecture 100 can include database server 101, network 102, and server 103. The network 102 is used to provide a medium for communication links between the database server 101 and the server 103. Network 102 can include a variety of connection types, such as wired, wireless communication links, fiber optic cables, and the like.

The database server 101 can be a back-end database server for various websites. For example, the database server 101 can be a background server of an e-commerce website for storing order data on the e-commerce website.

The server 103 can provide various services, for example, the server 103 can analyze the order data of the current order acquired from the database server 101, and output the processing result (for example, the billing duration of the current order).

It should be noted that the information output method provided by the embodiment of the present disclosure is generally performed by the server 103. Accordingly, the information output device is generally disposed in the server 103.

It should be understood that the number of database servers, networks, and servers in Figure 1 is merely illustrative. Depending on the implementation needs, you can have any number of database servers, networks, and servers.

With continued reference to FIG. 2, a flow 200 of one embodiment of an information output method in accordance with the present disclosure is illustrated. The information output method includes the following steps:

Step 201: Obtain order data of the current order.

In this embodiment, the electronic device (for example, the server 103 shown in FIG. 1) on which the information output method runs can be obtained from a database server (for example, the database server 101 shown in FIG. 1) by a wired connection or a wireless connection. Order data for the current order on a website (such as an e-commerce website). The current order may be an order placed by the registered user on the preset website currently on the item on the preset website. The order data may include, but is not limited to, at least one of the following: an order identifier, a merchant identifier of the merchant to which the order belongs, a city logo of the city to which the order belongs, a business circle logo of the business circle to which the order belongs, the total price of the articles in the order, and the articles in the order. The information, the category of the business to which the order belongs (such as fast food, buffet, home cooking, etc.), the brand of the merchant to which the order belongs, the time of the order for the order (usually the order data of the order is not included in the order data of the current order), and the order is placed. Category (usually the order size of the order is not included in the order data of the current order) and so on.

Step 202: Extract a feature vector of the current order from the order data of the current order.

In this embodiment, based on the order data of the current order acquired in step 201, the electronic device may extract the feature vector of the current order from the order data of the current order. Among them, the feature vector of the current order can be used to describe the characteristics of the current order.

In general, the feature vector may include, but is not limited to, at least one of the following: a base feature, a statistical feature, a combined feature, a sparse feature, and the like.

The basic feature may include, but is not limited to, at least one of the following: an order period (for example, an 8-point period, an 11-point period, and the like), an item of different categories in the order (usually the same item belongs to the same category), and an order The number of items, the number of orders for which the order belongs to the merchant, and so on.

The statistical characteristics may include, but are not limited to, at least one of: a ratio of the total price of the items in the order to the average total price of the items in the plurality of orders of the merchant to which the order belongs, the number of items of different categories in the order, and the merchant to which the order belongs The ratio of the average number of items in different categories in a plurality of orders, the ratio of the number of items in the order to the average number of items in the plurality of orders of the merchant to which the order belongs, and the items in the order in which the order is not currently placed by the merchant The total price, the number of items in different categories in the order that the order is not currently placed by the order, the quantity of items in the order that the order belongs to the current unsold order, and the average total of the items in the order that the order belongs to the current unsold order The ratio of the price to the average total price of the items in the multiple orders of the merchant to which the order belongs, the average number of items of different categories in the order that the order is not currently placed by the order, and the items of different categories among the multiple orders of the merchant to which the order belongs. The average number of items, the average number of items in the order that the order is not currently placed, and the business to which the order belongs The average ratio of the number of articles of a plurality of orders.

The combined feature may include, but is not limited to, at least one of the following: a combination feature of the merchant identification of the merchant to which the order belongs and a combination time of the time zone to which the order belongs, a combination feature of the time zone of the order and the delivery time of the order.

The sparse feature may be a sparse matrix obtained by encoding the order data, wherein the number of elements having a value of 0 in the sparse matrix is far more than the number of non-zero elements.

Step 203: Input the feature vector of the current order into the pre-trained billing time estimation model to obtain the billing duration of the current order.

In this embodiment, based on the feature vector of the current order extracted in step 202, the electronic device may input the feature vector of the current order into the pre-trained billing time estimation model, thereby obtaining the billing duration of the current order. The timeout period may be the time difference between the time of placing the order and the time when the delivery resource of the order arrives at the merchant to which the order belongs, wherein the distribution resource may include, but is not limited to, a delivery person, a delivery robot, and the like.

In this embodiment, the billing time prediction model can be used to characterize the correspondence between the feature vector and the billing duration. Here, the electronic device can train the single duration estimation model in a variety of ways.

In some optional implementation manners of the embodiment, the electronic device may generate, according to the statistics of the feature vector and the billing duration of the historical order, a correspondence table in which the correspondence between the plurality of feature vectors and the billing duration is stored. The correspondence table is used as an estimation model for the billing time. When the electronic device inputs the feature vector of the current order into the single-time estimation model, the billing time estimation model queries the correspondence table to obtain the billing time corresponding to the feature vector of the current order.

Step 204: Output the billing time of the current order.

In this embodiment, based on the billing time of the current order obtained in step 203, the electronic device can output the billing duration of the current order to other electronic devices (for example, a database server, a terminal device, etc.) that are communicatively connected thereto. As an example, the electronic device may send the billing time of the current order to the dispatcher of the current order or the terminal device of the manager of the delivery robot, so that the delivery person of the current order or the manager of the delivery robot can plan the delivery route.

The information output method provided by the embodiment of the present disclosure first obtains the order data of the current order, so as to extract the feature vector of the current order from the order data of the current order; and then input the feature vector of the current order into the billing time estimation model. In this way, the billing time of the current order is obtained; finally, the obtained billing time of the current order is output. The billing time prediction model which can represent the correspondence between the feature vector and the billing time is used to estimate the single duration, which improves the estimation accuracy of the billing time.

With further reference to FIG. 3, a flow 300 of one embodiment of a method of training a single duration estimation model is illustrated. The process 300 of the method for training a single time prediction model includes the following steps:

Step 301: Obtain a feature vector of the sample order and a billing time of the sample order.

In the present embodiment, the electronic device (for example, the server 103 shown in FIG. 1) on which the method for training the single duration estimation model is run can acquire the feature vector of the sample order and the billing time of the sample order. Here, the electronic device can obtain the feature vector of the sample order and the billing time of the sample order in various ways. As an example, a person skilled in the art can perform statistical analysis on the order data of a large number of orders, thereby obtaining a plurality of feature vectors and a plurality of billing durations, and correspondingly obtaining each of the obtained feature vectors and each billing time. It is then used as the feature vector for the sample order and the billing time for the sample order.

In step 302, the feature vector of the sample order is taken as an input, and the billing time of the sample order is taken as an output, and the single time duration estimation model is trained.

In this embodiment, based on the feature vector of the sample order acquired in step 301 and the billing time of the sample order, the electronic device can utilize the feature vector of the sample order and the billing time of the sample order, for example, DNN (Deep Neural Network, The deep neural network model is trained to obtain an order-time prediction model that can accurately represent the relationship between feature vectors and billing time.

In some optional implementation manners of the embodiment, the electronic device may be trained to obtain a single duration estimation model by the following steps:

First, perform the following training steps: input the feature vector of the sample order into the deep neural network, obtain the estimated billing time of the sample order, and determine the deep neural network by using the estimated billing time of the sample order and the billing time of the sample order. The estimated accuracy rate, if the estimated accuracy is greater than the preset accuracy threshold, the deep neural network is used as the billing time prediction model;

Here, if the estimated billing time of a sample order is the same as or similar to the billing time of the sample order (for example, the difference is less than a preset value), the deep neural network estimates the billing time of the sample order accurately; If the estimated billing time of the sample order is different from or different from the billing time of the sample order (for example, the difference is not less than the preset value), the deep neural network may not accurately predict the billing time of the sample order. In practice, the electronic device can use the ratio of the estimated number of sample orders to the total number of sample orders as the estimated accuracy of the deep neural network.

Then, in response to determining that the estimated accuracy is not greater than the preset accuracy threshold, the parameters of the deep neural network are adjusted and the training step continues.

Here, in the case that the estimated accuracy is not greater than the preset accuracy threshold, the electronic device may adjust the parameters of the deep neural network and return to perform the above training steps until the training of the feature vector capable of characterizing the order and the order time of the order are trained. The accuracy of the correspondence between the billing time and the estimated model.

The method for training a single duration estimation model provided by the embodiment of the present disclosure uses the feature vector of the sample order and the billing time of the sample order to perform training, thereby obtaining an order that can accurately represent the exact correspondence between the feature vector and the billing time. Duration estimation model.

With further reference to FIG. 4, a flow 400 of yet another embodiment of a method of training a single duration estimation model is illustrated. The process 400 of the method for training a single time estimation model includes the following steps:

Step 401: Obtain order data and billing time of the first historical order.

In this embodiment, the electronic device (for example, the server 103 shown in FIG. 1) on which the method for training the single duration estimation model is run may be from a database server by a wired connection or a wireless connection (for example, as shown in FIG. 1). The database server 101) obtains the order data and the billing time of the historical order in the first preset time period (for example, the previous month and the previous week) on the preset website (for example, an e-commerce website), and is the first The order data of the historical order and the billing time are long, and then steps 402 and 403 are performed.

Step 402: Extract a feature vector of the first historical order from the order data of the first historical order as a feature vector of the sample order.

In this embodiment, based on the order data of the first historical order acquired in step 401, the electronic device may extract the feature vector of the first historical order from the order data of the first historical order and use it as a feature vector of the sample order, and then Go to step 407. Wherein, the feature vector of the sample order can be used to describe the characteristics of the first historical order.

In step 403, the billing category of the first historical order is obtained.

In this embodiment, based on the order data of the first historical order acquired in step 401, the electronic device may acquire the billing category of the first historical order.

Usually, the order data of a historical order can include the billing category of the historical order. The billing category may include a first category and a second category. The first category can be used to characterize the situation where the order has not been placed after the delivery resource of the order arrives at the merchant to which the order belongs. The second category can be used to characterize the fact that the order has been placed before the order's delivery resource reaches the merchant to which the order belongs. The distribution resources may include, but are not limited to, a delivery person, a delivery robot, and the like.

Step 404: Determine whether the billing category of the first historical order is the first category or the second category.

In this embodiment, based on the billing category of the first historical order acquired in step 403, the electronic device may determine whether the billing category of the first historical order is the first category or the second category; the billing of the first historical order In the case where the category is the first category, step 405' is performed; in the case where the billing category of the first historical order is the second category, step 405 is performed.

In step 405 ′, the billing time of the first historical order is used as the billing time of the sample order.

In this embodiment, in the case that the billing category of the first historical order is the first category, the electronic device may use the billing duration of the first historical order as the billing duration of the sample order, and then perform step 407.

Step 405: Acquire a first billing time length corresponding to the first historical order, an average billing time length of the merchant to which the first historical order belongs, and a first time difference.

In this embodiment, in the case that the billing category of the first historical order is the second category, the electronic device may obtain the first billing time corresponding to the first historical order and the average billing of the merchant to which the first historical order belongs. Duration and first time difference. The first time difference may be a time difference between a time when the distribution resource of the first historical order reaches the merchant to which the first historical order belongs and a time when the merchant to which the first historical order belongs receives the first historical order. The first billing time corresponding to the first historical order may be obtained by statistical analysis of a large number of historical orders similar to the first historical order or a large number of historical orders of the merchant to which the first historical order belongs.

In some optional implementation manners of the embodiment, the first billing time corresponding to the first historical order may be obtained by the electronic device by the following steps:

First, the billing time of at least one second historical order of the merchant to which the first historical order belongs is obtained.

As an example, the electronic device may obtain the billing time of the historical order of the merchant to which the first historical order belongs in the second preset time period (for example, the previous month, the previous week), and serve as the billing time of the second historical order. .

Then, the billing duration of the preset number of second historical orders is selected from the billing duration of the at least one second historical order.

As an example, the electronic device may randomly select a billing duration of the preset number of second historical orders from the billing time of the at least one second historical order.

As another example, the electronic device may first sort the billing durations of the at least one second historical order according to the length of time; then, select the preset billing duration of the second historical order from the short side. Here, the electronic device may sort the billing durations of the at least one second historical order in order of length to length; or may sort the billing durations of the at least one second historical order in order of duration from short to long. .

Finally, the average value of the billing time of the selected second historical order is calculated, and is used as the first billing time corresponding to the first historical order.

Step 406: Generate a billing duration of the sample order based on the first billing time length corresponding to the first historical order, the average billing time of the merchant to which the first historical order belongs, and the first time difference.

In this embodiment, based on the first billing time length corresponding to the first historical order acquired in step 405, the average billing time of the merchant to which the first historical order belongs, and the first time difference, the electronic device may generate the sample order. Single time. Here, the electronic device can generate the billing time of the sample order in various ways. As an example, the electronic device may calculate the first billing time corresponding to the first historical order, the average billing time of the merchant to which the first historical order belongs, and the average of the first time difference, and serve as the billing time of the sample order, and then execute Step 407.

In some optional implementation manners of the embodiment, the electronic device may first obtain the first weight of the first billing time corresponding to the first historical order, and the second weight of the average billing time of the merchant to which the first historical order belongs. And a third weight that is different from the first time; and then based on the first billing time corresponding to the first historical order, the first weight, the average billing time of the merchant to which the first historical order belongs, the second weight, the first time difference, and the third Weight, the time the bill was generated for the sample order.

As an example, the electronic device may calculate a product of a first billing time and a first weight corresponding to the first historical order, a product of an average billing time and a second weight of the merchant to which the first historical order belongs, and a first time difference and a third The sum of the products of the weights, and the resulting sum is taken as the billing time of the sample order.

Specifically, the electronic device can generate the billing time ΔT of the sample order by the following formula:

ΔT = Δt ₁ × w ₁ + Δt ₂ × w ₂ + Δt ₃ × w ₃ ;

Where Δt ₁ is the first billing time corresponding to the first historical order, Δt ₂ is the average billing time of the merchant to which the first historical order belongs, Δt ₃ is the first time difference, w ₁ is the first weight, w ₂ is The second weight, w ₃ is the third weight.

It should be noted that the first weight, the second weight, and the third weight are preset according to the category of the first historical order, and the first weight, the second weight, and the third weight corresponding to the first historical order of different categories may be different.

In step 407, the feature vector of the sample order is taken as an input, and the billing time of the sample order is taken as an output, and the single time length estimation model is trained.

In this embodiment, based on the feature vector of the sample order obtained in step 402 and the billing time of the sample order obtained in step 405′ or the billing time of the sample order obtained in step 406, the electronic device can utilize the sample order. The feature vector and the order time of the sample order are trained, for example, on the DNN model, thereby obtaining a billing time prediction model capable of characterizing the exact correspondence between the feature vector and the billing time.

As can be seen from FIG. 4, the flow 400 of the method for training a single time prediction model in the present embodiment highlights the steps of obtaining the feature vector of the sample order and the time of placing the order, as compared with the embodiment corresponding to FIG. Therefore, the feature vector and the billing time of the sample order in the solution described in this embodiment are obtained by analyzing and processing the order data and the billing time of the first historical order, and the first history of different billing categories The order time of the order is different, so that the feature vector and the billing time of the sample order obtained are more objective and real.

With further reference to FIG. 5, as an implementation of the method shown in the above figures, the present disclosure provides an embodiment of an information output device, which corresponds to the method embodiment shown in FIG. 2, and the device may specifically Used in a variety of electronic devices.

As shown in FIG. 5, the information output apparatus 500 of the present embodiment may include an acquisition unit 501, an extraction unit 502, an estimation unit 503, and an output unit 504. The obtaining unit 501 is configured to acquire order data of the current order, and the extracting unit 502 is configured to extract a feature vector of the current order from the order data of the current order, where the feature vector of the current order is used to describe the current order. The estimating unit 503 is configured to input the feature vector of the current order into the pre-trained billing time estimation model to obtain the billing duration of the current order, wherein the billing time prediction model is used to represent the feature vector and The corresponding relationship of the billing time; the output unit 504 is configured to output the billing duration of the current order.

In this embodiment, the specific processing of the information obtaining apparatus 501, the extracting unit 502, the estimating unit 503, and the output unit 504 and the technical effects thereof can be referred to the steps in the corresponding embodiment of FIG. 2, respectively. 201, step 202, step 203 and step 204 are not described here.

In some optional implementation manners of the embodiment, the information output device 500 may further include a training unit (not shown), and the training unit may include: an acquisition subunit (not shown) configured to acquire The feature vector of the sample order and the billing time of the sample order; the training subunit (not shown) is configured to take the feature vector of the sample order as an input, and output the billing time of the sample order as an output, and train to obtain a bill. Duration estimation model.

In some optional implementation manners of the embodiment, the training subunit may include: a training module (not shown) configured to perform the following training step: input the feature vector of the sample order into the deep neural network, and obtain The estimated billing time of the sample order, using the estimated billing time of the sample order and the billing time of the sample order, determining the prediction accuracy of the deep neural network. If the estimated accuracy is greater than the preset accuracy threshold, then The deep neural network is used as a billing time estimation model; the adjustment module (not shown) is configured to adjust the parameters of the deep neural network in response to determining that the estimated accuracy is not greater than the preset accuracy threshold, and continue to perform the training. step.

In some optional implementation manners of the embodiment, the obtaining subunit may include: a first acquiring module (not shown) configured to acquire order data and billing time of the first historical order; and an extracting module ( Not shown in the figure), configured to extract a feature vector of the first historical order from the order data of the first historical order as a feature vector of the sample order, wherein the feature vector of the sample order is used to describe the characteristics of the first historical order a second acquisition module (not shown) configured to obtain an order category of the first historical order, wherein the billing category includes a first category and a second category, and the first category is used to characterize the delivery in the order The second category is used to indicate the situation that the order has been placed before the delivery resource of the order arrives at the merchant to which the order belongs, and the first generation module (not shown) is configured. If the billing category of the first historical order is the first category, the billing time of the first historical order is used as the billing time of the sample order.

In some optional implementation manners of the embodiment, the obtaining sub-unit may further include: a third obtaining module (not shown) configured to: if the billing category of the first historical order is the second category, Obtaining a first billing time corresponding to the first historical order, an average billing time of the merchant to which the first historical order belongs, and a first time difference, wherein the first time difference is that the distribution resource of the first historical order reaches the merchant of the first historical order And a time difference between the time when the first historical order is received by the merchant to which the first historical order belongs; the second generation module (not shown) is configured to be based on the first billing time corresponding to the first historical order, first The average billing time and the first time difference of the merchant to which the historical order belongs, and the billing time of the sample order is generated.

In some optional implementation manners of the embodiment, the second generating module may include: an acquiring submodule (not shown) configured to obtain the first time of the first billing time corresponding to the first historical order. The weight, the second weight of the average billing time of the merchant to which the first historical order belongs, and the third weight of the first time difference; generating a sub-module (not shown) configured to be based on the first corresponding to the first historical order The billing time, the first weight, the average billing time of the merchant to which the first historical order belongs, the second weight, the first time difference, and the third weight, generate the billing time of the sample order.

In some optional implementation manners of the embodiment, the generating submodule may be further configured to: calculate a product of a first billing time length corresponding to the first historical order and a first weight, and an average of the merchants to which the first historical order belongs The sum of the product of the billing time and the second weight and the product of the first time difference and the third weight, and the obtained sum is taken as the billing time of the sample order.

In some optional implementation manners of the embodiment, the third obtaining module may include a first billing time acquisition submodule (not shown), and the first billing time acquisition submodule is configured to: obtain the first a billing time of at least one second historical order of the merchant to which the historical order belongs; selecting a billing duration of the preset number of second historical orders from the billing duration of the at least one second historical order; calculating the selected second taken out The average of the billing time of the historical order, and the first billing time corresponding to the first historical order.

In some optional implementation manners of the embodiment, the first billing time acquisition sub-module may be further configured to: sort the billing durations of the at least one second historical order according to the length of time; the side from the short duration Start to select the preset duration of the second historical order.

Referring now to Figure 6, a block diagram of a computer system 600 suitable for use in implementing an electronic device of an embodiment of the present disclosure is shown. The electronic device shown in FIG. 6 is merely an example, and should not impose any limitation on the function and scope of use of the embodiments of the present disclosure.

As shown in FIG. 6, computer system 600 includes a central processing unit (CPU) 601 that can be loaded into a program in random access memory (RAM) 603 according to a program stored in read only memory (ROM) 602 or from storage portion 608. And perform various appropriate actions and processes. In the RAM 603, various programs and data required for the operation of the system 600 are also stored. The CPU 601, the ROM 602, and the RAM 603 are connected to each other through a bus 604. An input/output (I/O) interface 605 is also coupled to bus 604.

The following components are connected to the I/O interface 605: an input portion 606 including a keyboard, a mouse, etc.; an output portion 607 including, for example, a cathode ray tube (CRT), a liquid crystal display (LCD), and the like, and a storage portion 608 including a hard disk or the like. And a communication portion 609 including a network interface card such as a LAN card, a modem, or the like. The communication section 609 performs communication processing via a network such as the Internet. Driver 610 is also coupled to I/O interface 605 as needed. A removable medium 611, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory or the like, is mounted on the drive 610 as needed so that a computer program read therefrom is installed into the storage portion 608 as needed.

In particular, the processes described above with reference to the flowcharts may be implemented as a computer software program in accordance with an embodiment of the present disclosure. For example, an embodiment of the present disclosure includes a computer program product comprising a computer program embodied on a computer readable medium, the computer program comprising program code for executing the method illustrated in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network via communication portion 609, and/or installed from removable media 611. When the computer program is executed by the central processing unit (CPU) 601, the above-described functions defined in the method of the present disclosure are performed.

It should be noted that the above computer readable medium of the present disclosure may be a computer readable signal medium or a computer readable storage medium or any combination of the two. The computer readable storage medium can be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples of computer readable storage media may include, but are not limited to, electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read only memory (ROM), erasable Programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination of the foregoing. In the present disclosure, a computer readable storage medium may be any tangible medium that can contain or store a program, which can be used by or in connection with an instruction execution system, apparatus, or device. While in the present disclosure, a computer readable signal medium may include a data signal that is propagated in the baseband or as part of a carrier, carrying computer readable program code. Such propagated data signals can take a variety of forms including, but not limited to, electromagnetic signals, optical signals, or any suitable combination of the foregoing. The computer readable signal medium can also be any computer readable medium other than a computer readable storage medium, which can transmit, propagate, or transport a program for use by or in connection with the instruction execution system, apparatus, or device. . Program code embodied on a computer readable medium can be transmitted by any suitable medium, including but not limited to wireless, wire, fiber optic cable, RF, etc., or any suitable combination of the foregoing.

The flowchart and block diagrams in the Figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products in accordance with various embodiments of the present disclosure. In this regard, each block of the flowchart or block diagram can represent a module, a program segment, or a portion of code that includes one or more of the logic functions for implementing the specified. Executable instructions. It should also be noted that in some alternative implementations, the functions noted in the blocks may also occur in a different order than that illustrated in the drawings. For example, two successively represented blocks may in fact be executed substantially in parallel, and they may sometimes be executed in the reverse order, depending upon the functionality involved. It is also noted that each block of the block diagrams and/or flowcharts, and combinations of blocks in the block diagrams and/or flowcharts, can be implemented in a dedicated hardware-based system that performs the specified function or operation. Or it can be implemented by a combination of dedicated hardware and computer instructions.

The units described in the embodiments of the present disclosure may be implemented by software or by hardware. The described unit may also be provided in the processor, for example, as a processor including an acquisition unit, an extraction unit, an estimation unit, and an output unit. The names of these units do not constitute a limitation on the unit itself under certain circumstances. For example, the acquisition unit may also be described as “a unit that acquires order data of the current order”.

In another aspect, the present disclosure further provides a computer readable medium, which may be included in an electronic device described in the above embodiments, or may be separately present without being assembled into the electronic device. in. The computer readable medium carries one or more programs, when the one or more programs are executed by the electronic device, causing the electronic device to: obtain order data of the current order; and extract the current order from the order data of the current order. a feature vector, wherein the feature vector of the current order is used to describe the feature of the current order; the feature vector of the current order is input to the pre-trained billing time estimation model to obtain the billing time of the current order, wherein the billing time is The estimation model is used to characterize the correspondence between the feature vector and the billing time; output the billing duration of the current order.

The above description is only a preferred embodiment of the present disclosure and a description of the principles of the applied technology. It should be understood by those skilled in the art that the scope of the invention referred to in the present disclosure is not limited to the specific combination of the above technical features, and should also be covered by the above technical features or without departing from the above inventive concept. Other technical solutions formed by arbitrarily combining the equivalent features. For example, the above features are combined with technical features disclosed in the present disclosure, but not limited to, technical features having similar functions.

Claims

An information output method, wherein the method comprises:

Get the order data of the current order;

Extracting a feature vector of the current order from the order data of the current order, wherein a feature vector of the current order is used to describe a feature of the current order;

Entering a feature vector of the current order into a pre-trained billing time length estimation model to obtain a billing duration of the current order, wherein the billing time estimation model is used to represent the feature vector and the billing time Correspondence relationship

Output the billing time of the current order.
The method of claim 1 wherein said method further comprises the step of training a single duration estimation model, said step of training a single duration prediction model comprising:

Obtain the feature vector of the sample order and the billing time of the sample order;

Taking the feature vector of the sample order as an input, taking the billing time of the sample order as an output, training to obtain a billing time prediction model.
The method according to claim 2, wherein the taking the feature vector of the sample order as an input and outputting the billing time of the sample order as an output, training to obtain a billing time prediction model, comprising:

Performing the following training steps: inputting the feature vector of the sample order to the deep neural network, obtaining an estimated billing time of the sample order, using the estimated billing time of the sample order, and the billing of the sample order Determining an estimated accuracy rate of the deep neural network, and if the estimated accuracy is greater than a preset accuracy threshold, using the deep neural network as the billing time prediction model;

In response to determining that the estimated accuracy is not greater than a preset accuracy threshold, adjusting parameters of the deep neural network and continuing to perform the training step.
The method of claim 2, wherein the obtaining the feature vector of the sample order and the billing time of the sample order comprises:

Obtain order data and billing time for the first historical order;

Extracting, from the order data of the first historical order, a feature vector of the first historical order as a feature vector of a sample order, wherein a feature vector of the sample order is used to describe a feature of the first historical order;

Acquiring the billing category of the first historical order, wherein the billing category includes a first category and a second category, where the first category is used to represent that the order is not issued after the delivery resource of the order arrives at the merchant to which the order belongs The second category is used to characterize the fact that the order has been placed before the delivery resource of the order arrives at the merchant to which the order belongs;

If the billing category of the first historical order is the first category, the billing duration of the first historical order is used as the billing duration of the sample order.
The method of claim 4, wherein the obtaining the feature vector of the sample order and the billing time of the sample order further comprises:

If the billing category of the first historical order is the second category, acquiring the first billing time corresponding to the first historical order, the average billing time of the merchant to which the first historical order belongs, and the a time difference, wherein the first time difference is a time when the delivery resource of the first historical order reaches the merchant to which the first historical order belongs and a time when the merchant to which the first historical order belongs receives the first historical order Time difference;

The billing duration of the sample order is generated based on the first billing time length corresponding to the first historical order, the average billing time length of the merchant to which the first historical order belongs, and the first time difference.
The method according to claim 5, wherein the generating, based on the first billing time corresponding to the first historical order, the average billing time of the merchant to which the first historical order belongs, and the first time difference, are generated. The billing time for the sample order, including:

Obtaining a first weight of the first billing time corresponding to the first historical order, a second weight of the average billing time of the merchant to which the first historical order belongs, and a third weight of the first time difference;

a first billing time length corresponding to the first historical order, the first weight, an average billing time of the merchant to which the first historical order belongs, the second weight, the first time difference, and the The third weight is used to generate the billing time of the sample order.
The method according to claim 6, wherein the first billing time length corresponding to the first historical order, the first weight, the average billing time of the merchant to which the first historical order belongs, The second weight, the first time difference, and the third weight are generated, and the time for generating the sample order is generated, including:

Calculating a product of a first billing time length corresponding to the first historical order and the first weight, a product of an average billing time of the merchant to which the first historical order belongs, and the second weight, and the first The sum of the time difference and the product of the third weight, and the resulting sum is taken as the billing time of the sample order.
The method according to claim 5, wherein the first billing time corresponding to the first historical order is obtained by the following steps:

Obtaining a billing time of at least one second historical order of the merchant to which the first historical order belongs;

Selecting, according to the billing duration of the at least one second historical order, a billing duration of the preset number of second historical orders;

Calculating an average value of the billing time of the selected second historical order, and as the first billing time corresponding to the first historical order.
The method of claim 8, wherein the selecting a billing duration of the preset number of second historical orders from the billing duration of the at least one second historical order comprises:

Sorting the time of the billing of the at least one second historical order according to the length of time;

The time of the order for the preset number of second historical orders is selected from the side of the short duration.
An information output device, wherein the device comprises:

Obtaining a unit configured to obtain order data of a current order;

An extracting unit configured to extract a feature vector of the current order from the order data of the current order, wherein a feature vector of the current order is used to describe a feature of the current order;

An estimating unit, configured to input a feature vector of the current order into a pre-trained billing time estimation model, to obtain a billing duration of the current order, wherein the billing time prediction model is used for characterization Correspondence between feature vectors and billing time;

And an output unit configured to output a billing duration of the current order.
The apparatus of claim 10, wherein the apparatus further comprises a training unit, the training unit comprising:

Obtaining a subunit configured to obtain a feature vector of the sample order and a billing time of the sample order;

The training subunit is configured to take the feature vector of the sample order as an input, and output the billing time of the sample order as an output, and obtain a single duration estimation model.
The apparatus of claim 11 wherein said training subunit comprises:

a training module, configured to perform the following training step: inputting a feature vector of the sample order to a deep neural network, obtaining an estimated billing time of the sample order, and using the estimated billing time and the location of the sample order Determining the estimated time of the sample order, and determining the prediction accuracy of the deep neural network. If the estimated accuracy is greater than the preset accuracy threshold, using the deep neural network as the prediction model ;

The adjusting module is configured to adjust parameters of the deep neural network in response to determining that the estimated accuracy is not greater than a preset accuracy threshold, and continue to perform the training step.
The apparatus of claim 11 wherein said obtaining subunit comprises:

a first obtaining module, configured to acquire order data and a billing time of the first historical order;

An extraction module, configured to extract a feature vector of the first historical order from an order data of the first historical order as a feature vector of a sample order, wherein a feature vector of the sample order is used to describe the first Characteristics of historical orders;

a second obtaining module, configured to acquire an order type of the first historical order, where the billing category includes a first category and a second category, where the first category is used to represent the delivery resource arrival order of the order In the case that the order is not issued after the merchant, the second category is used to indicate that the order has been placed before the delivery resource of the order reaches the merchant to which the order belongs;

The first generating module is configured to use, when the billing category of the first historical order is the first category, the billing duration of the first historical order as the billing time of the sample order.
The apparatus of claim 13, wherein the obtaining subunit further comprises:

a third obtaining module, configured to acquire, when the billing category of the first historical order is the second category, the first billing time corresponding to the first historical order, and the first historical order The average time of the merchant and the first time difference, wherein the first time difference is the time when the distribution resource of the first historical order reaches the merchant to which the first historical order belongs and the merchant receiving the first historical order The time difference between the time of the first historical order;

a second generating module, configured to generate a sample order by based on a first billing duration corresponding to the first historical order, an average billing time of the merchant to which the first historical order belongs, and the first time difference duration.
The apparatus of claim 14, wherein the second generation module comprises:

Obtaining a sub-module, configured to acquire a first weight of the first billing time corresponding to the first historical order, a second weight of the average billing time of the merchant to which the first historical order belongs, and the first time difference Third weight;

Generating a sub-module, configured to be based on the first billing time length corresponding to the first historical order, the first weight, the average billing time of the merchant to which the first historical order belongs, the second weight, and the The first time difference and the third weight are used to generate a billing time for the sample order.
The apparatus of claim 15 wherein said generating sub-module is further configured to:

Calculating a product of a first billing time length corresponding to the first historical order and the first weight, a product of an average billing time of the merchant to which the first historical order belongs, and the second weight, and the first The sum of the time difference and the product of the third weight, and the resulting sum is taken as the billing time of the sample order.
The device according to claim 14, wherein the third obtaining module comprises a first billing time acquisition sub-module, and the first billing time acquisition sub-module is configured to:

Obtaining a billing time of at least one second historical order of the merchant to which the first historical order belongs;

Selecting, according to the billing duration of the at least one second historical order, a billing duration of the preset number of second historical orders;

Calculating an average value of the billing time of the selected second historical order, and as the first billing time corresponding to the first historical order.
The apparatus according to claim 17, wherein the first billing time acquisition submodule is further configured to:

Sorting the time of the billing of the at least one second historical order according to the length of time;

The time of the order for the preset number of second historical orders is selected from the side of the short duration.
An electronic device, wherein the electronic device comprises:

One or more processors;

a storage device for storing one or more programs;

The one or more programs are executed by the one or more processors such that the one or more processors implement the method of any of claims 1-9.
A computer readable storage medium having stored thereon a computer program, wherein the computer program is executed by a processor to implement the method of any of claims 1-9.