A method and apparatus for predicting the arrival time of delivery resources at merchants.

Abstract

This application provides a method and apparatus for predicting the arrival time of delivery resources at merchants. In the method, a first delivery object to be picked up and the corresponding first merchant information are determined. Then, the second merchant information corresponding to a second delivery object that has already been picked up is obtained. A merchant target location vector model is used to obtain the first merchant target location vector corresponding to the first merchant and the second merchant target location vector corresponding to the second merchant. Here, the merchant target location vector is calculated based on the initial location vector corresponding to each layer of the multi-layer graph structure. The multi-layer graph structure includes a near-field wireless node data layer graph structure and a historical behavior data layer graph structure of delivery resources. This multi-layer graph structure improves the accuracy of the merchant target location vector and also improves the prediction accuracy of the delivery resource's arrival at the merchant's location.

Description

Technical Field

[0001] This application relates to the field of computer technology, specifically to a method for predicting the arrival time of delivery resources at merchants. This application also relates to an apparatus, electronic device, and computer storage medium for predicting the arrival time of delivery resources at merchants. This application further relates to a method, apparatus, electronic device, and computer storage medium for predicting the target location of a merchant. This application also relates to a method, apparatus, electronic device, and computer storage medium for constructing a multi-layered graph structure. This application further relates to a method, apparatus, electronic device, and computer storage medium for predicting the arrival of delivery resources at a target merchant. This application also relates to a method, apparatus, electronic device, and computer storage medium for training a merchant target location vector model. Background Technology

[0002] Online service platforms have brought many conveniences to users' lives, allowing them to shop for goods, order food, and so on.

[0003] In existing technology, users order food through online food delivery platforms, merchants prepare the food according to the order, delivery resources arrive at the merchant to pick up the food, and then deliver the food to the user's designated delivery address. In this process, delivery delays frequently occur, but the reasons for these delays are often unclear. Therefore, predicting the arrival time of delivery resources at the merchant is a problem that needs to be solved. Summary of the Invention

[0004] This application provides a method for predicting the arrival time of delivery resources at merchants, addressing a problem that needs to be solved. This application also provides an apparatus, electronic device, and computer storage medium for predicting the arrival time of delivery resources at merchants. This application further relates to a method, apparatus, electronic device, and computer storage medium for predicting the target location of a merchant. This application also relates to a method, apparatus, electronic device, and computer storage medium for constructing a multi-layered geographic map structure. This application further provides a method, apparatus, electronic device, and computer storage medium for predicting the arrival of delivery resources at a target merchant. This application also provides a method, apparatus, electronic device, and computer storage medium for training a merchant target location vector model.

[0005] This application provides a method for predicting the arrival time of delivery resources at a merchant, comprising: obtaining information about a first merchant to which a first delivery object to be picked up belongs; inputting the first merchant information into a merchant target location vector model to obtain a first merchant target location vector output by the merchant target location vector model, wherein the merchant target location vector model is used to obtain the first merchant target location vector based on the initial location vector corresponding to each layer of the graph structure in the multi-layer graph structure, the multi-layer graph structure including a near-field wireless node data layer graph structure and a historical behavior data layer graph structure of delivery resources; obtaining information about a second merchant to which a second delivery object to which the delivery resource has been picked up belongs, and obtaining a second merchant target location vector corresponding to the second merchant information; inputting the second merchant target location vector and the first merchant target location vector as input data into a duration prediction network model to obtain first predicted duration information output by the duration prediction network model, representing the arrival time of the delivery resource from the location of the second merchant to the location of the first merchant.

[0006] Optionally, the step of inputting the first merchant information into the merchant target location vector model to obtain the first merchant target location vector corresponding to the first merchant information output by the merchant target location vector model includes: obtaining, based on the first merchant information, a first initial location vector of the first merchant corresponding to the first merchant information in the near-field wireless node data layer graph structure; obtaining, based on the first merchant information, a second initial location vector of the first merchant corresponding to the first merchant information in the historical behavior data layer graph structure of the delivery resources; calculating a first loss value between the first initial location vector and the second initial location vector; and obtaining the first merchant target location vector based on the first initial location vector, the second initial location vector, and the first loss value.

[0007] Optionally, obtaining the first merchant's target location vector according to a preset calculation method based on the first initial location vector, the second initial location vector, and the first loss value includes: obtaining a first adjusted location vector of the first merchant in the near-field wireless node data layer map structure based on the first initial location vector and the first loss value; obtaining a second adjusted location vector of the first merchant in the historical behavior data layer map structure of delivery resources based on the second initial location vector and the first loss value; and obtaining the first merchant's target location vector based on the first adjusted location vector and the second adjusted location vector.

[0008] Optionally, the multi-layer graph structure further includes a wireless local area network (WLAN) data layer graph structure; the step of inputting the first merchant information into the merchant target location vector model to obtain the first merchant target location vector corresponding to the first merchant information output by the merchant target location vector model includes: obtaining, based on the first merchant information, a first initial location vector of the first merchant corresponding to the first merchant information in the near-field wireless node data layer graph structure; obtaining, based on the first merchant information, a second initial location vector of the first merchant corresponding to the first merchant information in the historical behavior data layer graph structure of the delivery resources; obtaining, based on the first merchant information, a third initial location vector of the first merchant corresponding to the first merchant information in the WLAN data layer graph structure; calculating, based on the first initial location vector, the second initial location vector, and the third initial location vector, a second loss value is calculated; and based on the first initial location vector, the second initial location vector, the third initial location vector, and the second loss value, the first merchant target location vector is obtained.

[0009] Optionally, obtaining the first merchant's target location vector based on the first initial location vector, the second initial location vector, the third initial location vector, and the second loss value includes: obtaining a third adjusted location vector of the first merchant in the near-field wireless node data layer graph structure based on the first initial location vector and the second loss value; obtaining a fourth adjusted location vector of the first merchant in the historical behavior data layer graph structure of delivery resources based on the second initial location vector and the second loss value; obtaining a fifth adjusted location vector of the first merchant in the historical behavior data layer graph structure of delivery resources based on the third initial location vector and the second loss value; and obtaining the first merchant's target location vector based on the third adjusted location vector, the fourth adjusted location vector, and the fifth adjusted location vector.

[0010] Optionally, obtaining the first merchant information to which the first delivery object belongs includes: obtaining a first prompt message sent by the delivery resource terminal used to obtain the delivery resource to prompt the delivery resource to trigger the delivery operation of the first delivery object; and obtaining the first merchant information to which the first delivery object belongs based on the first prompt message and the first delivery object information.

[0011] Optionally, obtaining the second merchant information to which the second object to be delivered belongs, which has been claimed by the delivery resource, includes: querying the second object to be delivered that has been claimed by the delivery resource within a preset time range based on the first prompt message; and determining the second merchant information corresponding to the second object to be delivered based on the second object to be delivered.

[0012] Optionally, obtaining the second merchant target location vector corresponding to the second merchant information includes: inputting the second merchant information into the merchant target location vector model to obtain the second merchant target location vector output by the merchant target location vector model. The merchant target location vector model is used to obtain the second target merchant location vector based on the initial location vector corresponding to each layer of the multi-layer graph structure of the second merchant. The multi-layer graph structure includes a near-field wireless node data layer graph structure and a historical behavior data layer graph structure of delivery resources.

[0013] Optionally, the step of inputting the second merchant information into the merchant target location vector model to obtain the second merchant target location vector corresponding to the second merchant information output by the merchant target location vector model includes: obtaining the fourth initial location vector of the second merchant corresponding to the second merchant information in the near-field wireless node data layer graph structure according to the second merchant information; obtaining the fifth initial location vector of the second merchant corresponding to the second merchant information in the historical behavior data layer graph structure of the delivery resources; calculating the third loss value between the fourth initial location vector and the fifth initial location vector; and obtaining the second merchant target location vector according to the fourth initial location vector, the fifth initial location vector, and the third loss value, according to a preset calculation method.

[0014] Optionally, it further includes: determining the remaining time information of the delivery resource at the current time from the location of the first merchant; if the remaining time information is less than a first time information threshold, determining that the delivery resource has arrived at the location of the first merchant; and sending a first notification message to the first merchant terminal to indicate that the delivery resource has arrived at the location of the first merchant at the current time.

[0015] Optionally, determining the remaining time information of the delivery resource at the current moment from the location of the first merchant includes: detecting that the delivery resource receives a third wireless signal emitted by a third merchant at the current moment; obtaining third merchant information corresponding to the third merchant based on the third wireless signal, wherein the third merchant is at least one merchant passed by the delivery resource on its way to the location of the first merchant; inputting the third merchant information into the merchant target location vector model to obtain the third merchant target location vector output by the merchant target location vector model corresponding to the third merchant information; and determining the remaining time information of the delivery resource at the current moment from the location of the first merchant based on the first merchant target location vector and the third merchant target location vector.

[0016] This application embodiment also provides a method for predicting a merchant's target location, comprising: acquiring merchant information used to characterize merchant attributes; inputting the merchant information into a merchant target location vector model to obtain a merchant target location vector output by the merchant target location vector model corresponding to the merchant information; wherein, the merchant target location vector model is used to obtain the merchant target location vector based on the initial location vector corresponding to each layer of the graph structure in the multi-layer graph structure, the multi-layer graph structure including a near-field wireless node layer graph structure and a historical behavior data layer graph structure of delivery resources.

[0017] Optionally, it further includes: a first request message sent by the delivery resource terminal used to obtain the delivery resources to request the location of the object to be delivered; obtaining the merchant information to which the object to be delivered belongs based on the first request message, the merchant information including the merchant identification information of the merchant to which the object to be delivered belongs and the delivery information of the object to be delivered; the step of inputting the merchant information into the merchant target location vector model to obtain the merchant target location vector corresponding to the merchant information output by the merchant target location vector model includes: inputting the merchant identification information and the delivery information of the object to be delivered by the merchant as input data into the merchant target location vector model to obtain the merchant target location vector corresponding to the merchant information output by the merchant target location vector model, the merchant target location vector being a location vector used to represent the location of the delivery resources picking up the object to be delivered.

[0018] Optionally, it further includes: obtaining a second request message sent by a user terminal for requesting merchant target location information to reach the merchant; obtaining a first user location vector representing the user's location based on the second request message; and determining first path planning information for the user to reach the merchant target location vector based on the first user location vector and the merchant target location vector, wherein the merchant target location vector is the doorway location vector of the merchant.

[0019] This application also provides a method for constructing a multi-layer graph structure, comprising: constructing a near-field wireless node data layer graph structure representing the connection relationship between the target merchant and other merchants within a target area based on the near-field wireless node signals of the target merchant; constructing a wireless local area network (WLAN) data layer graph structure representing the connection relationship between the target merchant and other merchants within a target area based on the WLAN data information searched by the target merchant; acquiring historical behavior data of delivery resources within a first preset time period, and constructing a historical behavior data layer graph structure of delivery resources representing the connection relationship between the target merchant and other merchants within a target area; and constructing a multi-layer graph structure representing the connection relationship between the target merchant and other merchants within a target area based on the near-field wireless node data layer graph structure, the WLAN data layer graph structure, and the historical behavior data layer graph structure of delivery resources.

[0020] Optionally, the multi-layer graph structure includes a weight value for representing the connection relationship between the target merchant and other merchants, and a weight value corresponding to the connection relationship between the target merchant and other merchants; the method further includes: sorting the multiple weight values ​​for representing the connection relationship between the target merchant and other merchants; determining a target weight value according to the sorting order of the weight values; and using the target merchant corresponding to the target weight value and other merchants as merchants recommended to the user.

[0021] This application embodiment also provides a training method for a merchant target location vector model, comprising: constructing a multi-layer graph structure for representing the location of a target merchant within a target area where delivery resources arrive; the multi-layer graph structure including a near-field wireless node data layer graph structure and a historical behavior data layer graph structure of delivery resources; inputting the near-field wireless node data layer graph structure into a merchant initial location vector model to obtain a first initial location vector of the target merchant in the near-field wireless node data layer graph structure output by the merchant initial location vector model; inputting the historical behavior data layer graph structure of delivery resources into the merchant initial location vector model to obtain a second initial location vector of the target merchant in the historical behavior data layer graph structure of delivery resources output by the merchant initial location vector model; calculating a first loss value between the first initial location vector of the target merchant and the second initial location vector of the target merchant; and training the merchant initial location vector model based on the first loss value to obtain a merchant target location vector model, wherein the merchant target location vector model is used to obtain the merchant target location vector of the target merchant within the target area.

[0022] Optionally, the construction of a multi-layer graph structure for characterizing the location of the target merchant within the target area where the delivery resource arrives includes: constructing a near-field wireless node data layer graph structure based on the wireless signals received by the delivery resource from the merchant; and constructing a historical behavior data layer graph structure based on the time information of the delivery resource from the location of the first merchant to the location of the second merchant.

[0023] Optionally, the step of constructing a near-field wireless node data layer graph structure based on the wireless signals received by the delivery resources from the merchants includes: acquiring multiple wireless signal groups received by the delivery resources within a first preset time period, wherein each wireless signal group refers to the wireless signals received by the delivery resources from two merchants respectively within a second preset time period; calculating a first weight value for each wireless signal group in the multiple wireless signal groups; determining a first connection relationship between the two merchants in each wireless signal group based on the first weight value corresponding to each wireless signal group; and constructing a near-field wireless node data layer graph structure based on the first connection relationship between the two merchants in each wireless signal group.

[0024] Optionally, the wireless signal group is obtained by: acquiring the first wireless signal from the first merchant and the second wireless signal from the second merchant received by the delivery resource within a second preset time period; and using the first wireless signal and the second wireless signal received by the delivery resource within the second preset time period as the first wireless signal group received by the delivery resource within the second preset time period.

[0025] Optionally, determining the first connection relationship between two merchants in each wireless signal group based on the first weight value corresponding to each wireless signal group includes: determining the first connection distance relationship between two merchants in each wireless signal group based on the first weight value corresponding to each wireless signal group, wherein the first weight value and the first connection distance relationship have a negative correlation.

[0026] Optionally, constructing a historical behavior data layer graph structure for the delivery resource based on the time information of the delivery resource traveling from the location of the first merchant to the location of the second merchant includes: obtaining the time information of the delivery resource traveling from the location of the first merchant to the location of the second merchant; taking the process of the delivery resource traveling from the location of the first merchant to the location of the second merchant as a first behavior data group of the delivery resource; obtaining a second weight value corresponding to the first behavior data group of the delivery resource based on the time information; and constructing a historical behavior data layer graph structure for the delivery resource based on the second weight value corresponding to the first behavior data group.

[0027] Optionally, constructing the historical behavior data layer graph structure of the delivery resources based on the second weight value corresponding to the first behavior data group includes: determining a second connection distance relationship between the first merchant and the second merchant in the first behavior data group based on the second weight value, wherein the second weight value and the second connection distance relationship are positively correlated; and constructing the historical behavior data layer graph structure of the delivery resources based on the second connection distance relationship between the first merchant and the second merchant.

[0028] Optionally, training the merchant's initial location vector model based on the first loss value to obtain the merchant's target location vector model includes: adjusting the parameters in the trained merchant's initial location vector model according to the first loss value.

[0029] Optionally, the method further includes: stopping training the merchant initial location vector model if the first loss value is less than a first preset loss value threshold; or stopping training the merchant initial location vector model if the first loss value is greater than the first preset loss value threshold and the difference between multiple first loss values ​​is less than a first preset difference threshold.

[0030] Optionally, the multi-layer map structure further includes a wireless local area network (WLAN) data layer map structure; the method further includes: inputting the WLAN data layer map structure into the merchant initial location vector model to obtain the third initial location vector of the target merchant in the WLAN data layer map structure output by the merchant initial location vector model; calculating a second loss value among the first initial location vector of the target merchant, the second initial location vector of the target merchant, and the third initial location vector of the target merchant; and training the merchant initial location vector model based on the second loss value to obtain the merchant target location vector model.

[0031] Optionally, the construction of a multi-layer graph structure to characterize the location of the target merchant within the target area where the delivery resource arrives includes: constructing a near-field wireless node data layer graph structure based on the wireless signals received by the delivery resource from the merchant; constructing a historical behavior data layer graph structure based on the time information of the delivery resource from the location of the first merchant to the location of the second merchant; and constructing a wireless local area network (WLAN) data layer graph structure based on the WLAN data information searched by the target merchant.

[0032] Optionally, the step of constructing a wireless LAN data layer graph structure based on the wireless LAN data information searched by the target merchant includes: obtaining a list of wireless LAN data information searched by the target merchant, the list of wireless LAN data information including wireless LAN data information of multiple other merchants; determining a third connection distance relationship between the target merchant and other merchants based on the wireless LAN signal strength values ​​received from other merchants in the list of wireless LAN data information; and constructing a wireless LAN data layer graph structure based on the third connection distance relationship between the target merchant and other merchants.

[0033] This application embodiment also provides a method for predicting the arrival of delivery resources at a target merchant, including: obtaining information about a fourth merchant to which a fourth delivery object already picked up by the delivery resource belongs at the current time; inputting the fourth merchant information into a merchant target location vector model to obtain a fourth merchant target location vector output by the merchant target location vector model corresponding to the fourth merchant information, wherein the merchant target location vector model is used to obtain the fourth merchant target location vector based on the initial location vector corresponding to each layer of the graph structure in the multi-layer graph structure, the multi-layer graph structure including a near-field wireless node data layer graph structure and the historical behavior of the delivery resources. The data layer graph structure is as follows: Information about the fifth merchant to which the fifth delivery object belongs is obtained; the target location vector of the fifth merchant corresponding to the fifth merchant information is obtained; the target location vectors of the fourth and fifth merchants are used as input data and input into the duration prediction network model to obtain the second predicted duration information output by the duration prediction network model, representing the delivery resource's journey from the location of the fourth merchant to the location of the fifth merchant; if the second predicted duration information meets preset duration verification conditions, the fifth merchant is determined to be the target merchant that the delivery resource needs to reach after leaving the fourth merchant.

[0034] Optionally, the fifth merchant information includes at least one fifth candidate merchant information; the method further includes: for each fifth candidate merchant information in the at least one fifth candidate merchant information, obtaining second candidate predicted time information of the delivery resource from the location of the fourth merchant to the location of the fifth candidate merchant; the second predicted time information meets preset time review conditions, including: sorting multiple second candidate predicted time information according to the sorting rule of time from smallest to largest, to obtain second target predicted time information, wherein the second target predicted time information is less than other second candidate predicted time information in the multiple second candidate predicted time information.

[0035] Optionally, the step of inputting the fourth merchant information into the merchant target location vector model to obtain the fourth merchant target location vector corresponding to the fourth merchant information output by the merchant target location vector model includes: obtaining the sixth initial location vector of the fourth merchant corresponding to the fourth merchant information in the near-field wireless node data layer graph structure, and obtaining the seventh initial location vector of the fourth merchant corresponding to the fourth merchant information in the historical behavior data layer graph structure of the delivery resources; calculating the fourth loss value between the sixth initial location vector and the seventh initial location vector; and obtaining the fourth merchant target location vector according to a preset calculation method based on the sixth initial location vector, the seventh initial location vector, and the fourth loss value.

[0036] Optionally, obtaining the fourth merchant information to which the fourth delivery object to which the delivery resource has been claimed at the current time belongs includes: obtaining a third notification message sent by the fourth merchant terminal used by the fourth merchant, indicating that the delivery resource has been successfully claimed at the current time; and obtaining the fourth merchant information to which the fourth delivery object to which the delivery resource has been claimed at the current time belongs based on the third notification message.

[0037] Optionally, obtaining the fifth merchant target location vector corresponding to the fifth merchant information includes: inputting the fifth merchant information into the merchant target location vector model to obtain the fifth merchant target location vector output by the merchant target location vector model corresponding to the fifth merchant information; wherein, the merchant target location vector model is used to query the fifth merchant target location vector corresponding to the fifth merchant information in a pre-stored correspondence table between merchant information and merchant target location vectors based on the fifth merchant information, and the fifth merchant target location vector is obtained based on the initial location vector corresponding to each layer of the graph structure in the multi-layer graph structure, wherein the multi-layer graph structure includes a near-field wireless node data layer graph structure and a historical behavior data layer graph structure of delivery resources.

[0038] This application embodiment also provides an apparatus for predicting the arrival time of delivery resources at merchants, comprising: a first acquisition unit, configured to acquire first merchant information to which a first delivery object to be picked up belongs; and a first merchant target location vector acquisition unit, configured to input the first merchant information into a merchant target location vector model to obtain a first merchant target location vector corresponding to the first merchant information output by the merchant target location vector model, wherein the merchant target location vector model is used to obtain the first merchant target location vector based on the initial location vector corresponding to each layer of the multi-layer graph structure of the first merchant, the multi-layer graph structure including... The system includes a near-field wireless node data layer graph structure and a historical behavior data layer graph structure for delivery resources; a second merchant target location vector acquisition unit, used to acquire the second merchant information to which the second delivery object to be delivered belongs, and to acquire the second merchant target location vector corresponding to the second merchant information; and a first prediction duration information acquisition unit, used to input the second merchant target location vector and the first merchant target location vector as input data into the duration prediction network model, and to obtain the first prediction duration information output by the duration prediction network model, which represents the delivery resource's journey from the location of the second merchant to the location of the first merchant.

[0039] This application embodiment also provides an apparatus for predicting the target location of a merchant, comprising: a second acquisition unit for acquiring merchant information used to characterize merchant attributes; and a third acquisition unit for inputting the merchant information into a merchant target location vector model to obtain a merchant target location vector corresponding to the merchant information output by the merchant target location vector model; wherein, the merchant target location vector model is used to obtain the merchant target location vector based on the initial location vector corresponding to each layer of the multi-layer graph structure, the multi-layer graph structure including a near-field wireless node layer graph structure and a historical behavior data layer graph structure of delivery resources.

[0040] This application embodiment also provides an apparatus for constructing a multi-layer graph structure, comprising: a first construction unit, configured to construct a near-field wireless node data layer graph structure representing the connection relationship between the target merchant and other merchants within a target area based on the near-field wireless node signals of the target merchant; a second construction unit, configured to construct a wireless local area network data layer graph structure representing the connection relationship between the target merchant and other merchants within a target area based on the wireless local area network data information searched by the target merchant; a third construction unit, configured to acquire historical behavior data of delivery resources within a first preset time period, and construct a historical behavior data layer graph structure of delivery resources representing the connection relationship between the target merchant and other merchants within a target area; and a fourth construction unit, configured to construct a multi-layer graph structure representing the connection relationship between the target merchant and other merchants within a target area based on the near-field wireless node data layer graph structure, the wireless local area network data layer graph structure, and the historical behavior data layer graph structure of delivery resources.

[0041] This application embodiment also provides a training device for a merchant target location vector model, comprising: a fifth construction unit, configured to construct a multi-layer graph structure representing the location of a target merchant within a target area reached by delivery resources, the multi-layer graph structure including a near-field wireless node data layer graph structure and a historical behavior data layer graph structure of delivery resources; a first initial location vector acquisition unit, configured to input the near-field wireless node data layer graph structure into the merchant initial location vector model to obtain the first initial location vector of the target merchant in the near-field wireless node data layer graph structure output by the merchant initial location vector model; and a second initial location vector acquisition unit. The system includes a calculation unit for inputting the historical behavior data layer graph structure of the delivery resources into the merchant initial location vector model to obtain the second initial location vector of the target merchant in the historical behavior data layer graph structure of the delivery resources, output by the merchant initial location vector model; a calculation unit for calculating a first loss value between the first initial location vector of the target merchant and the second initial location vector of the target merchant; and a training unit for training the merchant initial location vector model based on the first loss value to obtain a merchant target location vector model, wherein the merchant target location vector model is used to obtain the merchant target location vector of the target merchant in the target area.

[0042] This application embodiment also provides an apparatus for predicting the arrival of delivery resources at a target merchant, comprising: a fourth acquisition unit for acquiring information about a fourth merchant to which a fourth delivery object already picked up at the current time belongs; a fourth merchant target location vector acquisition unit for inputting the fourth merchant information into a merchant target location vector model to obtain a fourth merchant target location vector output by the merchant target location vector model, wherein the merchant target location vector model is used to obtain the fourth merchant target location vector based on the initial location vector corresponding to each layer of the multi-layer graph structure, the multi-layer graph structure including a near-field wireless node data layer graph structure and a historical behavior data layer graph structure of delivery resources; and a fifth merchant target location vector acquisition unit for acquiring information about a fourth merchant to be delivered to a target merchant. The system includes a fifth merchant information unit, which obtains the target location vector of the fifth merchant corresponding to the fifth merchant information; a second prediction duration information acquisition unit, which inputs the target location vector of the fourth merchant and the target location vector of the fifth merchant as input data into the duration prediction network model to obtain the second prediction duration information output by the duration prediction network model, which represents the duration of the delivery resource from the location of the fourth merchant to the location of the fifth merchant; and an approval unit, which determines the fifth merchant as the target merchant that the delivery resource needs to reach in the next moment if the second prediction duration information meets the preset duration approval conditions, and the second prediction duration information is the target duration information of the delivery resource from the location of the fourth merchant to the location of the fifth merchant.

[0043] This application also provides an electronic device, which includes a processor and a memory; the memory stores a computer program, and the processor executes the above-described method after running the computer program.

[0044] This application also provides a computer storage medium storing a computer program, which, when run by the processor, executes the above-described method.

[0045] Compared with the prior art, the embodiments of this application have the following advantages:

[0046] This application provides a method for predicting the arrival time of delivery resources at a merchant, comprising: obtaining information about a first merchant to which a first delivery object to be picked up belongs; inputting the first merchant information into a merchant target location vector model to obtain a first merchant target location vector output by the merchant target location vector model, wherein the merchant target location vector model is used to obtain the first merchant target location vector based on the initial location vector corresponding to each layer of the graph structure in the multi-layer graph structure, the multi-layer graph structure including a near-field wireless node data layer graph structure and a historical behavior data layer graph structure of delivery resources; obtaining information about a second merchant to which a second delivery object to which the delivery resource has been picked up belongs, and obtaining a second merchant target location vector corresponding to the second merchant information; inputting the second merchant target location vector and the first merchant target location vector as input data into a duration prediction network model to obtain first predicted duration information output by the duration prediction network model, representing the arrival time of the delivery resource from the location of the second merchant to the location of the first merchant.

[0047] In the above method, the first delivery target and the corresponding first merchant information are determined. Then, the second merchant information corresponding to the second delivery target that the delivery resource has already picked up before picking up the first delivery target is obtained. The first merchant target location vector and the second merchant target location vector are obtained through a merchant target location vector model. In this process, the merchant target location vector is calculated based on the initial location vector of the merchant in each layer of the multi-layer graph structure, which includes a near-field wireless node data layer graph structure and a historical behavior data layer graph structure of the delivery resource. This multi-layer graph structure improves the accuracy of the merchant target location vector and also improves the prediction accuracy of the delivery resource arriving at the merchant's location. Based on this, a duration prediction network model is used to obtain the first prediction duration information between the first and second merchant target location vectors, improving the prediction of the delivery resource's arrival time at the target merchant.

[0048] This application embodiment also provides a method for predicting a merchant's target location, comprising: acquiring merchant information used to characterize merchant attributes; inputting the merchant information into a merchant target location vector model to obtain a merchant target location vector output by the merchant target location vector model corresponding to the merchant information; wherein, the merchant target location vector model is used to obtain the merchant target location vector based on the initial location vector corresponding to each layer of the graph structure in the multi-layer graph structure, the multi-layer graph structure including a near-field wireless node layer graph structure and a historical behavior data layer graph structure of delivery resources.

[0049] In the above method, based on merchant information, a merchant target location vector model is used to determine the merchant's target location vector. This is because the merchant target location vector model is based on the merchant's initial location vector in the near-field wireless node layer graph structure and its initial location vector in the historical behavior data layer graph structure of delivery resources. In other words, determining the merchant's target location vector references the merchant's near-field wireless node signals and historical behavior data, thus improving the accuracy of determining the merchant's target location.

[0050] This application provides a method for constructing a multi-layer graph structure, comprising: constructing a near-field wireless node data layer graph structure representing the connection relationship between the target merchant and other merchants within a target area based on the near-field wireless node signals of the target merchant; constructing a wireless local area network (WLAN) data layer graph structure representing the connection relationship between the target merchant and other merchants within a target area based on the WLAN data information searched by the target merchant; acquiring historical behavior data of delivery resources within a first preset time period, and constructing a historical behavior data layer graph structure of delivery resources representing the connection relationship between the target merchant and other merchants within a target area; and constructing a multi-layer graph structure representing the connection relationship between the target merchant and other merchants within a target area based on the near-field wireless node data layer graph structure, the WLAN data layer graph structure, and the historical behavior data layer graph structure of delivery resources.

[0051] The above method constructs a multi-layered graph structure to represent the connection relationship between the target merchant and other merchants in multiple dimensions. Based on the connection relationship between merchants in the multi-layered graph structure, the selection rate of the target merchant is determined, thereby improving the selection rate of the target merchant in the target area.

[0052] This application embodiment also provides a method for predicting the arrival of delivery resources at a target merchant, including: obtaining information about a fourth merchant to which a fourth delivery object already picked up by the delivery resource belongs at the current time; inputting the fourth merchant information into a merchant target location vector model to obtain a fourth merchant target location vector output by the merchant target location vector model corresponding to the fourth merchant information, wherein the merchant target location vector model is used to obtain the fourth merchant target location vector based on the initial location vector corresponding to each layer of the graph structure in the multi-layer graph structure, the multi-layer graph structure including a near-field wireless node data layer graph structure and the historical behavior of the delivery resources. The data layer graph structure is as follows: Information about the fifth merchant to which the fifth delivery object belongs is obtained; the target location vector of the fifth merchant corresponding to the fifth merchant information is obtained; the target location vectors of the fourth and fifth merchants are used as input data and input into the duration prediction network model to obtain the second predicted duration information output by the duration prediction network model, representing the delivery resource's journey from the location of the fourth merchant to the location of the fifth merchant; if the second predicted duration information meets preset duration verification conditions, the fifth merchant is determined to be the target merchant that the delivery resource needs to reach after leaving the fourth merchant.

[0053] The above method, after the delivery resource has received the information that the delivery resource has completed receiving the fourth delivery object, allows the server to determine the next delivery object to be received. Specifically, after determining the location information of the fourth merchant through a merchant target location vector model, and then determining the location information of the merchants corresponding to the multiple delivery objects to be received by the delivery resource, the distance between the location of the fourth merchant and the location of the fifth merchant is calculated. Then, a second predicted time information is determined for the delivery resource to travel from the location of the fourth merchant to the location of the fifth merchant. After judging the second predicted time information, the target merchant that the delivery resource needs to reach after receiving the fourth delivery object is determined. This method can prioritize matching the delivery resource with target merchants and receive target delivery objects from them, improving the efficiency of the delivery resource in receiving delivery objects. Attached Figure Description

[0054] Figure 1 This is an application scenario diagram showing how a delivery recipient picks up a delivery object within a target area, as provided in an embodiment of this application.

[0055] Figure 2 This is an application scenario diagram illustrating the predicted delivery time required for delivery resources to reach the target merchant, as provided in the embodiments of this application.

[0056] Figure 3This is an application scenario diagram illustrating the time information for predicting the arrival time of delivery resources from the location of the second merchant to the location of the first merchant, as provided in the embodiments of this application.

[0057] Figure 4 This is an application scenario diagram of the training method for the merchant target location vector model in the embodiments of this application.

[0058] Figure 5 This is an application scenario diagram of the multilayer map structure provided in the embodiments of this application.

[0059] Figure 6 This is a flowchart of a method for predicting the arrival time of delivery resources at merchants, provided in the first embodiment of this application.

[0060] Figure 7 This is a flowchart of a method for predicting the target location of a merchant, provided in the second embodiment of this application.

[0061] Figure 8 This is a flowchart of a method for constructing a multi-layered graph structure provided in the third embodiment of this application.

[0062] Figure 9 This is a flowchart illustrating a training method for a merchant target location vector model provided in the fourth embodiment of this application.

[0063] Figure 10 This is a flowchart illustrating a method for predicting the arrival of delivery resources at a target merchant, as provided in the fifth embodiment of this application.

[0064] Figure 11 This is a schematic diagram of an apparatus for predicting the arrival time of delivery resources at merchants, according to the sixth embodiment of this application.

[0065] Figure 12 This is a schematic diagram of a device for predicting the target location of a merchant, according to the seventh embodiment of this application.

[0066] Figure 13 This is a schematic diagram of an apparatus for constructing a multi-layered map structure according to the eighth embodiment of this application.

[0067] Figure 14 This is a schematic diagram of a training device for a merchant target location vector model according to the ninth embodiment of this application.

[0068] Figure 15 This is a schematic diagram of a device for predicting the arrival of delivery resources at a target merchant, according to the ninth embodiment of this application.

[0069] Figure 16 This is a schematic diagram of an electronic device provided in the eleventh embodiment of this application. Detailed Implementation

[0070] Many specific details are set forth in the following description to provide a full understanding of this application. However, this application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of this application; therefore, this application is not limited to the specific embodiments disclosed below.

[0071] The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to limit the application. The descriptive terms used in this application and the appended claims, such as "a," "first," and "second," are not intended to limit quantity or sequence, but rather to distinguish information of the same type from one another.

[0072] This application first provides a method, apparatus, electronic device, and computer storage medium for predicting the arrival time of delivery resources at merchants. This application also relates to a method, apparatus, electronic device, and computer storage medium for predicting the target location of merchants. This application further relates to a method, apparatus, electronic device, and computer storage medium for constructing a multi-layer map structure. This application also provides a method, apparatus, electronic device, and computer storage medium for predicting the arrival time of delivery resources at target merchants. This application also provides a method, apparatus, electronic device, and computer storage medium for training a merchant target location vector model.

[0073] The following are specific examples.

[0074] To more clearly explain the method for predicting the arrival time of delivery resources for merchants provided in this application, let's first introduce the application scenarios of this method.

[0075] The method for predicting the arrival time of delivery resources at merchants provided in this application can be applied to scenarios where delivery resources are located within a shopping mall environment, picking up items to be delivered from merchants. These items include goods, meals, etc. In existing technologies, within a shopping mall environment, users or delivery resources determine their specific merchant location information using Wi-Fi signal positioning. However, when the Wi-Fi signal within the mall is weak, the location of the delivery resource determined by the Wi-Fi signal may differ significantly from its actual location.

[0076] Therefore, there are instances where delivery resources experience delivery delays when delivering takeout meals to users' destinations, and it is impossible to determine whether the delays are caused by delivery resources failing to arrive at the merchant's store in a timely manner or by the merchant failing to prepare the meals in a timely manner.

[0077] Furthermore, regarding whether delivery resources arrive at the merchant's store in a timely manner, in actual operation, delivery resources often click the "arrived at merchant" trigger command before arriving at the merchant's store, or after the delivery resources have picked up the food from the merchant and entered the food delivery stage, the "arrived at merchant" trigger command is manually clicked. The server cannot accurately determine the legality of this operation.

[0078] To address the aforementioned issues, this application provides a method for predicting the arrival time of delivery resources at merchants, thereby improving the accuracy of such predictions. This is described in detail below.

[0079] Please refer to Figure 1 This is an application scenario diagram of a delivery object picking up a delivery object in a target area, as provided in the embodiments of this application.

[0080] This example illustrates how delivery resources can pick up multiple customers from various merchants within a target shopping mall. During peak delivery periods, the timeliness of food delivery is paramount. Therefore, improving the efficiency of delivery resources at each stage of food delivery is crucial for enhancing the overall efficiency of the food delivery chain.

[0081] A large food court includes multiple merchants. During peak delivery periods, a delivery resource needs to pick up food items from multiple merchants within a target time frame. To save time for the delivery resource to pick up food items from the merchants, this application embodiment uses a method to predict the arrival time of the delivery resource at the merchants. First, it determines the target location information of the first merchant corresponding to the food items to be picked up by the delivery resource. Then, it obtains the target location information of the second merchant corresponding to the second merchant to which the delivery resource has currently picked up food items. Based on the target location information between the two merchants, it determines the first predicted time information for the delivery resource to arrive at the location of the merchant to which the food items to be picked up belong.

[0082] For example, in Figure 1 In this scenario, the user corresponding to delivery resource 101 needs to collect the meals to be delivered from merchant 1, merchant 2, merchant 3, and merchant 4 within 10 minutes. In order to improve the efficiency of delivery resources in collecting the meals to be delivered, it is crucial to determine the time information of the delivery resources arriving at the location of the merchants.

[0083] like Figure 1 As shown, each merchant places the food to be picked up at the pickup point at the merchant's entrance. The time required for delivery resources to pick up the food from the pickup point at the merchant's entrance is shorter than the time required to pick up the food from the counter inside the merchant, which can improve the efficiency of delivery resources in picking up the food to be delivered.

[0084] Therefore, in order to improve the efficiency of delivery resources in picking up goods to be delivered, this embodiment first determines the merchant target location vector of the merchant to which the goods to be picked up belongs as the pick-up point at the merchant's door. Then, it calculates the time required for delivery resources to reach the pick-up point at the merchant's door. This method firstly improves the accuracy of obtaining the merchant target location information, and secondly saves the time required for delivery resources to reach the merchant to pick up the goods to be delivered.

[0085] In addition, please refer to Figure 2 This is an application scenario diagram illustrating the time required for predicted delivery resources to reach the target merchant, provided in the embodiments of this application.

[0086] S201: Delivery resource terminal 101 sends a notification message to server 102 to receive the first object to be delivered. S202: Obtain the first merchant information to which the first order to be delivered belongs.

[0087] Based on the notification message obtained in step S201, the server obtains the first merchant information of the first order to be delivered. The first merchant information includes the merchant name information of the first merchant, etc.

[0088] Here, after obtaining the first merchant's information, the merchant target location vector model is used to obtain the first merchant's target location vector corresponding to the first merchant's information. This merchant target location vector model is obtained by processing the initial location vector corresponding to each layer of the multi-layer graph structure. The multi-layer graph structure includes a near-field wireless node data layer graph structure and a historical behavior data layer graph structure for delivery resources. Here, the merchant's target location vector is determined using the location vectors of the merchant nodes in two layers of the graph structure. Alternatively, the multi-layer graph structure can also include three layers: a near-field wireless node data layer graph structure, a wireless LAN data layer graph structure, and a historical behavior data layer graph structure for delivery resources.

[0089] S203: Obtain the information of the second merchant to which the second object to be delivered belongs after the delivery resources have been claimed.

[0090] The server retrieves the last item picked up for delivery at the current moment and designates it as the second item to be delivered. It then obtains the information of the second merchant associated with this second item. For example... Figure 1 In the middle, the rider was about to pick up the food at the pickup point in front of Merchant 1 when he found that the rider had just picked up the food from Merchant 2.

[0091] This section retrieves the second object to be delivered that has been claimed by the delivery resource. This refers to the second object to be delivered that was last claimed before the current time. The location information of the delivery resource is determined based on the second object to be delivered.

[0092] S204: Obtain the first predicted time information for the delivery resource to travel from the location of the second merchant to the location of the first merchant. After obtaining the target location vectors of the first and second merchants in the above steps, the first predicted time information required for the delivery resource to travel from the second merchant to the first merchant is obtained according to the time prediction model.

[0093] For example, a delivery resource plans to pick up the first item to be delivered from merchant A's store at the current moment. At this time, it is found that the delivery resource has completed the pickup operation for the second item to be delivered from merchant B. Therefore, by calculating the predicted time it takes for the delivery resource to travel from the location of merchant B to the location of merchant A, the arrival time of the delivery resource at merchant A can be determined. In this process, to improve the accuracy of obtaining the arrival time information of the delivery resource at merchant A, this application first improves the accuracy of obtaining the merchant location information of the current location of the delivery resource through a merchant target location vector model. Based on this, the accuracy of obtaining the arrival time information of the delivery resource at the target merchant's location is improved.

[0094] S205: Send a message to the merchant that the delivery resources will arrive at the first merchant's location to pick up the first delivery object after the first predicted time information. S206: After the second time information, send a message to the merchant that the delivery resources are about to arrive at the first merchant's location. Please provide the first delivery object for the delivery resources as soon as possible. The time difference between the first predicted time information and the second time information is less than the preset time difference.

[0095] When the delivery resource is heading to the location of the first merchant, it obtains information about the third merchant based on the short-range wireless signals (e.g., Bluetooth signals) and wireless LAN signals received by the delivery resource from other merchants along the way. Based on the information received from the third merchant, it determines the location of the delivery resource and then obtains the target location vector of the third merchant.

[0096] If the time it takes for the third merchant's location to reach the first merchant's location is less than the first preset time, it means that the delivery resources have reached the vicinity of the first merchant, and it can be confirmed that the delivery resources have reached the first merchant, thus completing the in-store stage task.

[0097] S207: After the second duration information, a prompt message is sent to the delivery resource terminal to inform you that you have entered the area of ​​the first merchant and are waiting to pick up your meal.

[0098] Here, a notification message is sent to the delivery resource provider, indicating that the delivery resource has arrived at the location of the first merchant. This avoids manually triggering the store arrival operation after the delivery resource has arrived at the store. Figure 1The example shown illustrates how delivery resources can collect meals from multiple adjacent merchants within a short period, saving processing time during the in-store pickup process and improving the efficiency of in-store meal collection.

[0099] Please refer to Figure 3 This is an application scenario diagram illustrating the time information for delivery resources to travel from the location of the second merchant to the location of the first merchant, as provided in the embodiments of this application.

[0100] exist Figure 3 In the model, the input data 301 of the merchant target location vector model is either the first merchant information or the second merchant information. The output data 302 of the merchant target location vector model is either the first merchant target location vector corresponding to the first merchant information or the second merchant target location vector corresponding to the second merchant information.

[0101] The input data of the duration prediction network model is the output data 302 of the merchant target location vector, which is the first merchant target location vector and the second merchant target location vector as input data. The output data 303 of the duration prediction network model is the first prediction duration information, which refers to the time information of the delivery resource from the location of the second merchant to the location of the first merchant.

[0102] To improve the accuracy of predicting the arrival time of delivery resources at the first merchant, this embodiment first obtains the target location vector of the first merchant and the target location vector of the second merchant through a merchant target location vector model, thereby improving the accuracy of obtaining merchant location information.

[0103] The merchant target location vector model is used to obtain the merchant's target location vector based on the merchant's initial location vector in each layer of the multi-layered graph structure. This multi-layered graph structure includes a near-field wireless node data layer and a historical behavior data layer for delivery resources. In other words, the merchant target location vector model determines the first merchant's target location vector by referencing the first merchant's wireless signal and historical behavior data of delivery resources arriving at the first merchant from other merchants, thereby improving the accuracy of the obtained first merchant target location vector.

[0104] Specifically, the merchant target location vector model obtains the first merchant target location vector corresponding to the first merchant information using the following method. Please refer to [reference needed]. Figure 4 This is an application scenario diagram of the training method for the merchant target location vector model in the embodiments of this application.

[0105] First, a first initial position vector of the first merchant is obtained from the near-field wireless node data layer graph structure, and a second initial position vector of the first merchant is obtained from the historical behavior data layer graph structure of the delivery resources. A first loss value between the first and second initial position vectors is calculated according to a preset calculation method. Based on the first initial position vector and the first loss value, a first adjusted position vector of the first merchant in the near-field wireless node data layer graph structure is obtained. Based on the second initial position vector and the first loss value, a second adjusted position vector of the first merchant in the historical behavior data layer graph structure of the delivery resources is obtained. Based on the first and second adjusted position vectors, the target position vector of the first merchant is obtained.

[0106] The merchant target location vector model is obtained by training the initial merchant location vector model. The training method for the merchant target location vector model is described below; please refer to [the relevant documentation]. Figure 4 This is an application scenario diagram of the training method for the merchant target location vector model provided in the embodiments of this application.

[0107] The first step is to construct a multi-layer graph structure to characterize the location of the target merchant within the target area where the delivery resources arrive. The multi-layer graph structure includes a near-field wireless node data layer graph structure and a historical behavior data layer graph structure of the delivery resources.

[0108] The second step involves inputting the near-field wireless node data layer graph structure into the merchant's initial location vector model to obtain the first initial location vector of the target merchant in the near-field wireless node data layer graph structure, as output by the merchant's initial location vector model. The second initial location vector of the target merchant in the historical behavior data layer graph structure of the delivery resources is then input into the merchant's initial location vector model to obtain the second initial location vector of the target merchant in the historical behavior data layer graph structure of the delivery resources, as output by the merchant's initial location vector model.

[0109] The third step is to calculate the first loss value between the first initial position vector and the second initial position vector.

[0110] The fourth step is to train the merchant's initial position vector model based on the first loss value to obtain the merchant's target position vector model. The merchant's target position vector model is used to obtain the merchant's target position vector in the multi-layer graph structure.

[0111] The following sections will discuss these points separately:

[0112] The multi-layered graph structure constructed in the first step, such as Figure 4 The two-layer spectral structure in the middle, and Figure 5 The three-layer spectral structure in [the diagram]. Please refer to [the diagram]. Figure 5 This is an application scenario diagram of the multilayer map structure provided in the embodiments of this application. Figure 5 The three-layer graph structure includes a near-field wireless node data layer graph structure, a wireless local area network data layer graph structure, and a historical behavior data layer graph structure for delivery resources.

[0113] (1) The construction method of the near-field wireless node data layer graph structure 501 is as follows: For example, if the delivery resource receives wireless signals from merchant A and merchant B simultaneously within 2 seconds, it indicates that merchant A and merchant B are adjacent merchants. The above process is called the first wireless signal group received by the delivery resource.

[0114] Then, the relationship between merchant A and merchant B as adjacent merchants is verified:

[0115] The system retrieves the number of times the delivery service receives wireless signals from the first wireless signal group within a target time period (e.g., the number of times the first wireless signal group is collected within one month). Based on the number of times the first wireless signal group is collected, a first weight value for the first wireless signal group among multiple signal groups is determined. If the first weight value corresponding to the first wireless signal group is greater than a first preset weight value, it indicates that the verification of the relationship between Merchant A and Merchant B as adjacent merchants is successful.

[0116] Accordingly, multiple sets of connection relationships between two merchants and the first weight value corresponding to the connection relationship are obtained. For example, the first wireless signal group is the wireless signal between merchant A and merchant B, the second wireless signal group is the wireless signal between merchant A and merchant C, the third wireless signal group is the wireless signal between merchant B and merchant D, and the fourth wireless signal group is the wireless signal between merchant C and merchant D.

[0117] Based on the first weight value corresponding to each wireless signal group, the first connection distance relationship between two adjacent merchants in that wireless signal group is determined, wherein the first weight value and the first connection distance relationship are negatively correlated. For example, if the first weight value of the first wireless signal group is greater than the first weight value of the second wireless signal group, then the connection distance between merchant A and merchant B is less than the first connection distance between merchant A and merchant C.

[0118] (2) The construction method of the wireless local area network data layer graph structure 502 is as follows: For example, the server obtains the wireless local area network data list of merchant A. The wireless local area network data list of merchant A includes the wireless local area network information of multiple merchants whose distance information from the location of merchant A is less than a preset distance threshold.

[0119] Based on the signal strength of each Wi-Fi signal in Merchant A's Wi-Fi list, the distance relationship between Merchant A and the merchants corresponding to each Wi-Fi signal is determined, and the signal strength is positively correlated with the distance relationship.

[0120] (3) The construction method of the historical behavior data layer graph structure 403 of the delivery resources is as follows: For example, obtain the time data required for the delivery resources to go from merchant A to merchant E, obtain the number of times the delivery resources go from merchant A to merchant E within the target time and the time data of each time. For example, the historical behavior records of the delivery resources going from merchant A to merchant E within 1 month include 600 times, obtain the average time data of the historical behavior data in the 600 times, and use the average time data as the target time data of the delivery resources going from merchant A to merchant E.

[0121] Obtain target time data for multiple sets of delivery resources from the first merchant to the second merchant.

[0122] For example, the first set of behavioral data for delivery resources is from merchant A to merchant E, the second set of behavioral data for delivery resources is from merchant A to merchant C, the third set of behavioral data for delivery resources is from merchant A to merchant B, and the fourth set of behavioral data for delivery resources is from merchant B to merchant D.

[0123] Obtain the second weight value of the target duration data of each group of behavioral data in the sum of the target duration data of all behavioral data. Based on the second weight value corresponding to the target duration data of each group of behavioral data, determine the second connection distance relationship between the two merchants corresponding to each group of behavioral data. The second weight value and the second connection distance relationship are positively correlated.

[0124] For example, if the second weight value corresponding to the first set of behavioral data is greater than the second weight value corresponding to the second set of behavioral data, it means that the distance relationship between merchant A and merchant E is greater than the distance relationship between merchant A and merchant C.

[0125] The above is a detailed explanation of the first step in constructing the multi-layered graph structure.

[0126] In addition, the second step is to obtain the location vector information of each merchant node in each layer of the graph structure.

[0127] Here, the merchant initial location vector model is used to obtain location vector information for the multi-layer map structure.

[0128] The merchant's initial location vector model uses a multi-layer graph convolutional network (Graph Convolutional Networks (GCN)) as the initial network model to be trained.

[0129] The corresponding calculation formula is as follows: (1)

[0130]

[0131] Where X represents the feature information of the merchant node. Here, the feature information of the merchant node includes the basic information of the merchant and the connection relationship between the merchant and other merchant nodes in the corresponding graph structure.

[0132] A is a matrix that represents the topological structure;

[0133] D is the degree matrix of A;

[0134] σ is a non-linear activation function;

[0135] W i These are parameters that are randomly initialized.

[0136] The method for calculating the position vector of merchant nodes in the multi-layer graph structure according to the above formula (1) is as follows:

[0137] The near-field wireless node data layer graph structure is input into the merchant's initial location vector model. This model obtains basic information about the merchant nodes and the connections between them, thus acquiring the target merchant's first initial location vector within the near-field wireless node data layer graph structure. For example... Figure 4 In the 401 graph structure, the near-field wireless node data layer graph structure is used as the first layer graph structure. The first initial position vectors of merchant node A, merchant node B, merchant node C, merchant node D, and merchant node E are obtained.

[0138] Correspondingly, the historical behavior data layer graph structure of delivery resources is input into the merchant initial location vector model. Based on the basic information of the merchant nodes and the duration information between two merchants corresponding to the historical behavior data of delivery resources, the merchant initial location vector model obtains the second initial location vector of the target merchant in the historical behavior data layer graph structure of delivery resources. For example... Figure 4 In the 402 graph structure, the historical behavior data layer of the delivery resources is used as the second layer graph structure. The second initial position vectors of merchant node A, merchant node B, merchant node C, merchant node D, and merchant node E are obtained.

[0139] The third step is to calculate the first loss value between the vectors of the same merchant node in different data layer graph structures for the same merchant node. The first loss value is calculated according to the following formula (2):

[0140]

[0141] The meanings of each part in the above formula are as follows:

[0142]

[0143] This represents the first loss value between the position vectors of a node in a two-layer graph structure. For example, the first loss value between the first and second initial position vectors of merchant node A. Where u i v can represent the first initial position vector of merchant node A. i This can represent the second initial position vector of merchant node A.

[0144]

[0145] Represents the first layer of the spectral structure (e.g.) Figure 3 The first initial position vector of merchant node A in the 301st layer) and the second layer graph structure (such as... Figure 3 The similarity value between the second initial position vector of merchant node A in the 302nd layer of the graph structure, where the first initial position vector of merchant node A in the first layer of the graph structure and the second initial position vector of merchant node A in the second layer of the graph structure are positive samples of each other.

[0146]

[0147] Represents the first layer of the spectral structure (e.g.) Figure 3 The first initial position vector of merchant node A in the 301st layer) and the second layer graph structure (such as... Figure 3 The similarity value between the second initial position vector of merchant node B in the 302nd layer of the graph structure, where the first initial position vector of merchant node A in the first layer of the graph structure and the second initial position vector of merchant node B in the second layer of the graph structure are negative samples of each other.

[0148]

[0149] Represents the first layer of the spectral structure (e.g.) Figure 3 The first initial position vector of merchant node A in layer 301 and the first layer graph structure (such as...) Figure 3 The similarity value between the first initial position vector of merchant node B in the 301st layer of the graph structure, where the first initial position vector of merchant node A in the first layer of the graph structure and the first initial position vector of merchant node B in the first layer of the graph structure are negative samples of each other.

[0150] Therefore, the first loss value between the first initial position vector and the second initial position vector of merchant node A is calculated according to the above formula (2).

[0151] The first loss value calculated in formula (2) above is based on a two-layer graph structure, calculating the loss value between the position vectors corresponding to the same merchant node in the two-layer graph structure. If the multi-layer graph structure is a three-layer graph structure, then the loss value between the three position vectors corresponding to the same merchant node in the three-layer graph structure is calculated.

[0152] After obtaining the first loss value, the parameters of the merchant's initial location vector model are adjusted based on this first loss value. Adjusting the parameters of the merchant's initial location vector model means adjusting the parameter W in formula (1). i Adjustments will be made.

[0153] Here, each layer of the graph structure is input into a multi-layer graph convolutional network (GCN) model to obtain the position vector of each merchant node in that layer of the graph structure. Therefore, for parameter W... i When making adjustments, each layer of the spectral structure corresponds to an adjusted parameter W. i .

[0154] Each time the parameters of the merchant's initial position vector model are adjusted, the first adjusted position vector of merchant node A in the first layer of the graph structure and the second adjusted position vector of merchant node A in the second layer of the graph structure are obtained. The first loss value between the first adjusted position vector and the second adjusted position vector of merchant node A is calculated according to the merchant formula (2).

[0155] If, after one or more adjustments, the first loss value is less than the first preset loss value threshold, then the adjustment of the merchant's initial location vector model is stopped, and the merchant's target location vector model is obtained. At this time, the merchant's target location vector model is set with parameter W1 for the first layer of the graph structure and parameter W2 for the second layer of the graph structure.

[0156] At this point, the first-layer graph structure is input into the merchant target location vector model, resulting in the first adjusted location vector corresponding to each merchant node in the first-layer graph structure output by the merchant target location model. For example... Figure 4 In step 401, obtain the first adjusted position vectors of merchant node A, merchant node B, merchant node C, merchant node D, and merchant node E in the near-field wireless node data layer graph structure.

[0157] Inputting the second-layer graph structure into the merchant target location vector model yields the second adjusted location vector corresponding to each merchant node in the second-layer graph structure output by the merchant target location vector model. For example... Figure 4In section 402, the second adjusted position vectors of merchant node A, merchant node B, merchant node C, merchant node D, and merchant node E are obtained from the historical behavior data layer graph structure of delivery resources.

[0158] Specifically, the first loss value between the first adjusted position vector and the second adjusted position vector of merchant node A is less than the first preset loss value. Based on the first adjusted position vector and the second adjusted position vector of merchant node A, the first merchant target position vector of merchant node A is obtained.

[0159] The above describes the training method for the merchant target location vector model.

[0160] This application provides a method for predicting the arrival time of delivery resources at a merchant, comprising: obtaining information about a first merchant to which a first delivery object to be picked up belongs; inputting the first merchant information into a merchant target location vector model to obtain a first merchant target location vector output by the merchant target location vector model, wherein the merchant target location vector model is used to obtain the first merchant target location vector based on the initial location vector corresponding to each layer of the graph structure in the multi-layer graph structure, the multi-layer graph structure including a near-field wireless node data layer graph structure and a historical behavior data layer graph structure of delivery resources; obtaining information about a second merchant to which a second delivery object to which the delivery resource has been picked up belongs, and obtaining a second merchant target location vector corresponding to the second merchant information; inputting the second merchant target location vector and the first merchant target location vector as input data into a duration prediction network model to obtain first predicted duration information output by the duration prediction network model, representing the arrival time of the delivery resource from the location of the second merchant to the location of the first merchant.

[0161] In the above method, the first delivery target and the corresponding first merchant information are determined. Then, the second merchant information corresponding to the second delivery target that was already picked up before the delivery resource was ready to pick up the first delivery target is obtained. The first merchant target location vector and the second merchant target location vector are obtained through a merchant target location vector model. In this process, the merchant target location vector is calculated based on the initial location vector of the merchant in each layer of the multi-layer graph structure, which includes a near-field wireless node data layer graph structure and a historical behavior data layer graph structure of the delivery resource. This multi-layer graph structure improves the accuracy of determining the merchant target location vector and also improves the prediction accuracy of the delivery resource arriving at the merchant's location. Based on this, a duration prediction network model is used to obtain the first prediction duration information between the first and second merchant target location vectors, improving the prediction of the delivery resource's arrival time at the target merchant.

[0162] First Embodiment

[0163] The first embodiment of this application provides a method for predicting the arrival time of delivery resources at merchants, the specific process of which is as follows: Figure 6 As shown, it is a flowchart of a method for predicting the arrival time of delivery resources at merchants provided in the first embodiment of this application. Figure 6 The predicted arrival time of delivery resources for merchants is shown in steps S601 to S604.

[0164] like Figure 6 As shown, in step S601, the information of the first merchant to which the first object to be delivered belongs is obtained.

[0165] This step is used to obtain the first object to be delivered that the delivery resources are ready to pick up, so as to determine the merchant information of the first merchant to which the delivery resources are to be delivered, and thus determine the location vector of the first merchant in subsequent steps.

[0166] The step of obtaining the first merchant information to which the first delivery object belongs can include: a first prompt message sent by the delivery resource terminal used to obtain the delivery resource to prompt the delivery resource to trigger the delivery operation of the first delivery object; and obtaining the first merchant information to which the first delivery object belongs based on the first prompt message and the first delivery object information.

[0167] like Figure 6As shown, in step S602, the first merchant information is input into the merchant target location vector model to obtain the first merchant target location vector corresponding to the first merchant information output by the merchant target location vector model. The merchant target location vector model is used to obtain the first merchant target location vector based on the initial location vector corresponding to each layer of the multi-layer graph structure of the first merchant. The multi-layer graph structure includes a near-field wireless node data layer graph structure and a historical behavior data layer graph structure of delivery resources.

[0168] This step is used to determine the target location vector of the first merchant based on the first merchant information. This determination is achieved through a merchant target location vector model. The merchant target location vector model is obtained by training the initial merchant location vector. The specific training process can be obtained from the training method of the merchant target location vector model in the application scenario embodiment.

[0169] The step of inputting the first merchant information into the merchant target location vector model to obtain the first merchant target location vector corresponding to the first merchant information output by the merchant target location vector model includes: obtaining the first initial location vector of the first merchant corresponding to the first merchant information in the near-field wireless node data layer graph structure according to the first merchant information; obtaining the second initial location vector of the first merchant corresponding to the first merchant information in the historical behavior data layer graph structure of the delivery resources; calculating the first loss value between the first initial location vector and the second initial location vector; and obtaining the first merchant target location vector according to the first initial location vector, the second initial location vector, and the first loss value.

[0170] The step of obtaining the first merchant target location vector according to the first initial location vector, the second initial location vector, and the first loss value using a preset calculation method includes:

[0171] Based on the first initial position vector and the first loss value, a first adjusted position vector of the first merchant in the near-field wireless node data layer graph structure is obtained; based on the second initial position vector and the first loss value, a second adjusted position vector of the first merchant in the historical behavior data layer graph structure of delivery resources is obtained; based on the first adjusted position vector and the second adjusted position vector, the target position vector of the first merchant is obtained.

[0172] The above describes a method for determining the target location vector of the first merchant when the multi-layer graph structure is a two-layer graph structure. Furthermore, the multi-layer graph structure can also be a three-layer graph structure, including: a near-field wireless node data layer graph structure, a wireless LAN data layer graph structure, and a historical behavior data layer graph structure for delivery resources.

[0173] The step of inputting the first merchant information into the merchant target location vector model to obtain the first merchant target location vector corresponding to the first merchant information output by the merchant target location vector model includes: obtaining, based on the first merchant information, a first initial location vector of the first merchant corresponding to the first merchant information in the near-field wireless node data layer graph structure; obtaining, based on the first merchant information, a second initial location vector of the first merchant corresponding to the first merchant information in the historical behavior data layer graph structure of the delivery resources; obtaining, based on the first merchant information, a third initial location vector of the first merchant corresponding to the first merchant information in the wireless local area network data layer graph structure; calculating, based on the first initial location vector, the second initial location vector, and the third initial location vector, a second loss value is calculated; and based on the first initial location vector, the second initial location vector, the third initial location vector, and the second loss value, the first merchant target location vector is obtained.

[0174] The step of obtaining the first merchant target location vector based on the first initial location vector, the second initial location vector, the third initial location vector, and the second loss value includes:

[0175] Based on the first initial position vector and the second loss value, a third adjusted position vector of the first merchant in the near-field wireless node data layer graph structure is obtained; based on the second initial position vector and the second loss value, a fourth adjusted position vector of the first merchant in the historical behavior data layer graph structure of delivery resources is obtained; based on the third initial position vector and the second loss value, a fifth adjusted position vector of the first merchant in the historical behavior data layer graph structure of delivery resources is obtained; based on the third adjusted position vector, the fourth adjusted position vector, and the fifth adjusted position vector, the target position vector of the first merchant is obtained.

[0176] like Figure 6 As shown, in step S603, the second merchant information to which the second object to be delivered belongs is obtained, and the second merchant target location vector corresponding to the second merchant information is obtained.

[0177] After obtaining the message of the first delivery resource to be picked up based on step S601, in order to calculate the time for the delivery resource to arrive at the location of the first merchant, it is necessary to obtain the information of the current location of the delivery resource. Based on the merchant information of the delivery resource that has completed the task of picking up the delivery object, the location information of the second merchant is determined, and the location information of the second merchant is used as the information of the current location of the delivery resource.

[0178] Based on the above step S601, after the server obtains the first prompt information, it obtains the information of the first object to be delivered that is waiting to be picked up by the delivery resources.

[0179] Based on the first notification message, query the second delivery object that has been picked up within a preset time range; and determine the second merchant information corresponding to the second delivery object based on the second delivery object.

[0180] The step of obtaining the second merchant target location vector corresponding to the second merchant information includes: inputting the second merchant information into the merchant target location vector model to obtain the second merchant target location vector corresponding to the second merchant information output by the merchant target location vector model. The merchant target location vector model is used to obtain the second target merchant location vector based on the initial location vector corresponding to each layer of the multi-layer graph structure of the second merchant. The multi-layer graph structure includes a near-field wireless node data layer graph structure and a historical behavior data layer graph structure of delivery resources.

[0181] The step of inputting the second merchant information into the merchant target location vector model to obtain the second merchant target location vector corresponding to the second merchant information output by the merchant target location vector model includes:

[0182] Based on the second merchant information, obtain the fourth initial position vector of the second merchant in the near-field wireless node data layer graph structure corresponding to the second merchant information, and obtain the fifth initial position vector of the second merchant in the historical behavior data layer graph structure of the delivery resources; calculate the third loss value between the fourth initial position vector and the fifth initial position vector; and obtain the target position vector of the second merchant according to a preset calculation method based on the fourth initial position vector, the fifth initial position vector, and the third loss value.

[0183] like Figure 6 As shown, in step S604, the second merchant target location vector and the first merchant target location vector are used as input data and input into the duration prediction network model to obtain the first predicted duration information output by the duration prediction network model, which represents the delivery resource from the location of the second merchant to the location of the first merchant.

[0184] This step is used to obtain the predicted time information for the delivery resources to travel from the location of the second merchant to the location of the first merchant. The time prediction network model can be a Multi-Layer Perception (MLP) neural network model. The input data of this model are the target location vectors of the second merchant and the first merchant, and the output data is the first predicted time information.

[0185] In addition, the method also includes determining whether the delivery resource has reached the location of the first merchant by: determining the remaining time information of the delivery resource at the current time from the location of the first merchant; if the remaining time information is less than a first time information threshold, then determining that the delivery resource has reached the location of the first merchant; and sending a second notification message to the first merchant terminal to indicate that the delivery resource has reached the location of the first merchant at the current time.

[0186] The determination of the remaining time information of the delivery resource from the location of the first merchant at the current moment includes:

[0187] The system detects that the delivery resource receives a third wireless signal from a third merchant at the current moment. Based on the third wireless signal, it obtains the third merchant information corresponding to the third merchant, wherein the third merchant is at least one merchant that the delivery resource passes through on its journey to the location of the first merchant. The third merchant information is input into the merchant target location vector model to obtain the third merchant target location vector output by the merchant target location vector model. Based on the first merchant target location vector and the third merchant target location vector, the system determines the remaining time information of the delivery resource at the current moment from the location of the first merchant.

[0188] This application provides a method for predicting the arrival time of delivery resources at a merchant, comprising: obtaining information about a first merchant to which a first delivery object to be picked up belongs; inputting the first merchant information into a merchant target location vector model to obtain a first merchant target location vector output by the merchant target location vector model, wherein the merchant target location vector model is used to obtain the first merchant target location vector based on the initial location vector corresponding to each layer of the graph structure in the multi-layer graph structure, the multi-layer graph structure including a near-field wireless node data layer graph structure and a historical behavior data layer graph structure of delivery resources; obtaining information about a second merchant to which a second delivery object to which the delivery resource has been picked up belongs, and obtaining a second merchant target location vector corresponding to the second merchant information; inputting the second merchant target location vector and the first merchant target location vector as input data into a duration prediction network model to obtain first predicted duration information output by the duration prediction network model, representing the arrival time of the delivery resource from the location of the second merchant to the location of the first merchant.

[0189] In the above method, the first delivery target and the corresponding first merchant information are determined. Then, the second merchant information corresponding to the second delivery target that the delivery resource has already picked up before picking up the first delivery target is obtained. The first merchant target location vector and the second merchant target location vector are obtained through a merchant target location vector model. In this process, the merchant target location vector is calculated based on the initial location vector of the merchant in each layer of the multi-layer graph structure, which includes a near-field wireless node data layer graph structure and a historical behavior data layer graph structure of the delivery resource. This multi-layer graph structure improves the accuracy of the merchant target location vector and also improves the prediction accuracy of the delivery resource arriving at the merchant's location. Based on this, a duration prediction network model is used to obtain the first prediction duration information between the first and second merchant target location vectors, improving the prediction of the delivery resource's arrival time at the target merchant.

[0190] Second Embodiment

[0191] The second embodiment of this application provides a method for predicting the target location of a merchant, the specific process of which is as follows: Figure 7 As shown, it is a flowchart of a method for predicting the target location of a merchant provided in the second embodiment of this application. Figure 7 The predicted merchant target location shown includes steps S701 to S702.

[0192] like Figure 7As shown, in step S701, merchant information used to characterize merchant attributes is obtained.

[0193] This step is used to obtain merchant information so that the next step can determine the merchant's target location information based on the merchant information. Here, the merchant information that represents the merchant's attributes includes the merchant's name, the name of the shopping mall where the merchant is located, or the area information of the region where the merchant is located, as well as the merchant's Wi-Fi information, near-field wireless signal (e.g., Bluetooth signal), and historical behavioral data of delivery resources arriving at the target merchant in historical periods, etc.

[0194] like Figure 7 As shown, in step S702, the merchant information is input into the merchant target location vector model to obtain the merchant target location vector corresponding to the merchant information output by the merchant target location vector model. The merchant target location vector model is used to obtain the merchant target location vector based on the initial location vector corresponding to each layer of the multi-layer graph structure. The multi-layer graph structure includes a near-field wireless node layer graph structure and a historical behavior data layer graph structure of delivery resources.

[0195] This step is used to obtain the merchant's target location vector based on the merchant's target location vector model. The merchant's target location vector model is obtained by training the merchant's initial location vector model. Specifically, the merchant's target location vector model obtains the merchant's target location vector based on the initial location vector corresponding to the merchant in each layer of the multi-layer graph structure. The construction method of the multi-layer graph structure and the training method of the merchant's target location vector model can be found in the detailed description of these two parts in the application scenario embodiment.

[0196] This method provides users with the target location information of merchants by obtaining the trained merchant target location vector model, thereby improving the accuracy of obtaining merchant location information.

[0197] In addition, the delivery process includes the following steps: the delivery resources arrive at the merchant to pick up the items to be delivered, and after picking up the items, the delivery resources deliver the items to the user's location.

[0198] In the process of delivery resources picking up the items to be delivered from the merchant, the delivery resources need to obtain the location information of the items to be delivered. In order to improve the accuracy of determining the location information of the items to be delivered, the items to be delivered can be set at the pick-up point at the merchant's entrance, which is conducive to the delivery resources reaching the merchant to pick up the items to be delivered.

[0199] The method further includes: obtaining a first request message sent by the delivery resource terminal used by the delivery resource to request the location of the object to be delivered; obtaining merchant information to which the object to be delivered belongs based on the first request message, the merchant information including merchant identification information of the merchant to which the object to be delivered belongs and delivery information of the object to be delivered; and inputting the merchant information into a merchant target location vector model to obtain a merchant target location vector corresponding to the merchant information output by the merchant target location vector model, including: inputting the merchant identification information and the delivery information of the object to be delivered by the merchant as input data into the merchant target location vector model to obtain a merchant target location vector corresponding to the merchant information output by the merchant target location vector model, the merchant target location vector being a location vector used to represent the location of the delivery resource picking up the object to be delivered.

[0200] In addition, when users are looking for the store of a target merchant in an indoor shopping mall environment, the target location of the merchant can be determined by the merchant target location vector model. After obtaining the target location of the target merchant, the user is provided with a guide route to the location of the merchant, so as to avoid the user sitting in the wrong place or walking too much.

[0201] The method further includes: obtaining a second request message sent by a user terminal for requesting merchant target location information to reach the merchant; obtaining a first user location vector representing the user's location based on the second request message; and determining first path planning information for the user to reach the merchant target location vector based on the first user location vector and the merchant target location vector, wherein the merchant target location vector is the doorway location vector of the merchant.

[0202] This application provides a method for predicting a merchant's target location, comprising: acquiring merchant information to characterize merchant attributes; inputting the merchant information into a merchant target location vector model to obtain a merchant target location vector output by the merchant target location vector model corresponding to the merchant information; wherein, the merchant target location vector model is used to obtain the merchant target location vector based on the initial location vector corresponding to each layer of the graph structure in the multi-layer graph structure, the multi-layer graph structure including a near-field wireless node layer graph structure and a historical behavior data layer graph structure of delivery resources.

[0203] In the above method, based on merchant information, a merchant target location vector model is used to determine the merchant's target location vector. This is because the merchant target location vector model is based on the merchant's initial location vector in the near-field wireless node layer graph structure and its initial location vector in the historical behavior data layer graph structure of delivery resources. In other words, determining the merchant's target location vector references the merchant's near-field wireless node signals and historical behavior data, thus improving the accuracy of determining the merchant's target location.

[0204] Third Embodiment

[0205] The third embodiment of this application provides a method for constructing a multi-layer spectral structure, the specific process of which is as follows: Figure 8 As shown, it is a flowchart of a method for constructing a multi-layer map structure provided in the third embodiment of this application. Figure 8 The construction of the multi-layer map structure shown includes steps S801 to S804.

[0206] like Figure 8 As shown, in step S801, based on the near-field wireless node signals of the target merchant, a near-field wireless node data layer map structure is constructed to characterize the connection relationships between the target merchant and other merchants within the target area. The method for constructing the near-field wireless node data layer map structure in this step can be referred to the description in the application scenario embodiment, and will not be repeated here. Figure 8 As shown, in step S802, based on the wireless LAN data information searched by the target merchant, a wireless LAN data layer graph structure is constructed to characterize the connection relationships between the target merchant and other merchants within the target area. The method for constructing the wireless LAN data layer graph structure in this step can be referred to the description in the application scenario embodiment, and will not be repeated here. Figure 8 As shown, in step S803, historical behavior data of delivery resources within a first preset time period is acquired, and a historical behavior data layer graph structure of delivery resources is constructed to characterize the connection relationship between the target merchant and other merchants within the target area. The method for constructing the historical behavior data layer graph structure of delivery resources in this step can be referred to the description in the application scenario embodiment, and will not be repeated here. Figure 8 As shown, in step S804, a multi-layer graph structure is constructed to characterize the connection relationship between the target merchant and other merchants in the target area, based on the near-field wireless node data layer graph structure, the wireless local area network data layer graph structure, and the historical behavior data layer graph structure of the delivery resources.

[0207] It should be noted that the multi-layer graph structure includes a weight value for representing the connection relationship between the target merchant and other merchants, as well as a weight value for representing the connection relationship between the target merchant and other merchants.

[0208] The method further includes: sorting multiple weight values ​​used to characterize the connection relationship between the target merchant and other merchants; determining a target weight value according to the sorting order of the weight values; and using the target merchant corresponding to the target weight value and other merchants as merchants recommended to the user.

[0209] Based on the connection relationships between merchants in each layer of the multi-layer graph structure and the weight values ​​of these connections, the selection rate of each merchant by users is determined. Thus, the target weight value can be determined if the weight value is greater than the preset weight range. Merchants corresponding to the target weight value are then recommended to users, thereby increasing the selection rate of merchants in the target area.

[0210] This application provides a method for constructing a multi-layer graph structure, comprising: constructing a near-field wireless node data layer graph structure representing the connection relationship between the target merchant and other merchants within a target area based on the near-field wireless node signals of the target merchant; constructing a wireless local area network (WLAN) data layer graph structure representing the connection relationship between the target merchant and other merchants within a target area based on the WLAN data information searched by the target merchant; acquiring historical behavior data of delivery resources within a first preset time period, and constructing a historical behavior data layer graph structure of delivery resources representing the connection relationship between the target merchant and other merchants within a target area; and constructing a multi-layer graph structure representing the connection relationship between the target merchant and other merchants within a target area based on the near-field wireless node data layer graph structure, the WLAN data layer graph structure, and the historical behavior data layer graph structure of delivery resources.

[0211] The above method constructs a multi-layered graph structure to represent the connection relationship between the target merchant and other merchants in multiple dimensions. Based on the connection relationship between merchants in the multi-layered graph structure, the selection rate of the target merchant is determined, thereby improving the selection rate of the target merchant in the target area.

[0212] Fourth embodiment

[0213] The fourth embodiment of this application provides a training method for a merchant target location vector model, the specific process of which is as follows: Figure 9 As shown, it is a flowchart of a training method for a merchant target location vector model provided in the fourth embodiment of this application. Figure 9 The training method for the merchant target location vector model shown includes steps S901 to S905.

[0214] like Figure 9 As shown, in step S901, a multi-layer graph structure is constructed to characterize the location of the target merchant within the target area where the delivery resources arrive. This multi-layer graph structure includes a near-field wireless node data layer graph structure and a historical behavior data layer graph structure for the delivery resources. For example... Figure 9 As shown, in step S902, the near-field wireless node data layer map structure is input into the merchant initial location vector model to obtain the first initial location vector of the target merchant in the near-field wireless node data layer map structure output by the merchant initial location vector model. Figure 9 As shown, in step S903, the historical behavior data layer graph structure of the delivery resources is input into the merchant initial location vector model to obtain the second initial location vector of the target merchant in the historical behavior data layer graph structure of the delivery resources, output by the merchant initial location vector model. Figure 9 As shown, in step S904, a first loss value is calculated between the first initial position vector of the target merchant and the second initial position vector of the target merchant. Figure 9 As shown, in step S905, the merchant initial location vector model is trained based on the first loss value to obtain the merchant target location vector model. The merchant target location vector model is used to obtain the merchant target location vector of the target merchant in the target area.

[0215] The construction of a multi-layer graph structure to characterize the location of the target merchant within the target area where the delivery resource arrives includes: constructing a near-field wireless node data layer graph structure based on the wireless signals received by the delivery resource from the merchant; and constructing a historical behavior data layer graph structure based on the time information of the delivery resource from the location of the first merchant to the location of the second merchant.

[0216] The step of constructing a near-field wireless node data layer graph structure based on the wireless signals received by the delivery resources from merchants includes: acquiring multiple wireless signal groups received by the delivery resources within a first preset time period, wherein each wireless signal group refers to a wireless signal received by the delivery resources from two merchants within a second preset time period; calculating a first weight value for each wireless signal group in the multiple wireless signal groups; determining a first connection relationship between the two merchants in each wireless signal group based on the first weight value corresponding to each wireless signal group; and constructing a near-field wireless node data layer graph structure based on the first connection relationship between the two merchants in each wireless signal group.

[0217] The wireless signal group is obtained through the following method:

[0218] The delivery resource receives a first wireless signal from a first merchant and a second wireless signal from a second merchant within a second preset time period; the first wireless signal and the second wireless signal received by the delivery resource within the second preset time period are taken as the first wireless signal group received by the delivery resource within the second preset time period.

[0219] The step of determining the first connection relationship between two merchants in each wireless signal group based on the first weight value corresponding to each wireless signal group includes: determining the first connection distance relationship between two merchants in each wireless signal group based on the first weight value corresponding to each wireless signal group, wherein the first weight value and the first connection distance relationship have a negative correlation.

[0220] The step of constructing a historical behavior data layer graph structure for the delivery resources based on the time information of the delivery resources traveling from the location of the first merchant to the location of the second merchant includes:

[0221] The time information of the delivery resource from the location of the first merchant to the location of the second merchant is obtained, and the process of the delivery resource from the location of the first merchant to the location of the second merchant is taken as the first behavior data group of the delivery resource; according to the time information, the second weight value corresponding to the first behavior data group of the delivery resource is obtained; according to the second weight value corresponding to the first behavior data group, the historical behavior data layer graph structure of the delivery resource is constructed.

[0222] The step of constructing the historical behavior data layer graph structure of the delivery resources based on the second weight value corresponding to the first behavior data group includes:

[0223] Based on the second weight value, a second connection distance relationship between the first merchant and the second merchant in the first behavior data group is determined, and the second weight value and the second connection distance relationship are positively correlated; based on the second connection distance relationship between the first merchant and the second merchant, a historical behavior data layer graph structure of the delivery resources is constructed.

[0224] The step of training the merchant's initial location vector model based on the first loss value to obtain the merchant's target location vector model includes: adjusting the parameters in the trained merchant's initial location vector model according to the first loss value.

[0225] In addition, the method further includes: if the first loss value is less than a first preset loss value threshold, then stop training the merchant initial location vector model; or, if the first loss value is greater than the first preset loss value threshold and the difference between multiple first loss values ​​is less than a first preset difference threshold, then stop training the merchant initial location vector model.

[0226] The multi-layer graph structure further includes a wireless local area network data layer graph structure; the method further includes:

[0227] The wireless LAN data layer graph structure is input into the merchant initial location vector model to obtain the third initial location vector of the target merchant in the wireless LAN data layer graph structure output by the merchant initial location vector model; a second loss value is calculated between the first initial location vector, the second initial location vector, and the third initial location vector of the target merchant; the merchant initial location vector model is trained based on the second loss value to obtain the merchant target location vector model.

[0228] The construction of a multi-layer graph structure to characterize the location of a target merchant within the target area where the delivery resource arrives includes: constructing a near-field wireless node data layer graph structure based on the wireless signals received by the delivery resource from the merchant; constructing a historical behavior data layer graph structure based on the time information of the delivery resource from the location of the first merchant to the location of the second merchant; and constructing a wireless local area network (WLAN) data layer graph structure based on the WLAN data information searched by the target merchant.

[0229] The step of constructing a wireless LAN data layer graph structure based on the wireless LAN data information searched by the target merchant includes: obtaining a list of wireless LAN data information searched by the target merchant, the list of wireless LAN data information including wireless LAN data information of multiple other merchants; determining a third connection distance relationship between the target merchant and other merchants based on the wireless LAN signal strength values ​​received from other merchants in the list of wireless LAN data information; and constructing a wireless LAN data layer graph structure based on the third connection distance relationship between the target merchant and other merchants.

[0230] Fifth Embodiment

[0231] The fifth embodiment of this application provides a method for predicting the arrival of delivery resources at a target merchant, the specific process of which is as follows: Figure 10 As shown, it is a flowchart of a method for predicting the arrival of delivery resources at a target merchant provided in the fifth embodiment of this application. Figure 10The method shown for predicting the arrival of delivery resources at the target merchant includes steps S1001 to S1005.

[0232] like Figure 10 As shown, in step S1001, the information of the fourth merchant to which the fourth object to be delivered belongs, which has already received the delivery resource at the current time, is obtained; for example... Figure 10 As shown, in step S1002, the fourth merchant information is input into the merchant target location vector model to obtain the fourth merchant target location vector corresponding to the fourth merchant information output by the merchant target location vector model. The merchant target location vector model is used to obtain the fourth merchant target location vector based on the initial location vector corresponding to each layer of the multi-layer graph structure. The multi-layer graph structure includes a near-field wireless node data layer graph structure and a historical behavior data layer graph structure of delivery resources. Figure 10 As shown, in step S1003, the fifth merchant information to which the fifth delivery object to be picked up belongs is obtained, and the target location vector of the fifth merchant corresponding to the fifth merchant information is obtained; as shown Figure 10 As shown, in step S1004, the target location vectors of the fourth merchant and the fifth merchant are used as input data and input into the duration prediction network model to obtain the second predicted duration information output by the duration prediction network model, which represents the delivery resource's journey from the location of the fourth merchant to the location of the fifth merchant; as shown Figure 10 As shown, in step S1005, if the second predicted duration information meets the preset duration review conditions, then the fifth merchant is determined to be the target merchant that the delivery resources need to reach after leaving the fourth merchant.

[0233] Optionally, the fifth merchant information includes at least one fifth candidate merchant information; the method further includes: for each fifth candidate merchant information in the at least one fifth candidate merchant information, obtaining second candidate predicted time information of the delivery resource from the location of the fourth merchant to the location of the fifth candidate merchant; the second predicted time information meets preset time review conditions, including: sorting multiple second candidate predicted time information according to the sorting rule of time from smallest to largest, to obtain second target predicted time information, wherein the second target predicted time information is less than other second candidate predicted time information in the multiple second candidate predicted time information.

[0234] Optionally, the step of inputting the fourth merchant information into the merchant target location vector model to obtain the fourth merchant target location vector corresponding to the fourth merchant information output by the merchant target location vector model includes: obtaining the sixth initial location vector of the fourth merchant corresponding to the fourth merchant information in the near-field wireless node data layer graph structure, and obtaining the seventh initial location vector of the fourth merchant corresponding to the fourth merchant information in the historical behavior data layer graph structure of the delivery resources; calculating the fourth loss value between the sixth initial location vector and the seventh initial location vector; and obtaining the fourth merchant target location vector according to a preset calculation method based on the sixth initial location vector, the seventh initial location vector, and the fourth loss value.

[0235] Optionally, obtaining the fourth merchant information to which the fourth delivery object to which the delivery resource has been claimed at the current time belongs includes: obtaining a third notification message sent by the fourth merchant terminal used by the fourth merchant, indicating that the delivery resource has been successfully claimed at the current time; and obtaining the fourth merchant information to which the fourth delivery object to which the delivery resource has been claimed at the current time belongs based on the third notification message.

[0236] Optionally, obtaining the fifth merchant target location vector corresponding to the fifth merchant information includes: inputting the fifth merchant information into the merchant target location vector model to obtain the fifth merchant target location vector output by the merchant target location vector model corresponding to the fifth merchant information; wherein, the merchant target location vector model is used to query the fifth merchant target location vector corresponding to the fifth merchant information in a pre-stored correspondence table between merchant information and merchant target location vectors based on the fifth merchant information, and the fifth merchant target location vector is obtained based on the initial location vector corresponding to each layer of the graph structure in the multi-layer graph structure, wherein the multi-layer graph structure includes a near-field wireless node data layer graph structure and a historical behavior data layer graph structure of delivery resources.

[0237] This application provides a method for predicting the arrival of delivery resources at a target merchant, comprising: obtaining information about a fourth merchant to which a fourth delivery object already picked up at the current time belongs; inputting the fourth merchant information into a merchant target location vector model to obtain a fourth merchant target location vector output by the merchant target location vector model corresponding to the fourth merchant information, wherein the merchant target location vector model is used to obtain the fourth merchant target location vector based on the initial location vector corresponding to each layer of the graph structure in the multi-layer graph structure, the multi-layer graph structure including a near-field wireless node data layer graph structure and historical behavior data of delivery resources. According to the layer graph structure; obtain the information of the fifth merchant to which the fifth delivery object to be picked up belongs, and obtain the target location vector of the fifth merchant corresponding to the fifth merchant information; input the target location vector of the fourth merchant and the target location vector of the fifth merchant as input data into the duration prediction network model, and obtain the second predicted duration information output by the duration prediction network model to represent the delivery resource from the location of the fourth merchant to the location of the fifth merchant; if the second predicted duration information meets the preset duration review conditions, then determine the fifth merchant as the target merchant that the delivery resource needs to reach after leaving the fourth merchant.

[0238] The above method, after the delivery resource has received the information that the delivery resource has completed receiving the fourth delivery object, allows the server to determine the next delivery object to be received. Specifically, after determining the location information of the fourth merchant through a merchant target location vector model, and then determining the location information of the merchants corresponding to the multiple delivery objects to be received by the delivery resource, the distance between the location of the fourth merchant and the location of the fifth merchant is calculated. Then, a second predicted time information is determined for the delivery resource to travel from the location of the fourth merchant to the location of the fifth merchant. After judging the second predicted time information, the target merchant that the delivery resource needs to reach after receiving the fourth delivery object is determined. This method can prioritize matching the delivery resource with target merchants and receive target delivery objects from them, improving the efficiency of the delivery resource in receiving delivery objects.

[0239] Sixth Embodiment

[0240] Corresponding to the application scenario of the method for predicting the arrival time of delivery resources at merchants provided in this application, and the method for predicting the arrival time of delivery resources at merchants provided in the first embodiment, the sixth embodiment of this application provides an apparatus for predicting the arrival time of delivery resources at merchants. For example... Figure 11The diagram shown is a schematic representation of an apparatus for predicting the arrival time of delivery resources at merchants, according to a sixth embodiment of this application. Since the apparatus embodiment is fundamentally similar to the method embodiment, it is described simply; relevant details can be found in the description of the method embodiment. The apparatus embodiment described below is merely illustrative.

[0241] The sixth embodiment of this application provides an apparatus for predicting the arrival time of delivery resources at merchants. The apparatus includes: a first acquisition unit 1101, used to acquire first merchant information to which a first delivery object to be picked up belongs; and a first merchant target location vector acquisition unit 1102, used to input the first merchant information into a merchant target location vector model to obtain a first merchant target location vector corresponding to the first merchant information output by the merchant target location vector model. The merchant target location vector model is used to obtain the first merchant target location vector based on the initial location vector corresponding to each layer of the multi-layer graph structure in which the first merchant is located. The system includes a near-field wireless node data layer graph structure and a historical behavior data layer graph structure for delivery resources; a second merchant target location vector acquisition unit 1103, used to acquire the second merchant information to which the second delivery object to be delivered belongs, and acquire the second merchant target location vector corresponding to the second merchant information; and a first prediction duration information acquisition unit 1104, used to input the second merchant target location vector and the first merchant target location vector as input data into the duration prediction network model, and acquire the first prediction duration information output by the duration prediction network model, which represents the delivery resource from the location of the second merchant to the location of the first merchant.

[0242] Seventh Embodiment

[0243] Corresponding to the method for predicting merchant target locations provided in the second embodiment of this application, the seventh embodiment of this application provides an apparatus for predicting merchant target locations. For example... Figure 12 The diagram shown is a schematic representation of an apparatus for predicting the target location of a merchant according to a seventh embodiment of this application. Since the apparatus embodiment is basically similar to the method embodiment, it is described simply; relevant details can be found in the description of the method embodiment. The apparatus embodiment described below is merely illustrative.

[0244] The seventh embodiment of this application provides an apparatus for predicting the target location of a merchant. The apparatus includes: a second acquisition unit 1201, used to acquire merchant information used to characterize merchant attributes; and a third acquisition unit 1202, used to input the merchant information into a merchant target location vector model to obtain a merchant target location vector corresponding to the merchant information output by the merchant target location vector model. The merchant target location vector model is used to obtain the merchant target location vector based on the initial location vector corresponding to each layer of the multi-layer graph structure. The multi-layer graph structure includes a near-field wireless node layer graph structure and a historical behavior data layer graph structure for delivery resources.

[0245] Eighth embodiment

[0246] Corresponding to the method for constructing a multi-layered spectral structure provided in the third embodiment of this application, the eighth embodiment of this application provides an apparatus for constructing a multi-layered spectral structure. For example... Figure 13 The diagram shown is a schematic representation of an apparatus for constructing a multilayer map structure according to the eighth embodiment of this application. Since the apparatus embodiment is basically similar to the method embodiment, the description is relatively simple; relevant details can be found in the description of the method embodiment. The apparatus embodiment described below is merely illustrative.

[0247] The eighth embodiment of this application provides an apparatus for constructing a multi-layer map structure. The apparatus includes: a first construction unit 1301, configured to construct a near-field wireless node data layer map structure representing the connection relationship between the target merchant and other merchants within a target area based on the near-field wireless node signal of the target merchant; a second construction unit 1302, configured to construct a wireless local area network data layer map structure representing the connection relationship between the target merchant and other merchants within a target area based on the wireless local area network data information searched by the target merchant; a third construction unit 1303, configured to acquire historical behavior data of delivery resources within a first preset time period and construct a historical behavior data layer map structure of delivery resources representing the connection relationship between the target merchant and other merchants within a target area; and a fourth construction unit 1304, configured to construct a multi-layer map structure representing the connection relationship between the target merchant and other merchants within a target area based on the near-field wireless node data layer map structure, the wireless local area network data layer map structure, and the historical behavior data layer map structure of delivery resources.

[0248] Ninth Embodiment

[0249] Corresponding to the training method for the merchant target location vector model provided in the fourth embodiment of this application, the ninth embodiment of this application provides a training apparatus for the merchant target location vector model. For example... Figure 14The diagram shown is a schematic representation of a training apparatus for a merchant target location vector model according to the ninth embodiment of this application. Since the apparatus embodiment is basically similar to the method embodiment, the description is relatively simple; relevant details can be found in the description of the method embodiment. The apparatus embodiment described below is merely illustrative.

[0250] The ninth embodiment of this application provides a training device for a merchant target location vector model, the device comprising: a fifth construction unit 1401, configured to construct a multi-layer graph structure representing the location of a target merchant within a target area where delivery resources arrive, the multi-layer graph structure including a near-field wireless node data layer graph structure and a historical behavior data layer graph structure of delivery resources; a first initial location vector model 1402, configured to input the near-field wireless node data layer graph structure into the merchant initial location vector model to obtain the first initial location vector of the target merchant in the near-field wireless node data layer graph structure output by the merchant initial location vector model; and a second initial location vector acquisition unit 1. 403 is used to input the historical behavior data layer graph structure of the delivery resources into the merchant initial position vector model to obtain the second initial position vector of the target merchant in the historical behavior data layer graph structure of the delivery resources output by the merchant initial position vector model; calculation unit 1404 is used to calculate the first loss value between the first initial position vector of the target merchant and the second initial position vector of the target merchant; training unit 1405 is used to train the merchant initial position vector model based on the first loss value to obtain the merchant target position vector model, the merchant target position vector model is used to obtain the merchant target position vector of the target merchant in the target area.

[0251] Tenth Embodiment

[0252] Corresponding to the method for predicting the arrival of delivery resources at target merchants provided in the fifth embodiment of this application, the ninth embodiment of this application provides an apparatus for predicting the arrival of delivery resources at target merchants. For example... Figure 15 The diagram shown is a schematic representation of an apparatus for predicting the arrival of delivery resources at a target merchant, according to a ninth embodiment of this application. Since the apparatus embodiment is fundamentally similar to the method embodiment, it is described simply; relevant details can be found in the description of the method embodiment. The apparatus embodiment described below is merely illustrative.

[0253] The ninth embodiment of this application provides an apparatus for predicting the arrival of delivery resources at a target merchant. The apparatus includes: a fourth acquisition unit 1501, used to acquire information about a fourth merchant to which a fourth delivery object already picked up at the current time belongs; a fourth merchant target location vector acquisition unit 1502, used to input the fourth merchant information into a merchant target location vector model to obtain a fourth merchant target location vector output by the merchant target location vector model corresponding to the fourth merchant information, wherein the merchant target location vector model is used to obtain the fourth merchant target location vector based on the initial location vector corresponding to each layer of the multi-layer graph structure, the multi-layer graph structure including a near-field wireless node data layer graph structure and a historical behavior data layer graph structure of delivery resources; and a fifth merchant target location vector acquisition unit 1503. The system is used to obtain the fifth merchant information to which the fifth delivery object to be picked up belongs, and to obtain the target location vector of the fifth merchant corresponding to the fifth merchant information; the second prediction duration information acquisition unit 1504 is used to input the target location vector of the fourth merchant and the target location vector of the fifth merchant as input data into the duration prediction network model, and obtain the second prediction duration information output by the duration prediction network model to represent the delivery resource from the location of the fourth merchant to the location of the fifth merchant; the review unit 1505 is used to determine that the fifth merchant is the target merchant that the delivery resource needs to reach in the next moment if the second prediction duration information meets the preset duration review conditions, and the second prediction duration information is the target duration information of the delivery resource from the location of the fourth merchant to the location of the fifth merchant.

[0254] Eleventh Embodiment

[0255] Corresponding to the methods of the first to fifth embodiments of this application, the eleventh embodiment of this application also provides an electronic device. For example... Figure 16 As shown, Figure 16This is a schematic diagram of an electronic device provided in the eleventh embodiment of this application. The electronic device includes: at least one processor 1601, at least one communication interface 1602, at least one memory 1603, and at least one communication bus 1604. Optionally, the communication interface 1602 can be an interface for a communication module, such as an interface for a GSM module. The processor 1601 may be a CPU, an Application Specific Integrated Circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present invention. The memory 1603 may include high-speed RAM and may also include non-volatile memory, such as at least one disk storage device. The memory 1603 stores a program, and the processor 1601 calls the program stored in the memory 1603 to execute the methods of the first to fifth embodiments of the present invention.

[0256] Twelfth Embodiment

[0257] Corresponding to the methods of the first to fifth embodiments of this application, the twelfth embodiment of this application also provides a computer storage medium. The computer storage medium stores a computer program, which is executed by a processor to perform the methods of the first to fifth embodiments.

[0258] Although this application discloses preferred embodiments as described above, it is not intended to limit this application. Any person skilled in the art can make possible changes and modifications without departing from the spirit and scope of this application. Therefore, the scope of protection of this application should be determined by the scope defined in the claims of this application.

[0259] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, a network interface, and memory. Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0260] 1. Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include non-transitory computer-readable media, such as modulated data signals and carrier waves.

[0261] 2. Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

Claims

1. A method for predicting the arrival time of delivery resources at merchants, characterized in that, include: Obtain the information of the first merchant to which the first recipient of the delivery resources belongs; The first merchant information is input into the merchant target location vector model to obtain the first merchant target location vector corresponding to the first merchant information output by the merchant target location vector model. The merchant target location vector model is used to obtain the first merchant target location vector based on the initial location vector corresponding to each layer of the multi-layer graph structure of the first merchant. The multi-layer graph structure includes a near-field wireless node data layer graph structure and a historical behavior data layer graph structure of delivery resources. Obtain the second merchant information to which the second object to be delivered already claimed by the delivery resource belongs, and obtain the second merchant target location vector corresponding to the second merchant information; The second merchant's target location vector and the first merchant's target location vector are used as input data and input into the duration prediction network model to obtain the first predicted duration information output by the duration prediction network model, which represents the delivery resource's journey from the location of the second merchant to the location of the first merchant.

2. The method according to claim 1, characterized in that, The step of inputting the first merchant information into the merchant target location vector model to obtain the first merchant target location vector corresponding to the first merchant information output by the merchant target location vector model includes: Based on the first merchant information, obtain the first initial position vector of the first merchant corresponding to the first merchant information in the near-field wireless node data layer graph structure, and obtain the second initial position vector of the first merchant corresponding to the first merchant information in the historical behavior data layer graph structure of the delivery resources. Calculate the first loss value between the first initial position vector and the second initial position vector; The first merchant target location vector is obtained based on the first initial location vector, the second initial location vector, and the first loss value.

3. The method according to claim 2, characterized in that, The step of obtaining the first merchant target location vector according to the first initial location vector, the second initial location vector, and the first loss value using a preset calculation method includes: Based on the first initial position vector and the first loss value, the first adjusted position vector of the first merchant in the near-field wireless node data layer map structure is obtained; Based on the second initial position vector and the first loss value, obtain the second adjusted position vector of the first merchant in the historical behavior data layer graph structure of the delivery resources; The first merchant target location vector is obtained based on the first adjusted location vector and the second adjusted location vector.

4. The method according to claim 1, characterized in that, The multi-layer graph structure also includes a wireless local area network data layer graph structure; The step of inputting the first merchant information into the merchant target location vector model to obtain the first merchant target location vector corresponding to the first merchant information output by the merchant target location vector model includes: Based on the first merchant information, obtain the first initial position vector of the first merchant corresponding to the first merchant information in the near-field wireless node data layer graph structure, obtain the second initial position vector of the first merchant corresponding to the first merchant information in the historical behavior data layer graph structure of the delivery resources, and obtain the third initial position vector of the first merchant corresponding to the first merchant information in the wireless local area network data layer graph structure. Calculate the second loss value among the first initial position vector, the second initial position vector, and the third initial position vector; The first merchant target location vector is obtained based on the first initial location vector, the second initial location vector, the third initial location vector, and the second loss value.

5. The method according to claim 4, characterized in that, The step of obtaining the first merchant target location vector based on the first initial location vector, the second initial location vector, the third initial location vector, and the second loss value includes: Based on the first initial position vector and the second loss value, the third adjusted position vector of the first merchant in the near-field wireless node data layer map structure is obtained; Based on the second initial position vector and the second loss value, obtain the fourth adjusted position vector of the first merchant in the historical behavior data layer graph structure of the delivery resources; Based on the third initial position vector and the second loss value, the fifth adjusted position vector of the first merchant in the historical behavior data layer graph structure of the delivery resources is obtained; The first merchant's target location vector is obtained based on the third adjusted location vector, the fourth adjusted location vector, and the fifth adjusted location vector.

6. The method according to claim 1, characterized in that, The process of obtaining the first merchant information to which the first recipient of the delivery resources belongs includes: The delivery resource terminal used to obtain delivery resources sends a first prompt message to prompt the delivery resource to trigger the pick-up operation of the first object to be delivered; Based on the first notification message, the first merchant information to which the first object to be delivered belongs is obtained according to the first object to be delivered information.

7. The method according to claim 6, characterized in that, The step of obtaining the second merchant information to which the second object to be delivered belongs, which has already claimed the delivery resources, includes: Based on the first notification message, query the second delivery object that has been picked up within a preset time range by the delivery resource; Based on the second object to be delivered, determine the second merchant information corresponding to the second object to be delivered.

8. The method according to claim 1, characterized in that, The step of obtaining the second merchant target location vector corresponding to the second merchant information includes: The second merchant information is input into the merchant target location vector model to obtain the second merchant target location vector corresponding to the second merchant information output by the merchant target location vector model. The merchant target location vector model is used to obtain the second target merchant location vector based on the initial location vector corresponding to each layer of the multi-layer graph structure of the second merchant. The multi-layer graph structure includes a near-field wireless node data layer graph structure and a historical behavior data layer graph structure of delivery resources.

9. The method according to claim 8, characterized in that, The step of inputting the second merchant information into the merchant target location vector model to obtain the second merchant target location vector corresponding to the second merchant information output by the merchant target location vector model includes: Based on the second merchant information, obtain the fourth initial position vector of the second merchant corresponding to the second merchant information in the near-field wireless node data layer graph structure, and obtain the fifth initial position vector of the second merchant corresponding to the second merchant information in the historical behavior data layer graph structure of the delivery resources. Calculate the third loss value between the fourth initial position vector and the fifth initial position vector; The second merchant target location vector is obtained according to the fourth initial location vector, the fifth initial location vector, and the third loss value, using a preset calculation method.

10. The method according to claim 1, characterized in that, Also includes: Determine the remaining time information of the delivery resource from the location of the first merchant at the current moment; If the remaining time information is less than the first time information threshold, then it is determined that the delivery resource has arrived at the location of the first merchant; Send a first notification message to the first merchant terminal to indicate that the delivery resources have arrived at the location of the first merchant at the current moment.

11. The method according to claim 10, characterized in that, The determination of the remaining time information of the delivery resource from the location of the first merchant at the current moment includes: The delivery resource is detected to receive a third wireless signal from a third merchant at the current moment. Based on the third wireless signal, the third merchant information corresponding to the third merchant is obtained. The third merchant is at least one merchant that the delivery resource passes through on its way to the location of the first merchant. The third merchant information is input into the merchant target location vector model to obtain the third merchant target location vector corresponding to the third merchant information output by the merchant target location vector model; Based on the target location vector of the first merchant and the target location vector of the third merchant, the remaining time information of the delivery resource at the current time relative to the location of the first merchant is determined.

12. A method for constructing a multi-layered spectral structure, characterized in that, include: Based on the near-field wireless node signals of the target merchant, a near-field wireless node data layer map structure is constructed to characterize the connection relationship between the target merchant and other merchants in the target area; Based on the wireless LAN data information searched by the target merchant, a wireless LAN data layer graph structure is constructed to represent the connection relationship between the target merchant and other merchants in the target area. Acquire historical behavior data of delivery resources within a first preset time period, and construct a historical behavior data layer graph structure of delivery resources to characterize the connection relationship between the target merchant and other merchants in the target area; Based on the near-field wireless node data layer graph structure, the wireless local area network data layer graph structure, and the historical behavior data layer graph structure of the delivery resources, a multi-layer graph structure is constructed to characterize the connection relationship between the target merchant and other merchants within the target area.

13. The method according to claim 12, characterized in that, The multi-layer graph structure includes a connection relationship between the target merchant and other merchants, and a weight value corresponding to the connection relationship between the target merchant and other merchants. The method further includes: The weight values ​​used to characterize the connection relationship between the target merchant and other merchants are sorted. Based on the sorting order of the weight values, a target weight value is determined, and the target merchant corresponding to the target weight value, along with other merchants, is selected as the merchants recommended to the user.

14. A training method for a merchant target location vector model, characterized in that, include: A multi-layer graph structure is constructed to characterize the location of target merchants within the target area where delivery resources arrive. The multi-layer graph structure includes a near-field wireless node data layer graph structure and a historical behavior data layer graph structure of delivery resources. The near-field wireless node data layer graph structure is input into the merchant initial position vector model to obtain the first initial position vector of the target merchant in the near-field wireless node data layer graph structure output by the merchant initial position vector model. Input the historical behavior data layer graph structure of the delivery resources into the merchant initial location vector model to obtain the second initial location vector of the target merchant in the historical behavior data layer graph structure of the delivery resources output by the merchant initial location vector model; Calculate the first loss value between the first initial position vector of the target merchant and the second initial position vector of the target merchant; Based on the first loss value, the merchant initial location vector model is trained to obtain the merchant target location vector model, which is used to obtain the merchant target location vector of the target merchant in the target area.

15. The method according to claim 14, characterized in that, The construction of a multi-layered graph structure to characterize the location of target merchants within the target area where delivery resources arrive includes: Based on the wireless signals received by the delivery resources from the merchants, a near-field wireless node data layer graph structure is constructed. Based on the time information of the delivery resources from the location of the first merchant to the location of the second merchant, a historical behavior data layer graph structure of the delivery resources is constructed.

16. The method according to claim 15, characterized in that, The step of constructing a near-field wireless node data layer graph structure based on the wireless signals received from merchants by the delivery resources includes: The delivery resource receives multiple wireless signal groups within a first preset time period, wherein each wireless signal group refers to the delivery resource receiving wireless signals from two merchants respectively within a second preset time period. Calculate the first weight value of each wireless signal group in the plurality of wireless signal groups; Based on the first weight value corresponding to each wireless signal group, the first connection relationship between the two merchants in each wireless signal group is determined; Based on the first connection relationship between two merchants in each wireless signal group, a near-field wireless node data layer graph structure is constructed.

17. The method according to claim 16, characterized in that, The wireless signal group is obtained through the following method: The delivery resources receive the first wireless signal from the first merchant and the second wireless signal from the second merchant within a second preset time period. The first wireless signal and the second wireless signal received by the delivery resource within a second preset time period are taken as the first wireless signal group received by the delivery resource within the second preset time period.

18. The method according to claim 16, characterized in that, The step of determining the first connection relationship between two merchants in each wireless signal group based on the first weight value corresponding to each wireless signal group includes: Based on the first weight value corresponding to each wireless signal group, a first connection distance relationship between two merchants in each wireless signal group is determined, wherein the first weight value and the first connection distance relationship are negatively correlated.

19. The method according to claim 15, characterized in that, The step of constructing a historical behavior data layer graph structure for the delivery resources based on the time information of the delivery resources traveling from the location of the first merchant to the location of the second merchant includes: Obtain the time information of the delivery resource from the location of the first merchant to the location of the second merchant, and take the process of the delivery resource from the location of the first merchant to the location of the second merchant as the first behavior data group of the delivery resource; Based on the duration information, obtain the second weight value corresponding to the first row data group of the delivery resource; Based on the second weight value corresponding to the first behavior data group, construct the historical behavior data layer graph structure of the delivery resources.

20. The method according to claim 19, characterized in that, The step of constructing the historical behavior data layer graph structure of the delivery resources based on the second weight value corresponding to the first behavior data group includes: Based on the second weight value, a second connection distance relationship between the first merchant and the second merchant in the first behavior data group is determined, and the second weight value and the second connection distance relationship are positively correlated. Based on the second connection distance relationship between the first merchant and the second merchant, a historical behavior data layer graph structure of the delivery resources is constructed.

21. The method according to claim 14, characterized in that, The step of training the merchant's initial location vector model based on the first loss value to obtain the merchant's target location vector model includes: Based on the first loss value, the parameters in the merchant initial location vector model during training are adjusted.

22. The method according to claim 14, characterized in that, Also includes: If the first loss value is less than the first preset loss value threshold, then training of the merchant's initial location vector model is stopped; or, If the first loss value is greater than the first preset loss value threshold, and the difference between multiple first loss values ​​is less than the first preset difference threshold, then training of the merchant's initial location vector model is stopped.

23. The method according to claim 14, characterized in that, The multi-layer graph structure also includes a wireless local area network data layer graph structure; The method further includes: Input the wireless LAN data layer graph structure into the merchant initial location vector model to obtain the third initial location vector of the target merchant in the wireless LAN data layer graph structure output by the merchant initial location vector model. Calculate a second loss value among the first initial position vector of the target merchant, the second initial position vector of the target merchant, and the third initial position vector of the target merchant; Based on the second loss value, the merchant's initial location vector model is trained to obtain the merchant's target location vector model.

24. The method according to claim 23, characterized in that, The construction of a multi-layered graph structure to characterize the location of target merchants within the target area where delivery resources arrive includes: Based on the wireless signals received by the delivery resources from the merchants, a near-field wireless node data layer graph structure is constructed. Based on the time information of the delivery resources from the location of the first merchant to the location of the second merchant, a historical behavior data layer graph structure of the delivery resources is constructed. Based on the wireless LAN data information retrieved from the target merchant, a wireless LAN data layer graph structure is constructed.

25. The method according to claim 24, characterized in that, The step of constructing a wireless LAN data layer graph structure based on the wireless LAN data information searched from the target merchant includes: Obtain a list of Wi-Fi data information for the target merchant, the list of Wi-Fi data information includes Wi-Fi data information for multiple other merchants; Based on the wireless LAN signal strength values ​​received from other merchants in the wireless LAN data information list, the third connection distance relationship between the target merchant and other merchants is determined; Based on the third connection distance relationship between the target merchant and other merchants, a wireless LAN data layer graph structure is constructed.

26. A method for predicting the arrival of delivery resources at a target merchant, characterized in that, include: Obtain the information of the fourth merchant to which the fourth recipient of the delivery resource belongs at the current moment; The fourth merchant information is input into the merchant target location vector model to obtain the fourth merchant target location vector corresponding to the fourth merchant information output by the merchant target location vector model. The merchant target location vector model is used to obtain the fourth merchant target location vector based on the initial location vector corresponding to each layer of the multi-layer graph structure. The multi-layer graph structure includes a near-field wireless node data layer graph structure and a historical behavior data layer graph structure of delivery resources. Obtain the information of the fifth merchant to which the fifth object to be delivered belongs, and obtain the target location vector of the fifth merchant corresponding to the information of the fifth merchant; The target location vectors of the fourth merchant and the fifth merchant are used as input data and fed into the duration prediction network model to obtain the second predicted duration information output by the duration prediction network model, which represents the delivery resource from the location of the fourth merchant to the location of the fifth merchant. If the second predicted duration information meets the preset duration review conditions, then the fifth merchant is determined to be the target merchant that the delivery resources need to reach after leaving the fourth merchant.

27. The method according to claim 26, characterized in that, The fifth merchant information includes at least one fifth candidate merchant information; The method further includes: For each of the at least one fifth candidate merchant information, obtain the second candidate predicted time information of the delivery resource from the location of the fourth merchant to the location of the fifth candidate merchant; The second predicted duration information meets preset duration verification conditions, including: Multiple candidate prediction duration information are sorted according to the length order from smallest to largest to obtain the second target prediction duration information, wherein the second target prediction duration information is less than the other candidate prediction duration information among the multiple candidate prediction duration information.

28. The method according to claim 26, characterized in that, The step of inputting the fourth merchant information into the merchant target location vector model to obtain the fourth merchant target location vector corresponding to the fourth merchant information output by the merchant target location vector model includes: Based on the fourth merchant information, obtain the sixth initial position vector of the fourth merchant corresponding to the fourth merchant information in the near-field wireless node data layer graph structure, and obtain the seventh initial position vector of the fourth merchant corresponding to the fourth merchant information in the historical behavior data layer graph structure of the delivery resources. Calculate the fourth loss value between the sixth initial position vector and the seventh initial position vector; Based on the sixth initial position vector, the seventh initial position vector, and the fourth loss value, the fourth merchant target position vector is obtained according to a preset calculation method.

29. The method according to claim 26, characterized in that, The acquisition of the fourth merchant information to which the fourth recipient of the delivery resource belongs at the current moment includes: Obtain a third notification message sent by the fourth merchant terminal used by the fourth merchant, indicating that the delivery resource has successfully picked up the fourth delivery object at the current moment; Based on the third notification message, obtain the information of the fourth merchant to which the fourth object to be delivered belongs at the current moment.

30. The method according to claim 26, characterized in that, The step of obtaining the fifth merchant target location vector corresponding to the fifth merchant information includes: The fifth merchant information is input into the merchant target location vector model to obtain the fifth merchant target location vector output by the merchant target location vector model corresponding to the fifth merchant information; The merchant target location vector model is used to query the fifth merchant target location vector corresponding to the fifth merchant information in a pre-stored correspondence table between merchant information and merchant target location vectors, based on the fifth merchant information. The fifth merchant target location vector is obtained based on the initial location vector corresponding to each layer of the graph structure in the multi-layer graph structure. The multi-layer graph structure includes a near-field wireless node data layer graph structure and a historical behavior data layer graph structure of delivery resources.

31. An apparatus for predicting the arrival time of delivery resources at merchants, characterized in that, include: The first acquisition unit is used to acquire the first merchant information to which the first delivery object to be picked up belongs; The first merchant target location vector acquisition unit is used to input the first merchant information into the merchant target location vector model to obtain the first merchant target location vector corresponding to the first merchant information output by the merchant target location vector model. The merchant target location vector model is used to obtain the first merchant target location vector according to the initial location vector corresponding to each layer of the multi-layer graph structure of the first merchant. The multi-layer graph structure includes a near-field wireless node data layer graph structure and a historical behavior data layer graph structure of delivery resources. The second merchant target location vector acquisition unit is used to acquire the second merchant information to which the second delivery object to which the delivery resource has been claimed belongs, and to acquire the second merchant target location vector corresponding to the second merchant information; The first prediction duration information acquisition unit is used to input the second merchant target location vector and the first merchant target location vector as input data into the duration prediction network model, and obtain the first prediction duration information output by the duration prediction network model, which represents the delivery resource from the location of the second merchant to the location of the first merchant.

32. An apparatus for constructing a multi-layered spectrogram structure, characterized in that, include: The first building unit is used to construct a near-field wireless node data layer map structure to characterize the connection relationship between the target merchant and other merchants in the target area based on the near-field wireless node signals of the target merchant. The second construction unit is used to construct a wireless local area network data layer map structure to represent the connection relationship between the target merchant and other merchants in the target area based on the wireless local area network data information searched by the target merchant. The third construction unit is used to acquire historical behavior data of delivery resources within the first preset time period and construct a historical behavior data layer graph structure of delivery resources to represent the connection relationship between the target merchant and other merchants in the target area. The fourth construction unit is used to construct a multi-layer graph structure to represent the connection relationship between the target merchant and other merchants in the target area, based on the near-field wireless node data layer graph structure, the wireless local area network data layer graph structure, and the historical behavior data layer graph structure of the delivery resources.

33. A training device for a merchant target location vector model, characterized in that, include: The fifth construction unit is used to construct a multi-layer graph structure to characterize the location of the target merchant in the target area where the delivery resources arrive. The multi-layer graph structure includes a near-field wireless node data layer graph structure and a historical behavior data layer graph structure of the delivery resources. The first initial position vector acquisition unit is used to input the near-field wireless node data layer map structure into the merchant initial position vector model to obtain the first initial position vector of the target merchant in the near-field wireless node data layer map structure output by the merchant initial position vector model. The second initial position vector acquisition unit is used to input the historical behavior data layer graph structure of the delivery resources into the merchant initial position vector model to obtain the second initial position vector of the target merchant in the historical behavior data layer graph structure of the delivery resources output by the merchant initial position vector model. The calculation unit is used to calculate the first loss value between the first initial position vector of the target merchant and the second initial position vector of the target merchant; The training unit is used to train the merchant's initial location vector model based on the first loss value to obtain the merchant's target location vector model. The merchant's target location vector model is used to obtain the merchant's target location vector within the target area.

34. An apparatus for predicting the arrival of delivery resources at a target merchant, characterized in that, include: The fourth acquisition unit is used to acquire the information of the fourth merchant to which the fourth object to be delivered belongs at the current moment; The fourth merchant target location vector acquisition unit is used to input the fourth merchant information into the merchant target location vector model to obtain the fourth merchant target location vector corresponding to the fourth merchant information output by the merchant target location vector model. The merchant target location vector model is used to obtain the fourth merchant target location vector based on the initial location vector corresponding to each layer of the multi-layer graph structure of the fourth merchant. The multi-layer graph structure includes a near-field wireless node data layer graph structure and a historical behavior data layer graph structure of delivery resources. The fifth merchant target location vector acquisition unit is used to acquire the fifth merchant information to which the fifth delivery object to be picked up belongs, and to acquire the fifth merchant target location vector corresponding to the fifth merchant information; The second prediction duration information acquisition unit is used to input the target location vector of the fourth merchant and the target location vector of the fifth merchant as input data into the duration prediction network model to obtain the second prediction duration information output by the duration prediction network model, which represents the delivery resource from the location of the fourth merchant to the location of the fifth merchant. The review unit is used to determine the fifth merchant as the target merchant that the delivery resource needs to reach in the next moment if the second predicted duration information meets the preset duration review conditions. The second predicted duration information is the target duration information of the delivery resource from the location of the fourth merchant to the location of the fifth merchant.

35. An electronic device, characterized in that, The electronic device includes a processor and a memory; The memory stores a computer program, and after the processor runs the computer program, it executes the method described in any one of claims 1-30.

36. A computer storage medium, characterized in that, The computer storage medium stores a computer program, which, when executed by a processor, performs the method described in any one of claims 1-30.

Images